The multiple halved while operating profit kept growing — the entire investment debate

• De-rating, not de-growth. Down 42% in twelve months and 51% from the May 2025 peak, with trailing P/E at 14.6× against a nine-year average of 25.5× — the cheapest multiple in its public history. FY25 operating profit still grew 8% to £377M; H1 FY26 revenue rose 5% and net income 29%.
• What the bear is selling. The FY25 ARPR product lever printed £77 against a £120–130 guide and the stock lever inverted to −£22 from a +£20–£40 guide, framed by Investors Chronicle's December 2025 'Yellow Pages moment' headline and a CMA fake-reviews probe opened 27 March 2026.
• The peer the bull is buying. Rightmove — same UK marketplace template — trades at 18.6× P/E and 13.2× EV/EBITDA for similar margin and an inferior balance sheet. AUTO sits at 14.6× / 10.4× with higher ROIC, net cash, and a fatter buyback. The 25% gap is auto-specific AI fear, not a quality differential.

The 'Yellow Pages moment' frame ignores the dealer side that quietly became infrastructure

• Co-Driver AI hit 85% retailer adoption in year one. Combined with 1.8 million monthly retailer logins and 91 million monthly API calls, AUTO has stopped being a single-sided ad layer and become the operating system for 14,013 UK forecourts — integration depth no buyer-side AI agent can replicate.
• Forecourt count at an all-time high. 14,013 dealers today versus 13,452 at IPO in 2015, through a decade of UK dealer consolidation, COVID, the chip shortage, and the arrival of generative AI. Dealer count is the first dial that would move if disintermediation were real — it is moving up.
• The product-lever miss was one experimental SKU. The £77 vs £120–130 print sat almost entirely in the per-transaction Deal Builder fee, which has since been folded into the core advertising bundle. The price lever delivered at the top of its range and the stock lever was cyclical and pre-warned in May 2022.

63% operating margin, net cash, and an 8% total cash yield while the multiple sits at a 15-year low

The marketplace runs at near-100% incremental margin — people, marketing and engineering costs are largely fixed while ARPR compounds 5–10% per year on roughly £174M of segment cost against £565M of segment revenue. Management is using the de-rating to retire stock at the lowest P/E since IPO: roughly £275M returned in FY25, ~13% of the float retired since FY21, and CEO Nathan Coe sits on a £24.7M personal stake worth 36 years of base salary. For the bear thesis to hold, the next twelve months must show ARPR growth slipping below GDP and the dealer count starting to fall.

FY26 results on 21 May resolve the de-rating in one print

Three lines decide it. The ARPR product lever (FY25 printed £77 against a £120–130 guide), H2 retailer revenue inside or outside the 5–7% range, and the FY27 guide that tells the market whether 5–7% is a one-year pause or the new normal. A product lever back above £120 forces a sell-side reset upward; a second sub-£100 print confirms structural deceleration.

The number nobody discloses. Direct and branded traffic share — last reported around 82% — is the entire AI-disintermediation defence, and management does not publish a year-on-year trend. Any voluntary slide showing it held above 80% reverses the bear's central claim; any visible slip toward the mid-70s validates the 'Yellow Pages moment' framing.

Then the calendar empties. The annual report in late June carries the Vanarama goodwill test (the brand's useful life was cut from 15 to 5 years in October 2023 with no charge taken). After that the next real catalyst is H1 FY27 in early November — leaving a four-month interregnum dominated by the ~£9M-a-week buyback floor and the 545p / 447p technical levels.