Trinity Strategy Scoreboard

The bet, and the odds

~5%

Full path: £15–45M exit

Range 3–9%. Recalibrated after the pricing correction; the headline barely moves, but what must be true changed: a scalable paid channel and near-self-serve onboarding.

15–20%

Meaningful acquisition of any size

Tuck-ins happen well below £3M ARR (Motortech.ai was bought bootstrapped). A moderate outcome, not the goal.

12–24mo

Window on the independent segment

Before it is crowded and price-compressed. Clock evidence in section 02.

The bet in one paragraph Trinity sells an AI customer agent (calls, texts, emails) to small UK independent dealers at £100–300/mo, acquired through Meta ads and closed on a founder sales call, live in under an hour. The long-run vision is the Owner.com play, expanding into a free system of record once paying dealers pull for it. Bootstrapped now; a small operator-angel round only after unit economics are proven. The exit thesis is a sale to a PE-owned automotive software strategic at £3–5M ARR, which at this pricing is the same thing as segment leadership (~1,250–2,000 dealers).

The odds, gate by gate

Each gate's odds are conditional on the previous gate passing. What moves the odds fastest now, in order: a working funnel (cost per paying dealer under ~£600), then month-3 retention, then logo count.

GATE 1

First paying design partner

by ~Sep 2026

50–60%

conditional

Up slightly. A £100–300 yes is easier than a £2k yes, and ads open pipeline beyond the roadshow. Product readiness still gates.

GATE 2

£250k–£1M ARR (~100–420 dealers), retention proven

by end 2027

30–40%

conditional on G1

Needs 4–8x more logos than the old premium plan, but through a channel Carl has proven in an adjacent market. Churn on a £100–300 SMB book is the big unknown.

GATE 3

£3–5M ARR = ~1,250–2,100 dealers = segment leadership

2028–29

30–40%

conditional on G2

Exit threshold and leadership are now the same milestone. Requires the ad machine to hold CAC through scale and churn under ~2–3%/mo.

GATE 4

Acquired at £15–45M

exit window

55–70%

conditional on G3

Unchanged appetite. Buyers discount SMB books for churn, but 1,500+ independent logos is a distribution asset none of them can build.

Cumulative odds as the chain compounds

after G1

~55%

after G2

~19%

after G3

~7%

after G4 (exit)

~4–5%

Calibration. The base rate for a pre-revenue two-founder company reaching any £10M+ exit is low single digits; Trinity sits above it (proven demand category, two active UK acquirers with recent small-AI deals, clean cap table) and below its own ambition (the field is funded and moving). The pricing correction barely moved the headline. What changed is what must be true: a scalable paid channel and near-self-serve onboarding, instead of a premium sales motion. Each gate is cheap to test; failing one is a learning, not a crater.

The exit itself — who buys, why, and the path there — lives in the Getting acquired tab.

Why this can work — the wedge and the window

Verdict Yes, there is a clear wedge. Its moat is speed, not technology.

Every inbound enquiry (call, text, email, form) answered and converted for an independent dealer, live in under an hour, at £100–300 a month. The size of the leak this plugs (WTP research, 5 Jul): across ~750,000 UK dealership sales calls, 20% are never answered; of the answered, 40% are never logged anywhere; 40% of callers are offered no next step — the funnel leaks at three stages, not one.

Why it is real

Category-validated pain. 10+ funded competitors sell inbound AI to UK dealers; Carl proves the same product sells to the same buyer in Germany via paid social.
Incumbents structurally cannot serve it. Keyloop/Pinewood sales models do not reach the long tail profitably; their AI moves are franchise-first.
The price floor is really two different attacks — with two different defences (split 5 Jul). Attack 1, cheap standalone tools: the £30–97/mo garage products answer phones for MOT bookings but don't run dealer sales workflows; IONOS launched a £9–29/mo horizontal receptionist in the UK; Jodie has anchored garages at £49 for 8 months. Defence: sell the integration difference, never the price. Attack 2, free-via-finance bundles: DSG gives qualifying dealers (3+ financed deals/mo — a productive minority of its ~3,100 partners, not all of them) MotorSales.AI's text tool free (the ~£249–250/mo bundle, a discount off the £299 core list), and Auto Union now funds free dealer websites too — the perk playbook is spreading. This can't be price-matched (someone else's commission pays for it); the defence is channel coverage — their free tool is text-only and never rings a phone. One message per attack: "we integrate deeper than a cheap bot" vs "we cover the channel their free tool doesn't."
The empty square is verified (4 Jul 2026, ~90% confidence). Full market synthesis — 45 competitor cards, 3 ad-library sweeps over 501 advertisers, a 12-market Europe scan, a US deep-dive: no UK Meta advertiser combines automotive-only identity + AI customer agent + paid social. The entire mapped UK competitor set buys ~zero Meta ads; the contest is new entrants only. The UK is the last major European market with the slot open. The clock on that square: fonio.ai, the nearest DACH analog, went 0 → 7,000+ customers in ~21 months (€14.6M seed at a €120M valuation, June 2026) — the UK has no fonio-equivalent yet, but that is the speed an empty square fills at once a horizontal player breaks through. Evidence: work/competitor-map/market-overview-2026-07-04.md (+ tabbed HTML with UK deep-dive and US analysis).
The desk-level incumbent is Visitor Chat, not Moneypenny. The one tool Trinity's own prospects actually run: human chat, 5–7 day onboarding, no API, anti-AI positioning now softening. The sub-1-hour wedge is structural against a staffing model. ReceptionMate proves the garage-side integration play works solo (Garage Hive/TechMan live, 23+ garages, £200/mo) — partly taken in garages; dealers untouched.

Why it is fragile

The core tech is not defensible and the price floor is falling.
The window is 12–24 months before the segment crowds.
The actual moat (execution speed, onboarding friction, the conversation-data flywheel into a future CRM) does not exist yet. It gets built by moving fast, not by building more product.

Scope	Sites	Wedge ARR at £200 blended
UK independent car dealers (ICP)	~11,400	~£27M ceiling
Realistic leader share (15–20%)	~1,700–2,300	~£4–5.5M ARR ≈ the exit threshold
+ garages / bike dealers (adjacency)	~30,000+	~£70M+ ceiling
+ system-of-record expansion	same dealers	multiplies ARPU, not logos

Honest sizing. The wedge alone is a modest prize: a £20–50M exit business, not a unicorn. That matches the stated founder objective. The system-of-record expansion and adjacent verticals are what make it bigger, later.

The window: 12–24 months, and the clock evidence

Moneypenny launched an automotive AI receptionist Feb 2026. 2,000+ automotive clients on the human side, national distribution and trust.
Keyloop completed the Motortech.ai/AIME acquisition Apr 2026, folding AI into Fusion. Franchise end first, independents later.
Pinewood CARS ships DMS-agnostic with 7-day go-live, 2,000+ rooftops. No voice yet.
Autoflows claims 1,500+ dealerships in Europe and is live in the UK, Keyloop-deep.
Commodity floor forming at £9–97/mo (IONOS £9–29 UK launch Apr 2026, Team-Connect, Jodie £49, Electronic Receptionist) in the garage/generic segment.
Superchat entered UK Meta 30 Jun 2026 with dealer-specific creatives — Trinity's positioning near word-for-word ("call, chat and email"); its UK voice line is live and selling (Max test-called it 4 Jul). No UK org or GBP pricing yet. The first-mover claim is a this-quarter asset.
GetCarl shows cross-border ambition — English LinkedIn positioning ("AI Sales Operating System for Automotive Retail"), seven-figure angels, GTM roles open; site still German-only. 200+ customers in ~7 months is the existence proof and the clock.
Germany is the forecast — DE went from ~1 to 10+ dealer-AI players in ~6 months, 4 launching Meta ads within 3 weeks of June. Assume the UK follows within ~a quarter.
The rails are stirring — Dragon2000 ran its first-ever Meta ad Jul 2026 (a DMS ad, not voice). The US shows the endgame: AutoLeap shipped "AIR," a $99/mo AI receptionist inside its 15,000-shop install base (Apr 2026), closing the US standalone window from inside. A UK vendor shipping native voice is now a tracked kill signal.
Watch infrastructure is automated — weekly Meta radar (Mondays 07:00: new entrants on our keywords, Superchat/GetCarl/fonio deltas, empty-square re-verification) + monthly outside-in scan (funding, UK hires, voice launches). Views: outputs/html/meta-ads/competitor-radar.html · outputs/html/competitor-map/market-overview.html.

What speed buys. Logo count in the independent segment before incumbent distribution machines get there. Logos here are the one asset the strategics cannot easily build, because their sales models do not serve the long tail profitably.

What’s decided — and what it commits us to

Three founder decisions, all made 3 Jul 2026, all reopenable at the Sunday session — but each has consequences that are already shaping the plan below.

Decision 1 · Price at £100–300/mo — the volume play

Set by Max, 3 Jul 2026. Supersedes the £1,500–2,000/rooftop assumption in the June valuation notes (marked superseded there). The premium-vs-volume fork is resolved: Trinity is the volume play.

~£100

Small independent /mo

≤£500

Larger independent /mo

≤£1,000

Really large independent /mo

£200

Planning blended average (£100–300 band for the vast majority)

No premium shortcut exists.£3–5M ARR now means ~1,250–2,100 dealers. Exit threshold and market leadership merge into one milestone.

In-person selling cannot scale this.At £200/mo, a field visit per close never pays back. Paid digital + fast remote close is the only motion that fits the math.

Everything must be cheap to deliver.Sub-1-hour onboarding, low support load, and product-led retention are structural requirements now.

Still founder-calibrated, not transaction-proven.V2 stays open until real dealers pay real invoices at these levels. Carro at £45–120/mo argues the market may push even lower.

Setup fees change the funnel math. The decided plan includes done-for-you setup tiers at £999 → £2,500 → custom on top of the monthly price. A dealer acquired at the ~£600 CAC target with a £999 setup fee pays back on day one, making the ads experiment far more forgiving than subscription-only math suggests.

Decision 2 · Acquire dealers through Meta ads

Decision (Max, 3 Jul 2026): primary acquisition via Meta ads, per founder-to-founder input from Carl (getcarl.de), replacing door-to-door as the core motion.

Why it is right

At £100–300 pricing, field sales can never pay back per close. The pricing belief and the channel pivot are the same decision.
Ads are the only motion that reaches 1,250+ dealers with two founders.
The closest analog company has proven the channel for this product and this buyer in Germany.

The risks, named

Channel-transfer risk. One datapoint, different market. UK dealer CPLs on Meta are unknown until we spend. The Phase 1 experiment prices this.
The rejection relocates, it does not disappear. Every lead still ends in a founder sales call. The channel fails quietly if demos go unbooked or leads age past an hour. Watch this signal specifically (V8): a paid channel must not become a comfortable way to avoid the phone.
The roadshow pipeline is not written off. Warm conversations already paid for get closed or killed before any new field motion.

Decision 3 · Agent first, system of record later

The Owner.com-style expansion (free CRM/DMS/website for dealers) stays vision, not build plan: no system-of-record code before the £250k–1M ARR gate, and only if pilot dealers pull for it unprompted. The agent's conversation data is what will pre-populate that CRM when the time comes — the wedge builds the expansion.

The plan, phase by phase

The rollout is two distinct motions, in strict order — plus one option held for later. The phases below all live inside Motion 1 until the trigger fires.

Motion 1 · NOW

Direct

Acquire dealers ourselves: Meta ads → founder demo → close. This is where trust, retention proof, and unit economics get built.
Integrations here are plumbing only — MotorDesk/DealerKit public APIs so the agent answers with live stock data. Not a sales channel yet.

Owns everything until ~50–150 dealers.

Motion 2 · relationship clock starts at ~5–10 live dealers (refined 4 Jul); reseller pitch from ~50

Through the rails

The independents’ software vendors become distribution: partner/reseller conversations with ClickDealer (1,800 dealers) and Dragon2000 (1,000+).
The reseller motion cannot start first — channel partners only carry proven products. But the 4 Jul market synthesis moves the relationship clock earlier: every studied market was won on rails (US: AutoLeap AIR + Pie×Tekmetric; FR: Sandra×Nextlane; DE: Petra→OneQrew; UK: Visitor Chat's 2,000 accounts on pure ecosystem embedding), UK rails are stirring (Dragon2000's first Meta ad), and the US shows the window closes from inside once a rail ships its own voice.
Staging: light-touch contact (intro, integration listing, "we're building on your API") from ~5–10 live dealers; full reseller/co-marketing pitch from ~50 with retention data. Flagged for the Sunday stress-test — shifts Motion 2's start, not its logic.
Double-duty: these are the same moves as the acquisition path. Selling through the rails and embedding toward the Carlyle platforms is one motion, not two.

Reseller trigger unchanged: retention proof (GRR >90%) + outcome data per dealer.

Motion 3 · option, not plan

Agencies

Dealer-website agencies (AutoWeb 3,000+ sites, Bluesky, 67 Degrees) as white-label resellers.
Parked deliberately: agency incentives are weaker, and The Whole Caboodle shows agencies may build their own agent instead of reselling ours.

Revisit only if Motions 1–2 saturate.

The rails: whom to target, in what order, and how to run the conversations ADDED 4 JUL

Focus stays on paid ads — but the rails work now has a ranked list, named doors, and a conversation script, so no thinking is needed when the trigger fires. Full depth — scenarios, deal structures, named contacts, meeting scripts — lives in the "The rails" tab. Sequencing rule: V9 instruments every pilot dealer's stack; after ~10 dealers, the rail with the most Trinity customers gets the first formal approach. Pull beats theory.

Priority	Rail	Reach	Why + door
A1	ClickDealer (Centeca / Carlyle)	~1,800 dealers	Biggest single independent-dealer rail in the UK; embedding = simultaneously the acquisition path. Door: Duncan Josey (CSO); M&A executor Mat Hanson (CFO).
A2	Dragon2000 (Lacour / Carlyle)	~1,000+ dealers/garages	Same Carlyle machine, stated AI-modules thesis; just ran its first Meta ad — partner before they build. Door: Mark Cooper. On Max's case-study list: study before pitching.
B1	MotorDesk	indie DMS, growing	Public REST API (stock, leads, CRM, appointments, calls) — permissionless: Jacob builds today, no meeting needed. The live integration is the demo asset for every A-conversation. Listing talk after 3+ shared customers.
B2	DealerKit / Bluesky	indie long tail	API-open per the ICP stack CSV; same permissionless pattern.
C	Garage Hive / TechMan	garage-segment standard	Required for the garage-creative claim ("books into your Garage Hive diary"). ReceptionMate proves these vendors work with AI products — we arrive second, so differentiate (dealers + garages, multi-channel).
D	Ecosystem hooks: Kerfuffle listing · AutoTrader Messages API · AM ecosystem	ambient legitimacy	The Visitor Chat lesson: marketplace listings + trade press + award presence made every vendor conversation start warm. Cheap, compounding, non-exclusive.

The 3-stage conversation (copied from the players who did it)

LISTING — ask small. "N of your customers run Trinity on your API; list us in your directory." Sandra AI's path (Nextlane partner page first). Vendor cost ≈ zero; the listing converts cold dealers AND warms the vendor. Prerequisite: integration LIVE + 2–3 shared customers. Never pitch before pull.
CO-MARKETING — make their platform stickier. The Pie×Tekmetric framing, endorsed by Tekmetric's own CEO: your shops earn more and churn less with our layer on your rails. Ask: webinar/newsletter + referral intro flow. Give: white-glove onboarding for referred dealers + shared outcome data. Petra's variant: become the vendor's AI story before they have one.
COMMERCIAL — rev-share or bundle, only with retention data (GRR >90% + per-dealer outcomes — what channel partners require). Protect: no early exclusivity (Tier A rails are also acquisition candidates); keep the brand (white-labelling away the customer relationship — the Meistertelefon/fonio model — trades exit value for volume; wrong trade for a logo-count exit story); protect the setup fee in channel pricing.

The objection to prepare for: "we might build it ourselves"

The AutoLeap-AIR pattern is real — so the talk-track is build-vs-partner economics: voice AI is a product line, not a feature (telephony, latency, call QA, per-dealer tuning).
"We are live on your API today; revenue this quarter; zero engineering from you."
US precedent: platforms partner for this layer even at 15,000-shop scale (Pie×Tekmetric).
Every vendor interaction is quiet diligence — Pinewood-Seez and Petra-OneQrew both ran partner-first, acquire-later. Run these relationships like the exit depends on them; it may.
Guardrails: no Tier A formal approach before ~5–10 live dealers + one live integration (light-touch intro earlier is fine); rails work is Max-time-boxed to one conversation-advancing action per week until the trigger.

PHASE 0

Build the funnel

now · ~1–2 weeks

The offer. One landing page, one promise (never miss another enquiry; live in an hour; from £100/mo). Consider the Missed Revenue Audit repackaged as the digital lead magnet since the asset already exists.
Tracking before spend. Meta pixel, a booking calendar, a simple funnel sheet (spend → leads → demos → pilots). No spend until every step is measurable.
Speed-to-lead rule. Every lead gets a call within 1 hour. This is where the channel lives or dies, and it is the founder's job.

PHASE 1

The £1,000–1,500 experiment

~3–4 weeks

Launch 2–3 creative angles (missed-call pain; after-hours enquiries going to a competitor; AutoTrader leads going cold). £30–50/day.
Targets set before launch: cost per lead < ~£30–40 · cost per booked demo < ~£150 · cost per paying dealer < ~£600 (3-month payback at £200/mo).
Close 3–5 pilots across three price points (~£99 / £199 / £299) to put transaction data under the pricing belief.
Kill criterion, agreed in advance: if ~£1,200 of spend produces zero paying pilots, stop and diagnose in order: creative → offer → landing page → channel. Do not silently double the budget. (The full experiment is ~15–25% of one month's burn. Affordable once, not as a habit.)
In parallel, the warm roadshow pipeline gets closed or killed. Those conversations were paid for in days of driving.

PHASE 2

Repeatability

to ~25 dealers · Dec 2026 floor

Onboarding timed on every dealer; target under 1 hour; automate the slowest step after every cohort.
Retention instrumented from day one. Month-3 churn is the number that decides everything downstream.
CAC held stable while spend doubles once, then again. If CAC degrades faster than ~2x on scale-up, the segment ceiling on Meta is near.

PHASE 3

The machine

to ~100+ dealers · mid-2027

Scale spend against proven CAC/LTV; add the second channel only now (likely partnerships with the API-open DMS vendors: MotorDesk, DealerKit, Bluesky).
First expansion experiment (CRM-lite auto-populated from agent conversations) only if pilot dealers ask unprompted.
Earliest point where raising money makes sense (section 06).

The pace this must produce (at £200 blended)

When	Dealers (at £200 blended)	ARR run rate
Sep 2026	first paying pilots	—
Dec 2026	~25	~£60k · 6k EUR MRR survival floor
Mid 2027	~100–150	~£250–350k
Mid 2028	~500–700	~£1.2–1.7M
2029	~1,250–2,100 · #1 by logo count (ClickDealer sits at ~1,800)	£3–5M · exit window

What must be true — the validation board

Every card is a hypothesis with a test. Change a status (in the Markdown) the moment evidence lands.

Small independents will pay for an AI agent at all

In progress

Test: pilot closes from ads + warm roadshow pipeline.

Category proof exists (competitors + Carl) AND UK spend evidence (WTP research, 5 Jul): dealers already pay four figures/month for demand (AutoTrader), £99–259/mo for human-answered calls at 2,000+ automotive clients (Moneypenny), and £49–299/mo for text AI (MotorSales.AI) — £100–300 sits inside existing dealer spend, not above it. Zero Trinity-specific transaction proof yet.

ARPU £100–300 holds in real transactions

In progress

Test: three price points (~£99/£199/£299) across the first pilots.

Founder-calibrated 3 Jul 2026. Carro £45–120/mo argues even lower; no Trinity transactions yet. WTP research (5 Jul): the band is not the risk — the frame is (see V10). Sold as software, Trinity is compared to the £49 anchor and free bundles; sold as revenue capture, to a £25k+/yr hire and £2,000-per-car lost margin.

Retention holds past month 3 (novelty risk)

Unvalidated

Test: pilot cohort churn measured monthly.

None yet. The exit-critical metric.

Onboarding under 1 hour on legacy-email dealers

Unvalidated

Test: first 5 onboardings timed.

Electronic Receptionist claims 5-minute setup, so possible in-category.

Dealers pull for more software from the same vendor (Owner-style expansion)

Unvalidated

Test: count unprompted requests from pilot dealers.

Field observation: many run on legacy email. Anecdotal.

Agent conversation data meaningfully pre-populates a CRM (the bridge asset)

Unvalidated

Test: prototype on pilot data once 3+ months of history exists.

Logical, untested.

Meta ads produce UK dealer leads at workable cost

In progress

Test: Phase 1 experiment, £1,000–1,500 capped. CPL < ~£40 · demo < ~£150 · paying dealer < ~£600. Kill criterion pre-agreed + falsification tests (cost/meeting >2x DACH ref; 10+ meetings→0 pilots; "my DMS is adding that").

Playbook proven 3x in DACH (fonio, Carl, Petra) and 1x in the UK (Jodie, garages, 8 months). CORRECTED 4 Jul: the €600→3-meetings test was a friend's company, different vertical — Trinity's first-party data starts with Phase 1.

Founder speed-to-lead holds (every lead called within 1 hour)

Unvalidated

Test: funnel sheet, lead timestamp vs first-call timestamp.

The known failure mode of the channel. Watched explicitly.

Stack reality matches the CSV (what dealers actually run)

In progress

Test: field notes per dealer contact.

icp-software-stack CSV compiled; verification ongoing.

V10

The revenue-capture frame holds (dealers compare us to missed deals, not to £49 software)

Unvalidated

Test: track how pilot dealers talk about price in sales calls and early usage — do they compare Trinity to £49 tools and free bundles, or to missed calls and lost margin?

Falsification test from the WTP research (5 Jul): if pilots price-compare Trinity to £49 tools instead of missed deals, the frame — not the product — is what's failing. Cheap, early, decisive; the specific mechanism by which V1 could fail even if the product works.

Money, people, and borders

The money: bootstrapped now, small operator round on proof

Now

Do not raise

Pre-revenue and pre-funnel-proof, a raise costs months of founder time and prices terribly. Bootstrapping is also a documented acquisition asset: the closest comps (Motortech.ai, Calldrip) were bootstrapped, and clean cap tables close small deals easily. Keep the father financing cleanly papered (documented loan or simple convertible) so diligence never trips on it.

Trigger

Raise when the machine works

CAC under ~£600 holding through one doubling of spend, month-3 churn under ~3%, roughly 50–100 dealers / £10–20k MRR. At that point capital converts directly into share inside the 12–24 month window, which is the only good reason to take it. If the economics never prove, no capital fixes CAC > LTV. Size and shape: ~£250–750k angel/pre-seed, enough to 3–4x ad spend and make one onboarding/support hire. Not a VC A-round.

From whom, in order of fit

Exited UK dealer-group founders and operators. They know the pain, open dealer doors, and are fine with a £20–50M outcome. The sourcing workstream for exactly these people already exists (work/advisor-sourcing/board-and-angels.md).
Automotive SaaS angels (founders/operators of sold UK dealer-software companies). Same logic plus pattern recognition on the exit path.
Micro-funds comfortable with vertical SaaS outcomes. Only if a lead operator-angel exists first.
Avoid classic VC. The realistic prize (£20–50M) does not fit power-law fund math; that money creates pressure to chase a company we are not building.
Be careful with strategic money (anyone tied to Keyloop/Pinewood ecosystems). Partner with buyers; don't put them on the cap table before the exit conversation.

When to expand outside the UK

Triggered by evidence, not dates. Dates assume the mid-case pace; slip 6–12 months if gates slip.

Market	Trigger	Realistic timing	Reasoning
Ireland	UK machine repeatable: CAC stable through one spend doubling, churn <3%/mo, onboarding automated	~100–150 UK dealers · mid-to-late 2027	Nearly free: same language, same ads playbook, remote onboarding, ~600–800 independent dealers. A test, not a commitment.
Continental EU (Benelux/Nordics first; FR de-prioritized 4 Jul)	500+ dealers, raise closed, first hires in seat	decision mid-2028 · entry 2028–29 earliest	Needs localization, telephony/regulatory setup, and people. Only with outside capital. Europe scan (4 Jul): FR is now the MOST institutionalized EU market (~9 dedicated players, OEM/incumbent-armed: Sandra×Autosphere ~250 sites, Diago 250k calls/mo, Renault-backed Alhena) — closed like DE. Benelux/Nordics/PT verified thin-or-empty: the realistic sequence is IE → Benelux/Nordics.
Germany	Avoid	—	Carl owns the analog position on home turf (now 10+ native players). Fight where the incumbent analog is absent.
US	Not this company phase	—	Different buyer structure, saturated competitor field. Only relevant post-exit or with a US-proven buyer.

Two honest notes. Depth beats breadth for the exit: a clean UK #1 story prices better than a thin three-country footprint, so continental expansion pre-exit is a growth-narrative accelerant, not a requirement. And the thing most likely to pull Ireland forward is UK Meta-audience saturation — ~11,400 target businesses is a small ads pool. Watch CPL trend as the leading indicator.

Hiring: triggered by bottleneck, never by ambition

Hire only when a founder is the bottleneck on a proven, documented process. Never hire someone to figure something out. Planning ceiling: one closing founder sustains ~15–20 net dealer adds/month; the mid-2028 target needs ~25–30, which is why hire #2 is not optional on that path.

Hire	Trigger	Realistic timing	Cost
#1 Onboarding & support	~50–75 dealers; support hours start eating the demo calendar or Jacob's build time	Q2–Q3 2027	~£30–40k/yr
#2 Closer or media buyer	~150–250 dealers; sales motion scripted, Max's calendar full	late 2027 – early 2028	~£35–50k/yr + comm.
#3 Second engineer	~300+ dealers, or when the CRM/system-of-record build starts	2028	~£50–70k/yr

How much money, on each path

Founders take no salary; everything reinvests (Max, 3 Jul 2026). Father financing continues to cover living/base burn (EUR 6–10k/mo).

Path	Extra cash needed	Covers	Source
Bootstrap to ~150 dealers (end 2027)	~£30–60k cumulative headroom	Ads float ahead of collections, tooling, telephony COGS ramp	Extended father line; setup fees recycle into ad spend
Scale path to 500–700 (mid-2028)	~£250–750k	2–3 salaries (~£120–200k/yr), 3–4x ad spend, Ireland entry	The operator-angel round, raised late 2027 against proven CAC/churn
Continental expansion (2028–29)	in the round if sized at top, else ~£500k–1M more	Localization, in-market support, regulatory/telephony setup	Same investors, or delay and stay UK+IE until exit

The quiet risk in "reinvest everything": COGS. Voice minutes and AI inference are real marginal costs; at £100/mo small-dealer pricing with heavy call volume, gross margin can fall toward 60–70%. Instrument per-dealer COGS from pilot #1 (add it to the funnel sheet) — a £100/mo dealer costing £45/mo to serve changes every number above.

Getting acquired — the full picture

Written so someone with no industry knowledge can read it top to bottom in fifteen minutes and understand the market, where Trinity fits, who buys us, and the path there. Every fact is sourced; the canonical research lives in work/acquirer-research/acquisition-readiness.md. Everything follows from one distinction, so start there.

1 · The two worlds of UK car dealers

World 1 — NOT our market

The franchise world

Who: A few thousand rooftops selling NEW cars under contracts with manufacturers (BMW, Ford, VW). Mostly owned by large groups ranked in the industry’s “AM100” league table — multi-billion-revenue businesses with professional management. The scale of the money here: US giant Lithia bought Pendragon’s dealerships in 2023, and what remained — the software arm — became Pinewood.AI.
The rule that explains everything here: The manufacturers call the shots. OEMs mandate software and standards across their dealer networks, so winning one OEM contract means hundreds of rooftops at once. That is Keyloop’s moat — and why selling here means committees, 6–18 month cycles, and OEM politics.
Their software: Expensive enterprise DMS: Keyloop (global #1 — ~16,000 dealer sites in 90+ countries, 82.5% of BMW/MINI UK sites) and Pinewood.AI (UK #2, stock-market listed, 2,000+ rooftops on its CARS suite).
Who sells them AI today: Everyone tries here first, because the contracts are big: CitNOW (video, ~75% of UK franchised dealers), Autoflows (voice AI, Keyloop-integrated, 1,500+ dealerships in Europe), Keyloop’s own AIME (the former Motortech.ai), Impel, Connectivity.CX.
Why it matters to us anyway: This world’s companies have the money. They are the eventual buyers of Trinity, not its customers.

World 2 — Trinity’s market

The independent world

Who: ~15,600 used-car dealer businesses in the UK, of which ~11,400 are owner-operated (1–5 rooftops) — Trinity’s ICP. The owner does the selling, the part-exchanges, the finance referrals — and answers the phone. Per the canon stats: 16–22% of dealer calls go unanswered, and 45% of leads arrive after hours. That gap is Trinity’s product.
The rule that explains everything here: Auto Trader is the gravity center, and nobody owns the rest. ~14,000 retailers pay Auto Trader monthly for their demand. Beyond that, the market is wildly fragmented: the biggest software vendor (ClickDealer) has just ~1,800 dealers — roughly 11%. That fragmentation is why PE is rolling the layer up, and why Trinity’s leadership bar is only ~2,000 dealers.
Their software: Three tiers: integrated DMS+website at £50–300/mo (ClickDealer 1,800 · Dragon2000 1,000+ · Spidersnet 1,000+ · CarDealer5 ~2,300 claimed · MotorDesk); website-only agencies (AutoWeb 3,000+ sites, Bluesky); or literally just email and a diary. (Note, 5 Jul: a slice of this stack is finance-broker-funded, not dealer-paid — two brokers fund entire free dealer websites (DSG via Spidersnet; Auto Union, at its discretion, via AutoWeb) and finance calculators are widely lender-funded (Codeweavers, iVendi). A slice, not most of it — core DMS and most websites remain dealer-paid. See the Finance rails tab.)
Who sells them AI today: Almost nobody at scale — that’s the opening. The early movers are small: Carro (£45–120/mo voice agent), Clevaa, The Whole Caboodle (agents bundled with websites), and MotorSales.AI (added 5 Jul: its "Theo" agent is text-only — WhatsApp/SMS/email, no phone — £299/mo core list, £49 add-on hubs; DSG-affiliated in all but name, incorporated at DSG's own Stockport address with DSG's CEO on the board, running unregulated on DSG's FCA permissions. Per the ~22-provider census it is the one autonomous customer-comms tool in any finance-funded bundle). A £30–97/mo commodity floor is forming in the adjacent garage segment (Team-Connect, Electronic Receptionist).
Trinity’s play: Sell the AI agent at £100–300/mo via Meta ads, reach ~1,250–2,100 dealers = £3–5M recurring ARR = segment leadership. That is the exit-ready state.

15,600

UK used-car dealer businesses

The whole of world 2.

~11,400

Owner-operated — Trinity’s ICP

One decision-maker per business.

~14,000

Retailers paying Auto Trader

The demand layer everyone depends on.

Keep the two worlds separate and this whole page becomes simple. Keyloop lives in world 1 — which is exactly why it keeps appearing in the research but is NOT our buyer. Pinewood is the one world-1 company whose products already sell into world 2, which is why it is our #1 buyer hypothesis. And the PE firms are buying up world 2’s software companies right now — which is both our distribution opportunity and our second exit route.

2 · Who’s who — the money, the platforms, the players

Four layers. PE firms buy platforms. Platforms buy capabilities — that’s Trinity. Consolidators sweep the leftovers cheap. Click any box for its dossier and links.

Layer 1 · The moneyPE firms + permanent capital. They don't buy Trinity directly — they buy platforms, then have platforms buy us. Motive: make the platform worth more before their own exit in 3–7 years.

Layer 2 · Their platformsThe entities that actually write the cheque. They buy for capability (a product gap) or distribution (a customer base they can't reach).

Layer 3 · Independent strategicsListed companies that acquire for themselves, not for a PE sponsor.

Layer 4 · The fragmented long tailFounder-owned micro-vendors — Trinity's own layer. This layer cannot buy anyone; it gets bought (Dragon2000 already was). Our job: become its most valuable asset before consolidation finishes.

Click any box to expand its dossier. One open per layer; click again to close.

3 · What’s happening: the consolidation, year by year

Who bought whom, 2019 → today. Read the acceleration: one deal a year becomes five. The un-bought independent-dealer software layer is shrinking, and Trinity is building inside the shrinking part — which is the whole point.

FEB 2019
LDC takes a minority in Evolution Funding (£110M valuation)
Why: the classic PE first move — a minority stake in the segment’s biggest cash machine (the largest used-car finance broker) to test the market before committing to a full platform. It worked: four years later Carlyle bought the majority and the buy-and-build began. The lesson for reading everything below: PE enters where revenue is recurring and the market is fragmented — and that stake is now 7+ years old, which means an exit event inside Centeca is a matter of when, not if.
SEP 2020
Constellation/Perseus buys Catalyst
Why: Constellation’s whole model is buying 30-year-old software businesses with sticky customers at low prices and holding them forever — it prizes retention over growth, which is why it pays ~1x revenue while strategics pay 10x. Catalyst’s motor-trade DMS book (loyal since 1989) is exactly that. The lesson: there is always a buyer for a retentive software book — but the boring buyer pays boring prices. This is the floor under Trinity, not the goal.
2021
Francisco Partners carves Keyloop out of CDK · Keyloop buys RAPID RTC + enquiryMAX · Evolution buys ClickDealer · Cox buys Codeweavers
Why, deal by deal: Francisco bought a neglected corporate division (CDK’s international arm) cheap, renamed it Keyloop, and immediately bolted on lead-management tools — enquiryMAX came with a Jaguar Land Rover mandate attached, i.e. buying revenue that OEM politics protects. Evolution buying ClickDealer is the most instructive one for us: a finance broker bought the software its dealers use every day, because owning the dealer’s daily workflow means owning where the finance applications originate. Distribution logic, not software logic. Cox (US giant) bought UK checkout-finance tech to assemble digital retail. The lesson: platforms buy whatever deepens their grip on the dealer relationship.
2023 · THE PIVOT YEAR
Carlyle takes majority control of BOTH Lacour (Feb) and Evolution Funding (Jul)
Why: Carlyle had just closed a fresh €3bn European tech fund (Dec 2022) — a new fund must find platforms fast, and it picked two category anchors in the same segment within five months. The stated theses are near-identical: consolidate a fragmented market through further M&A, add software and AI capability. The lesson: the platform purchase is the starting gun, not the finish — everything a platform does afterwards (below) is about becoming worth more before the fund’s exit, which for both of these lands around 2028–2030. Also 2023: Lacour’s first bolt-on (KeplerVO, used-car stock software — widening the product surface); TekCor4 starts its own founder-led buy-and-build with Catalyst for Aftersales.
2024
Keyloop buys ATG (May) · Lacour buys Dragon2000 (Oct)
Why: the ATG deal shows sponsors trading assets between themselves — Keyloop (Francisco) bought omnichannel-retail capability from another PE firm (Inflexion) rather than building it. The Dragon2000 deal is the one to study closely: a 29-year founder-owned UK DMS with 1,000+ dealers and no institutional capital — the textbook succession sale. Mark Cooper got liquidity, stayed in charge, and Lacour got its UK entry in a single signature. The lesson: geographic entry via bolt-on is standard PE practice, and founder-owned businesses in our layer are being picked off one by one. Deal coverage.
2025
Pinewood completes Seez at ~10.5x ARR (Mar) · Lacour buys Proger, Italy (Oct) · Astira buys CallRevu (Oct) · TekCor4 buys Marketing Delivery (Nov) · Valsoft buys Quorum (Dec)
Why, and what each deal prices: Pinewood/Seez is the premium anchor — a listed strategic paid ~10.5x ARR because building the AI capability itself would have taken longer than its stock-market narrative allowed, and it had de-risked the deal by investing in Seez first (partner → invest → buy). Lacour/Proger is the country-a-year cadence confirming itself (FR→UK→IT). Astira/CallRevu is a brand-new $675M fund buying its platform explicitly to do M&A with it. And Valsoft/Quorum is the other end of the spectrum: a flat-growth DMS went for ~1.5x revenue. The lesson: the same year, the same industry — 10.5x for a growing AI capability, 1.5x for a flat legacy book. Growth story and retention ARE the price.
2026
Keyloop buys Motortech.ai (Jan–Apr) · Apax bids £575M for Pinewood, withdraws (Jan–Feb) · CallRevu buys Calldrip (Apr) · Perseus buys Starkwood (Apr) · Evolution rebrands as Centeca (Jun)
Why: the AI-bolt-on era opens. Keyloop bought Motortech.ai — bootstrapped, revenue undisclosed and probably small — because the AI agent filled a roadmap gap in Fusion; the price mattered less than the speed. It is the first pure AI-agent acquisition in the UK market, and it was bought for exactly the segment Trinity targets. Apax bidding ~10x EBITDA for all of Pinewood (withdrawn on market conditions, not on thesis — they can return after ~13 Aug 2026) shows PE appetite reaches the very top of the stack. CallRevu/Calldrip is the fund clock in action: first bolt-on within six months of Astira’s ownership. Centeca’s rebrand is what a platform does when it starts assembling its exit story. The lesson: every type of buyer Trinity needs — capability strategic, PE platform, top-of-stack sponsor — did a deal in the last 18 months.

4 · The rails: the software independents already run — and how easily we plug in

Trinity doesn’t replace this software; it sits on top of it, answering the dealer’s calls, texts and emails. So these vendors are our integration rails and doorways, ranked here by how easy the plumbing is. Two of them are also the side doors into the buyers.

Software	Dealers on it	How easy to integrate	Owned by	Trinity’s move
MotorDesk	growing	EASY — public REST API (stock, leads, CRM, even calls/VOIP); self-serve keys	Founder	Integrate first. Days of work, not months.
DealerKit	small, newer	EASY — public Integrators API + webhooks	Founder	Integrate first. Best-documented API in the layer.
Bluesky Interactive	hundreds of dealer sites	EASY — public Stock + Leads APIs, has onboarded AI vendors before	Employee trust	Integrate early.
Auto Trader Connect	~14,000 retailers reachable	MEDIUM — partner programme API (the marketplace layer, not a DMS)	Auto Trader plc (LSE)	Apply as partner; build the “AI follow-up lifts Auto Trader lead conversion” case study.
ClickDealer	1,800	HARD — no public API; sales-led partner programme (“120+ integrations”)	Centeca → Carlyle	The side door to Buyer #2B. A relationship conversation, not an API key. Target: Duncan Josey. The pitch currency (5 Jul): not "stickiness for your DMS" but "more financed deals per dealer" — Click Dealer is a finance broker's software asset whose economics run on financed volume.
Dragon2000	1,000+	HARD — integrations activated on request, no public docs	Lacour → Carlyle	The side door to Buyer #2A — priority. Target: Mark Cooper, founder/MD.
Spidersnet	1,000+	HARD — contact-form partnership only	Friday Media (family)	Later; reseller-bundle angle at most.
CarDealer5	~2,300 claimed	UNKNOWN — no dev portal, but has built custom feeds before	Two founders	One discovery call to find out what’s real.
The rest (Vehiso, DealerPal, Carbotix, TraderWay…)	small each	Mostly closed	Founders	Skip until a pilot dealer runs one.

The logic: start where the API is public (fast proof, zero permission needed), then use traction to earn the sales-led doors — which happen to be owned by the people most likely to buy us.

5 · Who actually buys us

First, the Keyloop confusion, settled. Keyloop appears everywhere in this research because it is the market’s biggest buyer of dealer software (5 deals since 2021, including the first pure AI-agent deal, Motortech.ai). But it lives in world 1: franchise-first, OEM-mandated, no stated interest in independents. Verdict: an opportunistic buyer at best. We stay visible to them cheaply (partner programme, IMDA listing) and build our plan around the three below instead.

BUYER #1 · PRIMARY HYPOTHESIS

Pinewood.AI — the bridge between the two worlds

Why them: the only world-1 strategic whose product already sells into our world — its CARS suite works with any DMS, goes live in 7 days, and is sold standalone. And the precedent is exact: it bought Seez, an AI capability filling a product gap, for ~$46M at ~10.5x ARR, after first investing in it (their documented pattern: partner → invest → buy). CARS has no voice today. The pitch their own history validates: “Seez v2, for voice and call intelligence.”

What they need to see: capability proof plus some dealer-group logos, not just independent volume. Route in: technology partner via pinewood.ai/partners; they exhibit at AM Live (NEC, 11–12 Nov 2026). Watch: Apax can return for Pinewood itself after ~13 Aug 2026.

Buyer #2 — the Carlyle machine. One PE firm owns two platforms in our world covering ~2,800 dealers between them. The mechanics, in one breath: PE buys a platform → the platform bolts on small companies cheaply (~1–2x revenue) → the same revenue is instantly worth 4–6x+ inside the group → after 3–7 years the PE firm sells everything. Bolt-ons are now ~73% of all PE deals; the platform’s own management initiates them, not the PE firm (so we talk to the people below, never to Carlyle); and the documented pattern is “try before you buy” — 1–4 years of partnership with measurable impact, then the acquisition. Carlyle bought both platforms in 2023, so its strongest buying window is roughly 2026–2028 — slightly before our full target profile exists. If an early offer comes, the choice is early money vs pushing on to the Pinewood profile, and a warm second buyer is what keeps that choice good. A related pattern on the watchlist (5 Jul, not yet a buyer archetype): finance brokers taking ownership of AI vendors they distribute rather than channel-partnering — DSG did exactly this with MotorSales.AI. Currently n=1; it becomes a named buyer archetype if a second broker repeats it.

BUYER #2A · THE PRIORITY DOORWAY

Groupe Lacour · Carlyle majority Feb 2023 · 12,000+ clients · groupe-lacour.fr

Company	What it does	Joined	Relevance to Trinity
Lacour core (IRIS, ALPHA2A, DIVA, MOSAIC)	Bodyshop / after-sales / insurer software, France. Its AI (MOSAIC, Oct 2025) is internal: photo/video/voice-to-quote estimating.	founding, 1985	No overlap — their AI is workshop-internal, not customer-facing.
KeplerVO	French used-vehicle stock management, 500+ clients.	Sep 2023	No overlap. Precedent: founder stayed post-deal.
Dragon2000	UK DMS, 1,000+ dealers/garages. APPraise AI = damage appraisal only. Live Chat = human-staffed website widget.	Oct 2024	The doorway. Integration surface into 1,000+ UK dealers. Watch item: if their human Live Chat ever goes AI, it becomes a competitor inside its own base.
Proger / ClipParts	Italian parts-search software, 1,500+ clients (40M SKUs).	Oct 2025	No overlap. Proves the country-a-year cadence.

How Trinity gets bought here: integrate with DragonDMS → prove call/booking outcomes across their dealer base → become the UK/EU "AI module" that Lacour's own public statements say they intend to add ("continue developing and integrating artificial intelligence into its product range"). They keep brands and founders post-deal, so Trinity would stay Trinity. The door has a name: Mark Cooper, Dragon2000's founder/MD, still in charge post-acquisition. Their deal counsel: Proskauer.

BUYER #2B · PARTNER NOW, BUYER LATER

Centeca · was Evolution Funding · Carlyle majority Jul 2023 + LDC since Feb 2019 · ~600 staff · centeca.com

Company	What it does	Joined	Relevance to Trinity
Evolution Funding	UK's largest used-car finance broker: 35+ lenders, £6.5bn advances facilitated/yr, 5,500+ transacting retailers (updated 5 Jul from the finance-rails research). 600k+ calls/yr — 100% monitored by rented Voyc AI.	founding, 2002	The rent-AI evidence: they pay a vendor for call QA rather than owning AI.
Click Dealer	ClickDMS + ClickWeb + ClickEngage (self-serve deal builder, explicitly no AI) + ClickLeads (manual CRM). 1,800 dealers.	2021 — their only software deal	The doorway. 1,800 dealers of exactly our ICP with zero AI anywhere in the suite. Trinity is the missing layer, not a rival.
Motion Finance	Finance broker for small/medium used-car dealers, Warrington.	May 2023	None — finance bolt-on.
Creditas	Leeds finance broker, Appointed Representative model.	May 2024	None — finance bolt-on.
Automotive Compliance	FCA Principal Firm — one of the two dominant AR (Appointed Representative) principals in UK motor retail (~300+ dealers under its umbrella; the other is ITC Compliance). Corrected 5 Jul from the finance-rails research.	Dec 2025	None directly — but CEO Lee Streets called it "the final piece of the platform vision": the shopping list looks closed near-term.

How Trinity gets bought here: not soon, and that's fine. Their pattern is buy-finance, rent-AI — so the near-term play is a reseller/integration partnership into ClickDealer's 1,800 dealers, and the buy decision comes later, when Trinity is embedded and a rival owning us would hurt. The door has a name: Duncan Josey, Chief Strategy Officer (ex-Auto Trader, 21 years), who owns the "digitising the customer experience" mandate. The M&A executor when the day comes: Mat Hanson, Group CFO, explicit deal mandate. Extra pressure in the background: LDC's stake is 7+ years old — some form of exit event is coming, and growth stories get assembled before exits.

BUYER #3 · THE US ENTRANT

CallRevu — owned by Astira Capital, the freshest money

Why them: the closest product surface to Trinity anywhere (US dealership call intelligence), owned since Oct 2025 by a young fund that must deploy $675M and has already bought once (Calldrip, within six months) with stated EMEA ambitions. The trigger to watch: their second concrete UK signal (a UK entity, UK hires, or a UK deal). Prepare now, cheaply: know Rhino Group / Oceros in Manchester — Calldrip’s exclusive UK reseller — before that day comes.

The floor (always there, never the goal)Constellation/Perseus · ~0.8–1.5x revenueValsoft · bought Quorum at ~1.5x= a guaranteed £3–15M exit at almost any stage

Watchlist (a glance per quarter)SoleraImpel (bought Outsell $100M+)Serent CapitalCitNOW (wakes if a rival proves voice)TekCor4 (small checks)

6 · The path, stage by stage

One plan, four stages. Each stage says what we do and what it unlocks. The selling motion itself (Meta ads → demo → close) lives on the Scoreboard tab; this is the acquisition layer on top of it.

STAGE 1

Now → first ~25 dealers

→ Dec 2026

Integrate where no permission is needed: MotorDesk + DealerKit (public APIs), apply to Auto Trader Connect.
Four hygiene items, free now, fatal later: IP-assignment signatures from everyone who touched the code; MRR reported separately from setup fees from invoice #1 (buyers value setup revenue at ~0.5–2x); change-of-control clause checked in the dealer contract template; GDPR basics for a company recording calls (data map, DPAs, retention policy).
Book AM Live (NEC, 11–12 Nov 2026 — Pinewood and CitNOW exhibit; booking closes early autumn).
Unlocks: per-dealer call-outcome data — the raw material of the entire capability story.

STAGE 2

~50–150 dealers

≈ 2027

Prove retention: GRR >90%, month-3 cohort churn measured and published internally. This single number moves the exit multiple more than anything else (~4x under 5% annual churn vs ~2.5x above 10%).
Open the side doors — this is GTM Motion 2 beginning: Dragon2000 integration conversation (Mark Cooper) and ClickDealer/Centeca conversation (Duncan Josey). The rails stop being plumbing and become a sales channel. Framed as partnership, never as sale. Their buying window (2026–28) is already open.
IMDA supplier listing — Motortech’s own pre-acquisition route.
Unlocks: tuck-in optionality exists; the try-before-you-buy clock starts ticking inside the Carlyle platforms.

STAGE 3

~150–400 dealers · £0.5–1M recurring ARR

late 2027 – 2028

First 1–2 recognisable dealer-group logos — what Pinewood specifically needs to see.
Pinewood technology-partner status (the Tjekvik-style entry to their partner → invest → buy pattern) + Catalyst integration (visibility inside Constellation).
Unlocks: the consolidator floor solidifies under the company; the Pinewood conversation becomes credible.

STAGE 4

~800–2,100 dealers · £2–5M recurring ARR

2028–29 · the window

The exit-ready state: 5–10 group logos, annual churn <5%, NRR >100%, live DMS integrations, and a second buyer genuinely warm (auction tension is the biggest price lever we control).
The likely endings: Pinewood buys the capability (Seez pattern, premium multiple), or a Carlyle platform buys its embedded AI layer (earlier, likely 3–6x), with the Constellation floor underneath throughout.
The standing rule at every stage: never signal “for sale.” Everything is partnership and category leadership until a buyer starts the conversation.

7 · The money

The three exits, and the revenue each needs

Exit shape	Customers / revenue needed	Price	Evidence
Capability tuck-in (Keyloop/Motortech pattern)	No ARR floor; ~50–150 dealers with strong call-outcome data	low single-digit £M (est.)	Motortech bought bootstrapped; price permanently undisclosed (micro-entity filings, verified via Companies House). Fallback, not the goal.
Strategic sale at the credible threshold (Seez pattern) — THE PLAN	~800–2,100 dealers = £2–5M recurring ARR + 5–10 group logos + live integrations	£15–45M at 3–9x	Seez: $4.4M ARR → ~$46M (~10.5x). Base market: SEG median ~4.1x, small-deal median 3.3x; vertical SaaS earns a 25–30% retention-driven premium.
Consolidator floor (Constellation/Valsoft pattern)	Several hundred dealers + 2 years of low churn	£3–15M only	CSU pays ~0.8–1.5x revenue; Valsoft bought Quorum DMS at ~1.5x. Always available; never the target.

Retention is the price. FE International's evidenced rule: ~4x ARR under 5% annual churn compressing to ~2.5x above 10%. SEG's screening bar before buyers engage seriously: GRR >90%, NRR >100%. On a £4M ARR base, the gap between 8% and 20% annual churn is roughly £10M+ of exit value. And it must be recurring revenue: setup fees price at ~0.5–2x, so MRR is reported separately from invoice #1.

What moves the price up: retention data, live integrations, group logos, a competing buyer, a clean cap table. What drags it down: churn, revenue concentrated in a few dealers, services-heavy revenue, integration claims that turn out aspirational.

What would kill it

Risk	Mechanism	Early signal	Mitigation
Meta CAC doesn't transfer to the UK	German channel economics fail to reproduce; funnel too expensive	Phase 1 misses targets after full £1,200–1,500 spend	Pre-agreed kill criterion; diagnose creative → offer → page → channel; fall back to warm pipeline + partnerships. Note 4 Jul: the "€600 → 3 meetings" datapoint was a friend's company in a different vertical — Trinity has zero first-party channel data until Phase 1; falsification tests wired in (cost/meeting >2x DACH ref · 10+ meetings→0 pilots · "my DMS is adding that" objections).
A UK rail ships native voice (the AutoLeap-AIR pattern)	Garage Hive / TechMan / MotorDesk / Dragon2000 bundles an AI receptionist into software dealers already pay for — in the US this closed the standalone window from inside (AutoLeap AIR, $99/mo, Apr 2026)	UK vendor product announcements; Dragon2000 already ran its first Meta ad Jul 2026; prospects citing "my DMS is adding that"	Motion 2 relationship clock started early — become their voice layer before they build one. Weekly radar watches announcements; the integration wedge is also the defence.
Superchat beds into the UK	A funded horizontal ($19M, ~100 staff) converts its 30 Jun UK entry into GBP pricing + a UK org before Trinity has logos	GBP pricing page · UK job posts · UK automotive case study (weekly radar checks all three)	Speed on Phase 0–1; automotive-only identity + integration depth a horizontal cannot template quickly.
Speed-to-lead failure	Leads age, demos unbooked; the channel becomes rejection-avoidance	Lead-to-first-call gap > 1 hour in the funnel sheet	V8 watched explicitly. This is founder behavior, not channel math.
Retention failure	£100–300 SMB book churns after novelty wears off	Month-3 cohort churn >10%	Workflow completion (bookings made, revenue recovered), not just call answering.
Incumbent distribution wins first	Moneypenny/Pinewood reach independents before we have 100+ dealers	Their independent-segment case studies appearing	Speed on G1/G2. Logo count now.
Commodity price floor	Voice AI at £30/mo makes the agent a feature, not a product	Pilot dealers anchoring to Team-Connect pricing	Data depth; the system-of-record layer is the long-run defence.
MotorSales.AI ships voice in the DSG bundle (added 5 Jul)	DSG's in-house-affiliated text tool (Theo) adds phone capability, closing the one confirmed free-stack gap — for the qualifying minority of DSG's ~3,100 partners on the 3+ deals/mo tier (single-digit-to-low-teens % of UK independents, not all 3,100)	MotorSales.AI product page / press / App Store / job ads mentioning voice or calls — watched weekly, by name	The research calls this the single event that turns the free-stack economy from context into direct threat. The defence stays phone + full loop + bolt-on integration depth vs their rip-out-your-stack model.
A finance broker bundles a rival's AI as a dealer perk (added 5 Jul)	Any of the ~22 censused providers (Close Brothers, Santander, Centeca…) — all comms-free today except DSG — signs a competitor into its dealer bundle before Trinity gets there, closing a rail-funded distribution door	A new AI-comms vendor name appearing in a lender/broker published dealer-perk stack	Speed on the finance-partner ladder (Finance rails tab, F9); the whitespace ranking (F10) is time-limited, not permanent. No exclusivity given if we land first.
Founder capacity	One engineer, one GTM. Any extended outage halves the company.	Slipped gates without external cause	Right-size the build. No system-of-record code before G2.
Runway	EUR 6–10k/mo burn, father-financed, no revenue; ad spend adds burn	Gates slipping while burn continues	Capped experiments, rank by speed to revenue. The Dec 2026 floor is real.
Single-comp valuation risk	The 10.5x Seez anchor does not repeat	Next UK dealer-AI deals pricing at 3–5x	Build the premium-case list. Plan on 4–6x.

How to keep this alive. The Markdown at work/strategy/strategy-scoreboard.md is canonical; this page is a generated view. When a gate passes or fails, a validation card changes status, funnel data arrives, or a new comp lands: update the Markdown, bump last-reviewed, add a log line, then ask Claude to "refresh the strategy scoreboard HTML". Generated 5 July 2026 (v20: cross-tab audit applied — 43-agent workflow swept every tab against the finance-rails research; commodity floor split into two attacks, kill-table tripwires, validation card V10 + WTP evidence, MotorSales.AI/Centeca/Evolution corrections on the acquired tab, rails guardrails aligned, vision WATCH tripwires, and the F9 precedent corrected; v19: Finance rails tab redesigned end-to-end — perk-stack census of ~22 providers, market map, corrected evidence base after an 8-agent adversarial recheck, and the internal change tracker for queued cross-tab deltas; v18: Finance rails added as a fifth tab — the finance-funded dealer-tooling economy, broker whitespace ranking, MotoNovo window, introducer verdict, deal structures, and cross-tab deltas, from the two-pass finance-rails research; v13: two-motion rollout frame added to the plan (direct → rails, agencies parked); v12: two-worlds section deepened (players, rules, who-sells-AI, market numbers) and timeline expanded with per-deal rationale; v11: scoreboard tab rebuilt Jacob-first into seven sections (bet+odds, wedge+window, decisions, plan, validation, money/people/borders, risks); v10: acquired tab rebuilt as a seven-part narrative; v2–v9: pricing, ads pivot, buyer research, dossiers, links).

The rails — the in-depth strategy

Why this tab exists. Every market we studied was ultimately won on rails, not ads: the US small-shop voice market was claimed by AutoLeap AIR (inside its own install base) and Pie ($23.7M raised specifically to ride Tekmetric's 15,000 shops); France by Sandra×Nextlane/Autosphere and Renault-armed Alhena; Germany's Petra rode OneQrew's ERP channel to its exit; and the UK's own incumbent, Visitor Chat, built 2,000+ accounts with zero ads through pure ecosystem embedding. Ads win customers; rails win segments. For Trinity the rails are triple-duty: distribution (the only path to 1,250–2,100 dealers with two founders), defence (a rail that carries us won't build against us), and the exit itself (both Tier-A rails are Carlyle platforms on the acquirer map; partner-first-acquire-later is the documented pattern: Pinewood→Seez, OneQrew→Petra). The scoreboard tab has the summary; this tab is for making the actual decisions.

R1 · The map — every rail, what it wants, and our way in

A second class of rail sits outside this table: finance partners, who can fund Trinity directly from their per-deal commission rather than distribute it through software — see the Finance rails tab. Partner-funded dealers count toward the same ~40%-per-rail concentration cap.

Rail	Who / reach	Their AI status + what they want	Our entry + the door
ClickDealer	Centeca platform (Carlyle; LDC’s 7-year stake adds exit pressure) · ~1,800 UK dealers	No customer-facing comms AI (verified gap). Platform logic: bolt-ons that raise per-dealer revenue and stickiness; “final piece of the platform vision” shopping posture near-term. Finance-rails upgrade (4 Jul): Centeca is a finance broker whose software runs on financed volume — pitch “more financed deals per dealer,” in their own currency (see Finance rails tab).	Formal approach only with proof (~50 dealers + GRR). Door: Duncan Josey (CSO) for partnership; Mat Hanson (CFO) executes M&A. Until then: track shared customers via V9.
Dragon2000	Lacour platform (Carlyle) · ~1,000+ dealers/garages · DragonDMS	Acquired Oct 2024 explicitly with an AI-modules thesis; country-a-year cadence; keeps founders. First-ever Meta ad Jul 2026 — waking up. No voice product.	Door: Mark Cooper. On the case-study list — study before pitching. Light-touch intro viable earlier than ClickDealer because the AI-modules intent is stated.
MotorDesk	Independent, modern, growing indie DMS	Public REST API (stock, leads, CRM, appointments, calls/VOIP, invoices; per-dealer keys, free custom feeds) — a platform that WANTS builders.	Permissionless: Jacob builds now. This is integration #1 — it powers the ad claim (“books into your DMS”) and becomes the demo asset for every other conversation. Listing talk after 3+ shared customers.
DealerKit · Bluesky	Indie long tail	API-open per the ICP stack CSV; small teams, partner-hungry.	Same permissionless pattern; priority decided by V9 (whichever stack pilots actually run).
Garage Hive · TechMan	Garage-segment management standard	Both already cooperate with an AI vendor (ReceptionMate books into their diaries) — precedent set, but we arrive second.	Technical path proven; differentiate on scope (dealers + garages, calls+texts+emails). Needed before scaling the garage ad lane.
Ecosystem hooks	Kerfuffle marketplace · AutoTrader Messages API · Motors.co.uk · AM ecosystem	The Visitor Chat channel: listings + trade press + awards = ambient legitimacy; vendors and dealers both check these surfaces.	Kerfuffle listing is self-serve — do it at ~10 dealers. AutoTrader Messages API per ecosystem-rails research. AM presence builds toward the buyer-scouting surfaces too.
Agencies (Motion 3 — parked)	AutoWeb 3,000+ sites · 67 Degrees	Weak incentives; Whole Caboodle shows they may build their own agent instead.	Do not activate. Revisit only if Motions 1–2 saturate.

R2 · If the rails work — three scenarios, one recommendation

Scenario 1 · Breadth: listed + integrated everywhere (recommended)

Non-exclusive integrations + listings across MotorDesk, DealerKit, Bluesky, Garage Hive, TechMan, ClickDealer, Dragon2000.
Trinity becomes the AI layer of the independent stack — every rail’s directory sells us; no rail owns us.
Exit effect: maximum auction tension — both Carlyle platforms (and Pinewood) can bid; logo count compounds from every rail’s referrals.
Cost: slower per-rail depth; more integrations to maintain (Jacob’s constraint — sequence by V9 data).

Scenario 2 · Depth: deep with one Carlyle platform

Fastest distribution (a bundle deal could add hundreds of dealers) and the shortest path to being bought.
Risks: exclusivity gravity (they will ask), the twin platform freezes us out, price/terms leverage shifts to them, single-channel concentration.
Only take this shape if the terms buy something irreplaceable (e.g., bundled default status) AND leave brand + billing + a second door open.

Scenario 3 · Freeze-out: rails build their own (the AutoLeap-AIR world)

Signal: a UK vendor ships native voice (tracked kill signal; Dragon2000’s ad activity is the early tremor).
Fallback: direct ads + ecosystem hooks + the segment rails never serve well (smallest independents on spreadsheets — no rail to ride, ours to win directly).
Defence is pre-emption: be live on their APIs with shared customers BEFORE their build/buy/partner meeting happens — it converts “build” agendas into “partner (or acquire)” agendas.

The recommendation, stated plainly

Breadth-first, non-exclusive (Scenario 1), while cultivating BOTH Carlyle doors symmetrically.
The twins share an owner — assume information flows between Lacour and Centeca. Symmetric treatment keeps both warm and preserves exit auction tension; favouring one early collapses it.
Concentration guardrail: no single rail above ~40% of new logos once channel referrals flow.
Data boundary (added 5 Jul): any partner receiving outcome data — referral rails and finance rails alike — gets aggregate reporting only (calls answered, bookings created, revenue per dealer), never raw conversation content. The customers are the dealer's.

R3 · The exchange — what each side wants, and our pre-committed positions

Deal shape	They get	We get	Our position (decided now, so we never negotiate from scratch)
1 · Directory listing	A richer integrations page; an AI answer for their dealers	Legitimacy; “works with your system” conversion; warm vendor relationship	Ask everywhere, immediately once live + 2–3 shared customers. Free both ways. No approval needed to keep building.
2 · Referral / intro flow	Stickier platform; happier dealers; zero build	Warm leads at near-zero CAC (vs Meta CAC — measure the delta)	Offer 10–15% of year-1 subscription as referral fee. White-glove every referred dealer. Report outcomes back monthly (the Pie framing: “your shops earn more and churn less”).
3 · Co-marketing	An AI story for their base before competitors have one (the Petra move)	Webinar/newsletter reach into thousands of dealers	Trade content + data, not margin. Joint case study with named shared dealers is the asset.
4 · Reseller / bundle	New revenue line; “AI included” differentiation vs other rails	Volume: the only realistic path past ~500 dealers	20–30% of MONTHLY only — never discount the setup fee (it funds growth). Requires GRR >90% evidence first. Trinity keeps the dealer contract, brand, and billing — no revenue-share that cedes billing (aligned 5 Jul with the finance-tab F12 guardrail).
5 · White-label	Their brand on our product	Volume without identity	Refuse. The Meistertelefon/fonio model trades the customer relationship away — fatal to a logo-count exit story. Walk away; offer shape 4 instead.
6 · Exclusivity	A moat against the twin platform	A bigger cheque, maybe	Refuse early; price it absurdly if pushed later. Exclusivity with one Carlyle platform kills the other as buyer AND partner. Only conceivable inside an actual acquisition conversation.

R4 · Who to talk to, when — the contact plan tied to dealer milestones

Milestone	Action	Who, exactly
Now (0 dealers)	Jacob builds MotorDesk integration (permissionless). Max studies the Dragon2000 case (already queued). No outreach.	Nobody — build first.
~5–10 live dealers	Listing requests to every rail our pilots actually run (V9 decides). Kerfuffle self-serve listing. Light-touch intro note to Dragon2000 (“building on the indie stack, customers on DragonDMS, wanted to introduce ourselves” — 4 sentences, no ask).	MotorDesk/DealerKit partner contacts (generic ok); Mark Cooper (Dragon2000) — intro only.
~25 dealers	Garage Hive/TechMan listing (garage lane creative unlock). First joint case study with the most-shared rail. AM ecosystem presence begins.	Vendor partner teams; AM Live planning.
~50 dealers + GRR >90%	Formal A-tier conversations — referral/co-marketing pitch with retention + outcome data. Both Carlyle platforms in the same fortnight (symmetry).	Duncan Josey (ClickDealer/Centeca CSO) · Mark Cooper (Dragon2000/Lacour).
~150 dealers	Reseller/bundle negotiation with whichever rail referred most; the other twin informed the same week. Pinewood partner-programme entry (acquirer surface).	Same doors; Mat Hanson (CFO) appears if it turns commercial-structural — that’s the M&A tell.

First-meeting shape (30 min): 5 min — their world (ask what their dealers demand about AI; listen for build plans). 10 min — live demo of a SHARED customer: the agent answering, then the booking appearing in THEIR system. 10 min — the exchange: listing + intro flow offer, referral fee, white-glove promise, monthly outcome reporting. 5 min — one concrete next step (listing copy sent today). Materials checklist before any A-tier meeting: live integration · 2–3 named shared dealers · one-page outcome sheet (calls answered, bookings created, revenue per dealer) · retention figure · the build-vs-partner one-pager.

R5 · Objection scripts + rails-specific risks

The four objections

“We might build this ourselves.” — “Voice is a product line, not a feature: telephony, latency, call QA, per-dealer tuning. AutoLeap took a year with $50M raised. We’re live on your API today, revenue to you this quarter, zero engineering. Even 15,000-shop Tekmetric partnered for this layer (Pie).”
“You’re too small.” — “That’s why it’s free to say yes: a listing costs nothing, the intro flow is opt-in, and your dealers already asked” (show the shared customers).
“White-label or nothing.” — decline; offer bundle-with-brand (shape 4). If truly nothing: keep the integration live anyway, revisit in two quarters with bigger numbers.
“Exclusive or nothing.” — “We’ll be your best partner, not your only vendor’s captive”; escalate only inside an acquisition conversation.

Risks specific to this motion

Channel conflict with our own ads: a rail-referred dealer must pay the same price as an ad-acquired one (referral fee absorbs the difference). Never undercut the channel publicly.
Carlyle twins share an owner: assume notes travel between Lacour and Centeca. Say nothing to one you wouldn’t want the other to read.
Arriving second in garages: ReceptionMate is already on Garage Hive/TechMan rails — expect them to have partner goodwill; differentiate on scope, don’t badmouth.
Dependency: the ~40% concentration guardrail; the direct ads machine never stops — it is the leverage in every rail negotiation.
Time sink: rails talk is seductive and slow. Max’s time-box stands: one conversation-advancing action per week until the ~50-dealer trigger.

Finance rails — the market under our dealers

The executive read · one screen, judged from the founder seat

What this market is

Finance partners fund a real slice of dealer software — but not most of it. The finance calculators on dealer websites are usually lender-paid, and two brokers fund entire free websites and tool bundles: DSG (~3,100 dealers, at 3+ financed deals/month) and Auto Union (case by case). Most dealers still buy their core software themselves — the free layer is a loyalty weapon, not the norm. We censused all ~22 providers: exactly one tool they give dealers can talk to customers (MotorSales.AI's text assistant in DSG's bundle — it replies and, by its own claim, chases leads on WhatsApp, SMS and email; it has no phone capability). Nobody's free stack answers a phone.

What it means for Trinity

The frame: sell revenue capture (the answered phone), never software — dealers already pay 4-figures for demand and £99–259 for answered calls.

A second buyer exists: a finance partner can fund Trinity at £70–240/dealer/mo depending on tier (~£120 is our working example) and make roughly 3–4x back per extra financed deal — same product, someone else pays.

Will dealers pay? Yes, in our judgment — dealers already pay in this price band for similar things. What's missing is proof from dealers exactly like ours; the first pilots exist to create it, with monthly "money captured" reports as the evidence.

What we do — and the clocks

Now (costs nothing): record which finance partner every pilot dealer uses (the funnel tracker's stack column), and ask EZY Access — the dealer whose free website started this research — the three lock-in questions in F2.

At ~10 dealers: start the partner ladder (F9). In short: prove Trinity creates financed deals at dealers a partner already knows, and let the partner's own field reps carry that proof inward. No cold pitching before then.

Two timers and one alarm: MotoNovo, a car-finance lender holding ~10% of the market (its own claim), is being sold — which puts its dealer relationships up for grabs over the next ~12 months, and no rival lender has moved yet. The alarm we check weekly: MotorSales.AI adding voice to its free bundle. And Germany shows the speed limit — fonio went from zero to 7,000 customers in 21 months; the UK's version of that hasn't happened yet.

DSGthe perk-model inventor

Funds MotorSales.AI (its own affiliate) + Spidersnet websitesDoor taken — its AI seat is in-house

Centecaowned by Carlyle

Owns Evolution Funding (finance broker) and Click Dealer — the dealer-software company that is our top partnership targetRuns NO free-perk program despite owning the software — we can bring them the playbook

The big lendersClose Bros · Santander · Black Horse

Buy dealer tools from vendors (Codeweavers, iVendi, ATG/Keyloop)Proven perk-BUYERS — Trinity can be bought the same way

The plumbingCox/Codeweavers · iVendi

Neutral quoting tech everyone rentsNo phone AI has connected to either — a first-mover seat

Sequencing, unchanged: nothing here pulls a minute from Phase 1 (Meta ads) — the finance rail is a door that opens around the first 10–20 dealers; everything below is so we walk through it prepared. How to read this tab: Act I explains how the market works · Act II judges whether we can win · Act III says what we do, and when. Depth is deliberate; repetition with this summary is deliberate. Facts are sourced (canonical: work/strategy/finance-rails-research-2026-07-04.md + the WTP and Germany docs); modelled numbers and judgment calls are labelled in place.

Act I · How this market works

F1 · Four numbers that carry the whole story

60–85%

of dealer enquiries arrive by phone (Mediahawk; the 85% figure is from a 2016 study of 200+ UK retailers — treat the range, not the top, as the claim)

£2,181

average UK retail margin per used car (Cox Automotive Dealer Auction monitor, Dec 2025) — what one lost buyer costs

19%

of used-car sales carry point-of-sale finance (FCA, 2024) vs 80%+ of new — four in five used-car buyers are NOT financed at the dealer; every provider wants that headroom

3/mo

financed deals a month is what DSG asks of a dealer for the free website + £299-value software bundle — loyalty, bought monthly

F2 · The machine — how software becomes "free"

1 · Lender pays commissionper financed car

A fixed fee per deal, worth a few hundred pounds. The exact 2026 rate is published nowhere (industry-wide); the FCA's last analysis put flat fees in a band around ~£700. We ask a partner directly in the first meeting.

2 · Broker keeps a marginand spends it on loyalty

DSG, the proven example: £620m advances, 33,034 cars/yr, ~3,100 dealer partners, ~30 field repsShares part of the commission with the dealerBuys the perk stack with the rest

3 · Dealer gets perkswhile he qualifies

Free website (Spidersnet et al. — DSG pays the provider directly)Free finance calculators (Codeweavers/iVendi)Free MotorSales.AI software bundle (£299/mo list)Review + social-posting tools

The arithmetic, walked (our model)

Assume a conservative £300–500 commission per financed deal. (The exact 2026 rate is published nowhere. Secondary analyses of FCA-era data put flat fees near ~£700 — we deliberately model well below that, since the figure is dated and we'd rather understate the case.)
A dealer at the 3-deals/month threshold earns the broker £900–1,500 a month.
His perk stack costs the broker £250–400 a month — roughly a quarter to 45% of that commission. The ratio gets better for every dealer doing more than three deals. The model clears with room.
Most DSG dealers never hit 3/month — they average ~10–11 financed deals per year. The machine is real; the fully-subsidized dealer is a productive minority.
Good news twice: most dealers we meet have less free tooling than the headlines suggest, and the perk budget concentrates on exactly the dealers a partner would fund Trinity for.

What nobody publishes — the three EZY questions

What happens when a dealer drops below the threshold or leaves is disclosed nowhere. Only a dealer can answer. (EZY Access = the Ruislip dealership from the June roadshow that triggered this research.)
1 · "Ever had a quiet month under three deals — what did DSG say?"
2 · "If you switched finance partner, do you keep the website and pay Spidersnet yourself, or does it go away?"
3 · "Was the website a separate thing you signed, or part of the finance agreement?"

F3 · The market map — every provider that matters, sized

The denominator: UK used-car point-of-sale finance was £21.8bn across 1.41m cars in 2025 (FLA). Behind it: 106 regulated lenders (the FCA counts "over 100"), ~11,000 broker permissions (mostly dealerships themselves), and heavy concentration — in new-car finance the top 10 lenders hold 90% of volume; the used-car segment is where the long tail of smaller lenders lives. Nobody publishes named market shares; blocks below carry the best verified number for each provider (loan book, volumes, or dealer count), and the one stated share is MotoNovo's own ~10% claim.

The giantsfranchise/new-car heavy — mostly above our segment

Black Horse (Lloyds) · market leader, book not published; £1.95bn redress provision is the scale proxy · 4,500+ dealersCaptives (VW FS, Ford Credit ~$7.7bn UK, Mercedes ~£6.4bn, BMW, Toyota, Stellantis) · dominate NEW-car finance, structurally absent from independent used

In motionthe consolidation wave — redress-driven

MotoNovo/Aldermore · ~10% of UK car finance (self-claim) · FOR SALE BofA advising, bidders eye the SME book not motorSantander Consumer UK · 3,500–4,500 retailers · provisions ~£640m · ring-fencing read as pre-sale positioningBlue Motor · sale process (BNP Paribas, Aug 2025)Startline · sale candidateV12/Secure Trust · EXITED book sold to LCM for £458.6m, Feb 2026 — exits complete, at a premium

The mid-market lendersour segment's heart

Close Brothers Motor · £1.99bn book (FY2025, trade-press sourced) · 5,300+ partners incl. small independents · shrinking book, growing dealer offerNorthridge (BoI) · ~€3bn book (late 2024), shrinking (~£800m off; £170m FY25 loss)Mann Island (Investec) · ~£1.1bn book · 1,000+ dealers · stableAlphera (BMW's non-captive arm) · 1,200 partners

The specialistsnon-prime — deepest independent-dealer relationships

Advantage (S&U) · £317.1m receivables, deals +44% YoY growingOodle (KKR) · 300+ dealers (their own site, Jul 2026) growingFirst Response (ITOCHU) · 2,750+ dealersMoneybarn (Vanquis) · £709m, deliberately shrinkingMarsh · small (<£15m turnover), family-run

The broker layerintermediaries — they place deals with the lenders above; don't double-count

Centeca (Evolution + Click Dealer, Carlyle) · £6.5bn advances facilitated/yr · 5,500+ retailers · owns our A1 software rail growingDSG · £620m/yr · ~3,100 dealers · the perk-model inventor · stableZuto · £62m turnover +26% · Bridgepoint majority Nov 2025 growingCarFinance247 · >£54m turnoverCarMoney · £309m brokered 2025 · 1,500+ dealers · ITOCHU-owned May 2026

Do they fight for dealers? Yes — continuously

A dealer multi-homes across a panel of lenders and routes each deal to whoever offers the best rate, acceptance, and payout speed that week.
So loyalty must be re-bought monthly — that is what the perk stacks are for.

And the fight is about to intensify

The redress bill is forcing consolidation: one exit completed (V12), one giant in a sale process (MotoNovo), one ring-fencing (Santander), two specialists on the block.
Fewer lenders will fight over the same dealers — every survivor's loyalty problem gets bigger. An AI perk none of them has is exactly the class of weapon this fight buys (Act III).

F4 · Who owns what — the four clusters that matter

Cluster	What it holds	Why it matters to us
Carlyle / Centeca	Evolution Funding (largest used-car finance platform, 5,500+ retailers) + Click Dealer (our A1 software rail, 1,550+ dealers) + Automotive Compliance (regulatory umbrella for 300+ dealers)	Our top software rail is a finance broker's asset — we pitch "more financed deals per dealer," in their currency.
Carlyle / Lacour	Dragon2000 (our A2 rail, 1,000+ dealers)	Carlyle's second, separate bet on the same space. Treat both doors symmetrically (see The rails tab).
DSG Group	Broker + own lender (Unity) + MotorSales.AI — DSG in all but name (registered at DSG's address, DSG's CEO on the board, no FCA permissions of its own)	DSG's "bundled AI tool" is in-house — that door was never winnable. But their setup is the legal template we reuse in F11: the software layer needs no FCA permission at all.
Cox Automotive / Auto Trader	Codeweavers (the calculators on most dealer sites; public API; lender-funded; 30-day terms) · AutoConvert (Auto Trader's finance CRM, sold separately)	Neutral plumbing. Codeweavers/iVendi are integration candidates, not threats — and no phone AI has connected to either yet.

Census finding (4 Jul), elaborating the row above: despite owning a DMS/website business, Centeca does NOT run the DSG perk model — its whole software cross-sell to finance dealers is a £124 setup discount. The opening for Trinity: offer Centeca the playbook DSG invented. DSG had to pay outside website companies (like Spidersnet) to run its perk program; Centeca already owns Click Dealer, so it could deliver the same program in-house — with Trinity as the AI piece.

Words used on this tab — plain-English definitions (click to open)

Lender: holds the loan on its balance sheet (Close Brothers, Santander).
Broker: sits between dealer and a panel of lenders, places each deal, earns commission (DSG, Evolution/Centeca, CarMoney). Most dealers themselves hold a limited FCA permission to introduce finance.
Commission / flat fee: since a 2021 FCA rule change, lenders pay a fixed fee per financed deal. Before 2021, dealers could inflate the customer's interest rate for a bigger cut ("DCA") — that's what the scandal is about.
The redress scheme: is the FCA's plan to compensate customers who were overcharged between 2007 and 2024 (about £9bn industry-wide); currently part-suspended in court. It concerns the past — today's flat-fee model is lawful and settled.
FCA permission: the licence needed to arrange or advise on finance; software vendors avoid needing one by running on the dealer's own permission.
Rail / A1 / A2: our term for a partner whose installed base can distribute Trinity (see The rails tab). A1 (ClickDealer) and A2 (Dragon2000) are our two top software rails. We also cap concentration risk: no single rail may ever source more than about 40% of our new customers.
GRR: gross revenue retention; >90% is our own bar (rails doctrine) before offering any partner deep commercial terms.
IONOS: a large web-hosting company whose generic £9–29/mo AI receptionist marks the commodity price floor for undifferentiated voice AI.
Connex AI: the conversational-AI vendor CarMoney already uses on its own service lines.
EZY Access: the Ruislip dealership (June roadshow) whose free DSG-funded website triggered this research.

Act II · Can we win here?

F5 · What the "free" tools actually do — and don't

We audited all eight free-or-bundled tools dealers get. The conclusion first: with one exception, every free tool just captures a lead's details and hands them to a human. The exception is MotorSales.AI's "Theo," which genuinely replies, claims bookings, and chases leads — on text channels only. Nothing free touches a phone call. The table is the evidence:

Tool (how dealers get it)	Replies to enquiries?	Phone?	Books on its own?	Follows up?
Auto.Social · Car Dealer Reviews (free via DSG)	No — outbound posts / review collection	No	No	No
Finance calculators & toolkits — Codeweavers, iVendi, Santander, MotoNovo, Close Brothers (free)	No — customer self-serves a quote; a human must call back	No	No	No
Spidersnet Autopromotor website/DMS (free via DSG)	No — staff-operated; its chat add-on is Visitor Chat = paid humans	No	Staff-keyed	No
AutoTrader bundled tools	Basic Chat is free but the dealer's own staff must answer it; the automated backstops (Chat Assist, Chat Manager) are paid add-ons	No	Deal Builder accepts submissions for humans	No
MotorSales.AI "Theo" (free via DSG at 3+ deals/mo; £49 entry tier; £299/mo core list)	Yes — genuinely autonomous on WhatsApp/SMS/email when the dealer switches it on. Stock-aware; runs Sundays (two dealer case studies)	No — confirmed (full product/press/app/job-ad sweep)	Claims automatic test-drive booking (text channels)	Yes (claimed)

The one real overlap, sized

Theo is the only free tool doing real AI work — on typed channels. Autonomy is dealer-controlled (a per-lead "ENGAGE AI" toggle in their UI — not always-on); stock-aware; runs Sundays.
But it asks the dealer to replace his website/CRM (not bolt on), and its traction footprint is negligible: no Trustpilot, an app with too few ratings to score, ~3 named customers (one is the founder's own dealership).
Most DSG dealers never hit the 3/month bar (F2) → dealers walking in with a working free AI text-responder = single-digit-to-low-teens % of the UK's ~15.5k independent dealers (IBISWorld 2026). Most have nothing, or a manual chat widget.

Census-grade, and watched

The census (below) makes the whitespace a fact, not a claim: across ~22 providers, exactly ONE product responds to customers autonomously (Theo, text-only) — plus a one-way welcome SMS (Black Horse) and paid human live-chat (Visitor Chat). Everything else: quoting, calculators, dashboards, leads-for-humans-to-call.
The tripwire — actively monitored: MotorSales.AI shipping VOICE inside DSG's bundle turns this from context into direct threat. It sits in the weekly competitor radar by name.

The full perk-stack census — all ~22 providers: software offered, who builds it, comms overlap (click to open)

Provider	Software offered to dealers	Built by	Comms?
DSG	Full perk stack: website, calculators, reviews, social, MotorSales.AI bundle (free at 3+ deals/mo)	Buys/affiliates: Spidersnet, Codeweavers, iVendi, Auto.Social, MotorSales.AI	responds-text (Theo) — the market's one real comms tool
Close Brothers	Showroom quoting, calculators, data dashboards, e-learning, stock funding, Pod Point EV perk (all free/bundled)	Buys: Codeweavers, iVendi, Experian, AutoTrader, Pod Point	none
Black Horse	Dealer portal, Finance Online quoting, Portfolio 3Sixty retention leads, toolkit, e-learning	Buys: Autofutura/ATG-Keyloop + in-house	one-way welcome SMS only; leads are for humans to call
Santander	Dealer Hub, free Digital Toolkit, Gateway 3Sixty AI renewals	Buys: ATG/Keyloop — the same "3Sixty" family as Black Horse: one vendor holds that seat at two giants, the exact analogue of the seat Trinity wants for comms	none
MotoNovo	Dealer Hub, MotoPro, Quote & Propose, Self-Serve, free CSAT surveys, Lender Switch	Buys: Codeweavers + in-house	none
Alphera	FAMOS quoting, Partner Hub, end-of-contract Lead Management, eMaster segmentation, training academy, Consumer Duty+ code	In-house/BMW Group	none — leads surfaced for humans
Northridge	NETSOL Transcend wholesale platform + dealer app, JV profit-share	Buys: NetSol	none
Mann Island · First Response · Marsh · Startline	Portals (MIDOS / Dealer Portal / SalesDot / Insights analytics) + integrations, marketing collateral, accreditation schemes	Mixed: Codeweavers, iVendi, DealTrak, ClearSky, in-house	none
Blue · Jigsaw · CarFinance247	Blue: in-house "Dealer App Store" + ML decisioning · Jigsaw: in-house Optimus POS, white-label forms, embeddable calculators (free, commission-funded) · CF247: in-house portal + DealTrak	In-house builders — lower buy-a-perk propensity	none / capture-only
Oodle · Paragon · Moneybarn · Billing · Zuto	Generic application portals or nothing at all (confirmed)	—	none
Auto Union Finance	The second broker-funded website program after DSG: Autoweb Design builds AU dealers a finance-integrated site, live in ~1 week; "AU may be able to help fund it" (their page keywords: "Free Car Dealer Websites"). Discretionary, softer than DSG's — proof the perk playbook is spreading beyond its inventor	Autoweb Design (third-party)	capture-only
Platform sweep (vendor side)	iVendi's Finance Navigator panel now includes newer online-only lenders (Zopa, Admiral Car Finance, Tandem, Ayan Capital) — they compete through platforms rather than field reps, so they can't use perks the way the big lenders do · DealTrak F&I platform (~1,100 dealerships) · NextGear stock-funding app · Payment Assist free aftersales BNPL (integrates Garage Hive/TechMan — garage-lane relevant)	iVendi, DealTrak/AutoProtect, Cox, Payment Assist	none / capture-only — the sweep double-confirmed: no lender-funded AI comms anywhere
Evolution / Centeca	In-house finance tech (quote/soft-search APIs); Click Dealer cross-sell is only a £124 setup discount — NOT the DSG model; Automotive Compliance ARs get Evolution tech embedded; Click Dealer chat = Visitor Chat (paid humans)	In-house + owns Click Dealer	capture-only
CarMoney	Quotation apps over Codeweavers/AutoConvert/iVendi/Experian; Connex AI on its own service lines only; no new tools post-ITOCHU yet	Buys	none dealer-facing

Build-vs-buy, the strategic sort: the proven tool-BUYERS (DSG, Close Brothers, Black Horse, Santander, Northridge, First Response, Marsh, CarMoney, Startline) already pay third-party vendors for dealer perks — they can buy Trinity the same way. The in-house builders (Blue, Jigsaw, Evolution's finance tech, Alphera, CF247) would sooner build than buy. Our whitespace ranking survives this test: #1 Close Brothers and #2 ITOCHU cluster are both buyers.

F6 · Will they pay? The verdict

What dealers already pay (evidence)

Demand: AutoTrader costs four figures a month for as few as 10–29 advertised cars (named case: Spencers Car Sales, an 80-car dealer that can only afford a 29-car package). Price rises of 8% then 5.5% two years running; ~100 dealers downgraded or cancelled in the recent backlash. They pay multiples of our price — resentfully.
Operating software: MotorDesk £79/£149/£269 a month; Dragon2000 from £99.95 per module. Our band is their normal band.
Answered calls: Moneypenny — 2,000+ automotive clients paying ~£99–259/mo for HUMAN answering, now pushing an automotive AI receptionist. The behaviour "dealer pays monthly for answered calls" already exists at scale.
Capture-only chat: Visitor Chat — 2,000+ live accounts, price published nowhere (a field task: ask a prospect who runs it).
Text AI: priced £49 (MotorSales.AI entry) to £299 (its core list). We enter an existing price band, we don't invent one.

The phone gap, in numbers (UK)

60–85% of enquiries ring the phone (F1 caveats apply).
The funnel leaks three times: across ~750,000 UK dealership sales calls — 20% never answered; of the answered, 40% never logged anywhere; 40% of callers offered no next step.
21% of contacts arrive 6pm–9am; 55% of buyers expect a reply to email/SMS/WhatsApp enquiries within 3 hours or they look elsewhere.
Callers are buyers: at least 28% of dealership callers go on to purchase, and phone converts ~4x email — a 2019 US study (the only conversion data that exists; treat as directional, not UK fact).
Do the maths for a 20–40-car dealer (our model): selling 20–40 cars a month with most buyers arriving by phone implies roughly 70–140 buyer-intent calls a month (using the 28% caller-to-buyer rate above). One in five unanswered means 14–28 missed calls a month. If even two or three of those were buyers who never called back, that is £4,000–6,000 of margin at risk every month (at ~£2,000 per car — the F1 figure of £2,181, rounded down). No single study computes this end-to-end; every input is individually sourced, the chain is ours.
The recovery is measurable: ReceptionMate's best published garage case study shows £21–34k captured at one site (its other cases range ~£6.5–29k) — the receipt format we copy from pilot #1. (Status, honestly: no Trinity pilots have run yet — Phase 1 is launching. These receipts are to be created, not claimed.)

The frame decides the fight

If we sell this as "software / AI chat," we compete against £49 tools, the £9–29 commodity floor, and free bundles — a fight we lose.
If we sell it as "revenue capture on the phone," we sit in the same mental bucket as AutoTrader and Moneypenny (money in, not admin) — competing with a £25k/year hire and £2,000 lost margins per missed buyer. That fight we win.
"Isn't voice a thin layer?" Raw voice IS thin — the £9 floor proves it. Our product is the full loop: calls+texts+emails, stock-aware, booked into their diary, followed up — on the channels where enquiries actually arrive. Text-response commoditising to £49-and-free proves the defensible layer was never text: it's the phone + the loop + integrations.

The honest caveat — and its test

The category is proven in UK garages and trades, priced (not yet traction-proven) in UK dental, and proven at scale in US dealerships. Not yet for UK independent car dealers. Nobody has scaled it here — the square is empty because the category is ~18 months old everywhere, funded players went to easier-to-count verticals first, and thin cheap attempts starved.
The proof burden is ours: receipts-first.
The falsification test (proposed for the validation board): if pilot dealers price-compare us to £49 tools instead of missed deals, it's the FRAME failing — cheap to detect, early, decisive.

F7 · Where we win — and the three objections, pre-scripted

Five differentiators, in pitch order

The phone. Majority of enquiries, zero free-stack coverage. First line of every ad: "Who answers your phone at 6pm?"
End-to-end autonomy. Booked into their actual diary/DMS — not a scheduling link, not a Monday-morning pile, not a toggle-assisted draft.
Bolt-on, not rip-out. We plug into the stack they have; the free alternative asks them to replace website+CRM.
One agent across calls+texts+emails. The free stack covers fragments of typed channels; unification is the product.
Receipts. Calls answered / viewings booked / £ captured, monthly, from pilot #1 — the ReceptionMate format, applied to dealers.

The objections

"I get MotorSales.AI free from DSG." → "That's WhatsApp and email replies — good ones. It doesn't ring. Most of your enquiries do. Keep it; we cover the phone."
"There are £49 tools." → "£49 tools capture messages. What did your last missed call cost? Your margin is two grand a car."
"AI can't handle my customers." → "Two thousand dealerships already pay Moneypenny for answered calls. We do it stock-aware, booked straight into your diary — and you can hear every call."

F8 · Germany vs UK — is anyone playing an easier game?

Short answer: no. Germany is somewhat easier to sell into (no free-software economy — German dealers pay for everything; louder staffing pain) but harder to defend in (the category wave already broke there), and on pure demand the two markets are near-equal. The load-bearing rows:

Factor	Germany	UK	Easier for…
Free-tooling economy	None. Banks hand dealers single-bank finance tools only; dealers pay for DMS (€40–300) and mobile.de (€390–1,235/mo) themselves	Broker-funded free stacks (DSG model)	DE to sell — but the UK broker layer is also our potential FUNDER, which Germany lacks entirely
Labour pain (the ad hook)	Louder: ~18,000 unfilled positions, 93% of workshops understaffed	Real but less quantified	DE
Digital readiness	Weaker: only 37% of dealerships use ANY AI	Higher SMB baseline	UK
Category competition	fonio.ai: 7,000+ customers by Jun 2026, €14.6M seed at €120M valuation, serves the auto vertical + several vertical players. GetCarl: F4 Fund-backed (confirmed), traction self-reported, zero named customers	No fonio-equivalent at scale yet; Superchat entering; commodity floor forming	UK for Trinity — the horizontal wave broke over DACH; ours is still offshore

The decisive read: the UK's free-stack headwind is narrow (F5) and comes bundled with a funding channel Germany doesn't have. The true DACH lesson is velocity — fonio went 0 → 7,000 customers in ~21 months. The UK's fonio hasn't happened yet. That clock, not GetCarl's comfort, is what we're racing. Full comparison: germany-uk-structural-comparison-2026-07-04.md.

Act III · What we do about it — and when

F9 · The way onto the rails — the ladder, and the offer

You don't get on a finance rail by pitching it — you make a dealer they know into your proof and let their own field force carry it inward. (The cautionary tale: MotorSales.AI, the one AI tool in any broker's bundle, is DSG's in-house affiliate — incorporated at DSG's own address with DSG's CEO on the board from the start. DSG chose to own its AI layer rather than rent one, which is why that door was never winnable and why our guardrails below are pre-committed: brokers may prefer owning to renting.)

Step 0 · Nowcosts nothing

Funnel tracker records every pilot dealer's finance partner + perks (the stack column). With Close at 5,300 and CarMoney at 1,500 partners, our first 10–20 ads-acquired dealers will include partners of several — the funnel data picks the door

Step 1 · Receiptsin THEIR currency

"At [dealer X — your partner], Trinity handled N enquiries, booked M viewings, and K of the resulting sales were financed — through you." Plus: how often callers asked about finance (the agent counts it free)

Step 2 · The field forcethe warm way in

The pilot dealer shows his account manager ("look what answers my phone — your deals went up"); the rep reports it up. Every partner runs field reps: Close Brothers' account managers, DSG's 30 Connectors, Marsh's regional team. A dealer-instigated intro beats any pitch

Step 3 · Learn smallprotect the A-door

First formal conversation with the smallest receptive partner (Marsh or CarMoney) to learn the questions. Close Brothers — the prize with the perk precedent — gets approached with rehearsed pitch + solid receipts

Step 4 · Three rungsask small, grow the ask

Rung 1: co-marketing / a preferred rate for their dealers — costs them nothing and gives them something to announce as their first AI perkRung 2: funded pilot — "fund Trinity at 10 of your dealers for 6 months; monthly attribution report"Rung 3: the per-dealer subsidy at scale (F12)

The offer, on one card

They pay: £70–240 per dealer per month by tier (our £100–300 list minus a 20–30% channel discount, on the monthly only; ~£120 is the working example).
They get: more financed deals per dealer (the receipt proves it), a perk no rival lender has, dealer loyalty re-bought monthly, and a monthly attribution report — plus a compliance upgrade: every call recorded and structured, which makes dealer phone conduct auditable for the first time (Consumer Duty gift).
The dealer gets: Trinity free (or at partner rate) while he qualifies — and if the partner ever stops paying, direct billing at list. Deliberately the opposite of the website lock-in dealers fear.
We keep: the dealer contract, our brand, our billing, the setup fee at full price, and the freedom to do the same deal with every other lender.

Why they'd say yes (their math)

One extra financed deal a month is worth about £400 in commission to them, against a roughly £120 subsidy for Trinity — about a 3.3x return before counting the loyalty effect. (Our model; we confirm their actual flat fee directly in the first meeting.)
Post-redress consolidation means fewer lenders fighting for the same dealers — every survivor's loyalty problem grows.
The MotoNovo land-grab: ~10% of the market's dealer relationships come loose within ~12 months, and nobody has moved. "First AI perk in your bundle, in time for that fight" works even on an AI-skeptic.
Non-exclusive by design: the goal is the Codeweavers position — the neutral AI layer many lenders fund, captive to none. Worth more at exit than any single deal.

F10 · The whitespace — who to talk to, ranked

None of the eight researched has any AI perk in its bundle. Every named person independently re-verified — verification caught five stale names across the wider research, two of whom would otherwise appear in this table. Ranked by receptivity × reach:

#	Partner	Why this rank	The door
1	Close Brothers Motor Finance · 5,300+ partners	Already runs the perk model (multi-year Pod Point EV-charging deal for its dealers, May 2026; iVendi partnership Dec 2024). Investing through ~£300m redress pain — motor is core, not for sale. Pitch: "your Pod Point deal, but for the phone line."	John Cassidy, MD Sales recheck · Matt King, Head of Partners recheck · Seán Kemple, CEO verified
2	ITOCHU cluster: CarMoney + First Response · 1,500+ + 2,750+ dealers	Fresh owner (May 2026) merging a digital broker with a lender's dealer network; CarMoney already trusts conversational AI internally (Connex AI). Timing: 12–18-month integration cycle.	Andrew Marshall, Head of Marketing, Partnerships & Propositions verified · Alastair Grier, CEO verified
3	Marsh Finance · small	Highest receptivity per pound: family-run, CEO personally fronts tech partnerships, its MVP accreditation scheme is already a perk bundle. Small reach is the trade — and why it's the learning conversation, not the prize.	Andrew Marsh, CEO verified · Sean Ryan, Head of Field Sales verified
4	Santander Consumer UK · 3,500–4,500 retailers	Tech-perk-friendly (free Digital Toolkit; AI renewals platform with Keyloop) but in organisational flux (provisions ~£640m, ring-fencing signals, new leadership). Big prize, uncertain moment.	Steve Franklin, Commercial Director verified · Adam Goldhagen, CEO verified
5	Northridge (BoI)	Still investing (new platform Jan 2026) but bruised (£170m FY25 loss) and its perks aim above our segment.	Spencer Halil, MD verified · James Dempster, Head of Sales verified
6	Mann Island (Investec) · 1,000+	Healthiest lender here but thinnest perk stack; people-not-tech culture; MD door closed (resigned Mar 2025 — caught by verification).	Andy Muir, Head of Sales verified
7	Advantage (S&U)	Healthy and growing, but ~90% broker-routed — no direct dealer panel to perk. Route via its introducer brokers if ever relevant.	Karl Werner, CEO (ex-MotoNovo) verified

F11 · Two live windows — and the regulatory question, answered

The MotoNovo window — real, undefended

FirstRand is selling MotoNovo's parent Aldermore (Apr 2026; BofA advising). ~10% of UK car finance with a tooling-rich dealer offer.
Reported bidders (Lloyds, Shawbrook) want the SME book, not the motor business; a 13-lender sweep found zero public campaigns for MotoNovo's dealers. An undefended flank, not a battle we missed.
Watch triggers (in the monthly scan): first lender campaign aimed at MotoNovo dealers; an announced Aldermore buyer and what it does with the motor book.

Can our agent quote finance in-call? Partner, don't build

The industry norm: the software holds no FCA permissions. Codeweavers and iVendi run on the dealer's licence, registered at setup. MotorSales.AI is unregulated — DSG's licence carries everything.
When dealer pull demands it: integrate Codeweavers or iVendi under the dealer's own licence — plumbing, no Trinity permission needed. No phone AI has connected to either yet — a first-mover seat.
Until then: "finance options are available on this car — want a quote when the team calls back?" keeps us cleanly on the legal side.

Reference: the full regulatory path, priced (click to open)

Three routes if Trinity ever wants to earn finance commission itself: Introducer Appointed Representative (~£0, weeks — but scope excludes live-call quoting; fine for passive lead-passing only) · Full Appointed Representative under a principal (motor-retail principals: Automotive Compliance — a Centeca/Carlyle asset — or ITC Compliance; unpublished pricing, ~4–6 weeks + 30-day FCA notice; allows advising/arranging) · Direct limited permission for credit broking as a secondary activity (~£500–1,500 application, ~6-month FCA determination, ongoing Consumer Duty burden). Quoting a live APR on a call is likely "arranging" in FCA terms — beyond the cheap route. Verdict stands: partner now; direct permission is a post-G2, dealer-pull-triggered option. Full detail: research doc §11.

F12 · The deal — three structures, guardrails pre-committed

Structure	Shape	Anchor economics	Trade-off
A · Flat per-dealer subsidy (the DSG template)	Partner pays a fixed monthly fee per enrolled dealer; partner sets and polices its own volume threshold	Our list is £100–300/mo; a 20–30% channel discount takes the partner's price to £70–240 per dealer by tier. Against the £900–1,500+ that dealer's deals gross them monthly, that's roughly 5–25% of commission — around 10–15% at the typical mid-band	Predictable, simple. Risk: the partner becomes a gatekeeper — concentration cap applies
B · Per-deal bounty	Partner pays per financed deal attributable to a Trinity-handled enquiry	£50–100 per deal; ~2–4 attributable deals/mo matches our price floor	Perfect alignment; but needs attribution plumbing and makes our revenue volatile
C · Hybrid — our opening position	Base £75–100/mo per dealer + £25–50 per attributed deal, capped ~£300/mo; dealer pays for upgrades beyond the funded tier himself	Floor for us + aligned upside for them; our ARPU growth stays off their bill	Slightly more to explain; still needs the attribution report

Guardrails — pre-committed, never negotiated in the room

No exclusivity, at any price.
Trinity keeps the dealer contract, brand, and billing. The partner subsidizes our invoice; if they stop paying, the dealer gets direct billing at list — our anti-lock-in design, and a selling point.
Setup fee never discounted. Monthly discount capped at 20–30%; deeper only after retention proof (GRR >90%).
Partner-funded dealers count toward the ~40%-per-rail concentration cap. Aggregate reporting only — never raw conversations. 30-day rolling per-dealer enrollment inside an annual frame.

Cost & sequencing

Product cost of all this: a report, a tag, and a billing flag. Days of work — built only when a partner signs a pilot.
Sequencing, unchanged: nothing fires before live dealers and retention data. Ads first. Receipts first.

Internal: what this research changes on the other tabs (change tracker — click to open)

Where	What changes	Status
Rails tab · ClickDealer row	Finance pitch-currency annotation ("more financed deals per dealer")	applied
Scoreboard · wedge + risks	Commodity-floor mechanism split (cheap-standalone vs free-with-financed-deals); empty-square nuance + fonio velocity clock; new tripwires in the kill table; wedge sized with the three-stage funnel leak	applied 5 Jul (v20, 43-agent cross-tab audit + joint review)
Scoreboard · validation board	The frame test (software-vs-staff price comparison) as falsification card V10; V1 strengthened by WTP evidence; V2 frame note	applied 5 Jul (v20)
Vision tab · WATCH	Add proximate tripwires: MotorSales.AI voice; fonio/Superchat UK scale-up; CRM-lite bolt-on shape constraint	applied 5 Jul (v20)
Acquired tab	MotorSales.AI in the world-2 vendor list; Centeca + Automotive Compliance corrections; Evolution figures updated to £6.5bn/5,500+; broker-owns-vendor pattern on the watchlist (n=1, not yet a buyer archetype)	applied 5 Jul (v20)
Rails tab	R3 reseller position aligned with F12 (billing never ceded); finance partners named as a second rail class; data-boundary guardrail added; F9 precedent corrected (MotorSales.AI is DSG's in-house affiliate from origin — the earlier "founder proved it, DSG took board control" story was unsupported by the research and has been removed)	applied 5 Jul (v20)
Competitor map (separate artifact)	MotorSales.AI capability re-read + £299 list; GetCarl F4-funding confirmed/traction company-origin; fonio 7,000+ customers, €14.6M at €120M	queued
Scoreboard · funnel tracker	`stack` column captures finance partner + perks per pilot dealer (starts with pilot #1)	in canon

Product vision — from inbound agent to the whole stack

What this tab is. The parked long-run map: where Trinity's value expands after the agent, analyzed against the two live reference companies running this play today — Owner.com (restaurants) and Pie (Main Street SMBs, $23.7M, ex-Square/Toast, 4 days out of stealth). This is back-of-mind material by design: nothing here gets built before its trigger fires (the scoreboard rule stands — no system-of-record code before G2; every expansion is pulled by evidence via V5/V6, never pushed by this tab). Analyzed 4 Jul 2026 from their live sites.

V1 · The two reference models — and what each teaches

Owner.com — the "operate the business" direction

Their stack today: AI website + restaurant SEO + listings + reviews engine → online ordering with smart upsells → branded app + loyalty + email/SMS/push automation → operations app + analytics.
Positioning: "keep all money from sales" vs the 25–35% aggregators take — they replace the parasite layer, not add a tool.
The lesson: enter by capturing demand, expand into the operating system, and monetize by absorbing spend that ALREADY leaves the merchant's pocket. The dealer version of that existing spend: websites, listings, AutoTrader fees, call answering, agencies.
Fit with the bet: this IS the stated long-run play (free CRM/DMS/calendar/website once paying dealers pull) — the analysis confirms the model, not just the ambition.

Pie — the "capture more demand" direction

Their stack today: Growth (hyperlocal channel presence — "always there when customers need you") · AI Front Desk (24/7 answering + bookings "in a voice and tone that fits yours") · AI Search (be the answer when customers ask AI assistants).
Distribution: through vertical software rails (Tekmetric's 15,000 shops) — the rails tab's thesis, funded at $23.7M.
The lesson: the receptionist as the ENTRY into a growth suite is being validated with real money right now — same wedge as Trinity, different expansion order (they went visibility-first, we go workflow-first).
The new idea we didn't have: AI-search visibility. When buyers ask ChatGPT/Gemini "best used pickup dealer near Leeds," someone will sell dealers the placement. Trinity's agent data (stock, reviews, response quality) is exactly the substrate. Not ours to build yet — ours to watch (the monthly scan already tracks Pie).

V2 · Trinity's expansion map — two axes from one agent

Everything grows from the same asset: the agent owns the dealer's conversations (every call, text, email — structured). That data feeds two expansion directions, and both multiply ARPU on the same logos (the wedge-table property: expansion multiplies ARPU, not logos — and NRR >100% is the #1 premium lever in the exit case).

Axis	The ladder	Reference	Why it's earned, not built on spec
Axis 1 · Capture more demand (the Pie direction)	Review requests after completed work → listings/local presence → outbound campaigns & dormant-lead reactivation (the V-Cast / AI Retain analogs) → AI-search visibility	Pie Growth + AI Search; Visitor Chat's V-Cast proves dealers buy outbound campaigns	The agent already owns the phone number, the conversation history, and the moment-of-satisfaction (booking completed) — each rung uses data we uniquely hold.
Axis 2 · Operate the business (the Owner.com direction)	CRM-lite auto-filled from conversations (V6, the bridge asset) → calendar/diary → DMS-lite → free website	Owner.com's full stack; the bet's stated endgame	V5 gates it: build only what pilot dealers ask for unprompted. Field observation says many run on legacy email — the pull may come fast, but it must COME. And when it comes, CRM-lite layers on the stack dealers already run (ClickDealer/MotorDesk/Dragon2000) — the bolt-on discipline, not MotorSales.AI's replace-the-website+CRM rip-out (their own model, per their product pages).

V3 · Build now · earn next · build later · watch

Horizon	What	Trigger
NOW (G1 phase)	The agent (calls, texts, emails) + booking + stock plumbing (MotorDesk API) + the funnel sheet + per-dealer COGS instrumenting. Nothing else.	Already triggered — this is the company.
NEXT — earned by pull	CRM-lite: the conversation log as a readable customer view (the V6 bridge asset — cheap because the data already exists) · post-work review-request flow (first Axis-1 rung, near-zero build)	V5: pilot dealers asking unprompted, logged and counted. Not a date.
LATER (post-G2, ~100+ dealers)	The system-of-record decision: CRM → diary → DMS-lite → free website (Owner-style free-tier framing). Likely what the operator-angel round part-funds.	G2 passed + retention proven + the raise. A decision point, not a commitment.
WATCH (not ours yet)	AI-search visibility for dealers (the Pie AI Search analog) · Pie's own trajectory as the leading indicator for receptionist→growth-suite expansion economics · MotorSales.AI shipping voice inside the DSG free bundle (added 5 Jul) · fonio/Superchat scaling into the UK (added 5 Jul)	Monthly outside-in scan already tracks Pie; revisit when they publish traction or a dealer asks for it. MotorSales.AI voice = the single event that turns the free-stack economy from context into direct threat — watched weekly, by name. fonio/Superchat = the first horizontal breaking out in the UK the way fonio did in DACH (0→7,000 customers in ~21 months) narrows the vertical window the same way.

The guardrail, restated. Building ahead of validation is the documented gravity well. This tab exists precisely so the vision is PARKED — visible, ordered, and inert until its triggers fire. The only expansion work permitted in the current phase is passive: log every unprompted dealer request (V5) and keep the conversation data clean, because that data is the option on everything above.