Business

Bottom Line

Auto Trader is a near-monopoly UK car classifieds platform with structurally exceptional unit economics — 70% segment operating margin, 50% FCF margin, 60%+ return on capital employed. The consolidated 63% group margin understates the marketplace because it carries Autorama, a small loss-making vehicle-leasing brokerage; the asset to underwrite is the marketplace, not the group P&L. The market has just re-rated the stock 46% lower from its May 2025 high of 908p to ~493p on the fear that AI search disintermediates the buyer-side network — the right debate, but one where the evidence (75% of automotive marketplace minutes, 81.6m monthly visits, 10x the next classified competitor) still favours the moat.

How This Business Actually Works

Auto Trader sells advertising slots and adjacent software to ~14,000 UK car dealers, and the price per dealer compounds. Buyers come to the marketplace because it has the most cars; dealers pay because it has the most buyers. Everything else — Co-Driver AI tools, Deal Builder, valuations, finance leads — is incremental product layered onto that flywheel.

The economics are dictated by the ARPR equation: price × stock × product. FY2025 ARPR rose 5% to £2,854/month, with +£78 from the annual price event, +£77 from product (mostly Trended Valuations + enhanced Retail Check), and −£22 from the stock lever as cars sold faster than expected. With £174m of total Auto Trader segment costs against £565m of segment revenue, every additional pound of ARPR drops through to operating profit at near-100% incremental margin — people costs and marketing are largely fixed, software development is mostly expensed in-period, and depreciation is trivial (£6.3m).

The Playing Field

Auto Trader looks more like a UK property portal (Rightmove) than a US auto classified (CarGurus, Cars.com). Same playbook: dominant national network, fragmented supply side, multi-decade compounding ARPR, capital-light. The US auto peers operate in a more contested market against multiple at-scale rivals; the gap in operating margin and ROCE is structural, not a temporary execution gap.

The peer set tells you what dominance is worth. Two-sided national marketplaces with no credible rival cluster at 60–70% operating margins and trade at 13–18x EV/EBITDA. Contested marketplaces collapse to single-digit margins and trade at 7–13x. Carvana is the cycle-and-leverage trade — capital-intensive, retail margins, gigantic share-count and balance sheet. The right peer for AUTO is RMV; the right cautionary peer is CARS.

Is This Business Cyclical?

Less than the market thinks, and the cycle hits revenue, not transaction volume. UK used-car volumes have moved in a tight 6.5–8.2m band over 14 years; the car parc grows ~1% per year, the change cycle is 3–4 years, and supply lags new-car output by years. So the volume base is structurally stable. What does cycle is dealer profitability — when prices fall, dealer margins compress, dealers reduce paid stock, ARPR's stock lever turns negative, and ARPR growth slows to mid-single digits instead of high-single. That is exactly what is happening now (stock lever was −£22 in FY2025, was −£54 in H2).

The only real downturn in 14 years was FY2021 (−29% revenue) and that was a self-inflicted choice — Auto Trader gave dealers two months of free advertising during the COVID lockdown. Operating margin fell to 61% but never went near loss. The FY2023 margin dip is the Autorama acquisition, not a pricing-power event; the core marketplace ran at 70%+ throughout. The cycle is the pace of compounding, not the direction.

The Metrics That Actually Matter

Don't look at P/E, P/S, or EBITDA growth. Look at ARPR × forecourts. That product equals 80% of revenue and 100% of the differentiated economics. Watch the price lever (annual April pricing event), the product lever (Co-Driver AI, Deal Builder, Trended Valuations attach rates), and the stock lever (cyclical). If the price lever ever stops working, the moat is broken.

What Is This Business Worth?

This is one economic engine, not a sum of parts. Autorama is too small (£36m revenue, £4m loss) to justify SOTP — strip it out and AT segment is £565m revenue at 70% margin. The right valuation lens is earnings power times reinvestment runway: trailing P/E and EV/EBITDA, anchored by the durability of ARPR growth.

At the FY2025 close (£7.44), AUTO traded at 23.6x earnings, 16.4x EV/EBITDA, 4.6% FCF yield. At ~£4.94 today (May 2026), the multiple compresses materially — closer to 16x trailing earnings on the same earnings stream. The market is pricing in a structural growth slowdown, not just a cyclical one. If you believe ARPR keeps compounding at 6–8% and forecourt count stays flat, the current price implies an unusually generous setup. If you believe AI search permanently rewires discovery, the multiple may still be too high. The valuation question is not "what's fair P/E" — it's "is the buyer-side network actually breaking."

What I'd Tell a Young Analyst

1. Track ARPR by lever, not in aggregate. A 5% ARPR print can mean different things: +£78 price, +£77 product, −£22 stock is healthy and cyclical; +£40 price, +£10 product, +£100 stock would be late-cycle and fragile. The composition tells you whether the moat is widening or just riding the cycle.

2. Watch forecourt count quarterly. It is the first dial that moves if dealers ever decide they don't need this platform. A 13,452 → 14,013 path over 10 years is the moat. A drop below 13,500 with stable used-car volumes is the thesis-breaker.

3. Don't conflate the share price with the business. AUTO has fallen 46% from 908p (May 2025) to ~493p (today) on AI-disintermediation fear. Revenue, margin, and ARPR are still up — every operating KPI in Q2 FY26 was in line with guidance. The market is voting on a future risk, not on current execution.

4. The Autorama loss is a rounding error, not a thesis. £4m operating loss vs. £394m AT segment profit. It exists because management wanted optionality on digital retailing; if it stays loss-making, they will close it. Do not build a SOTP around it.

5. The thing that would actually break the thesis is buyer-side disintermediation — generative AI assistants (ChatGPT, Gemini) becoming the discovery layer for car buying. Today 18% of AUTO traffic comes via organic search; the rest is direct/app. Watch that organic-search share. If it climbs and AUTO can't get onto AI-assistant SERPs as a preferred result, the network unravels slowly.

6. The thing the market is probably underweighting is the dealer side. Co-Driver hit 85% adoption in its first year, dealers are now logging into the Retailer Portal 1.8m times per month, and API calls hit 91m/month. AUTO is becoming embedded operating infrastructure for UK dealerships — not just a place to advertise. That switching cost is what justifies the premium even in a worst-case buyer-side scenario.