Trustpilot: Selling Trust as the AI Answer Engine's Tollbooth

Trustpilot: Selling Trust as the AI Answer Engine's Tollbooth

For most of its listed life Trustpilot has been a subscription reviews business that happened to sit on a very large pile of consumer opinion. The H1 2026 call reframes the whole thing: the pitch is no longer "read our reviews," it is "be the source the machines cite." That is a genuine change in emphasis, and the keyword set confirms it — the top of the curated list for the quarter is dominated by large language models, a freshly coined cited review platform claim, and the mechanics behind it, server logs.

CEO Adrian Blair led with it: “Trustpilot is the number one cited review platform globally, according to research by Seer Interactive. We saw a more than 400% year-on-year increase in ChatGPT citations.” — Adrian Blair, Chief Executive Officer · 2026-09-15 The supporting infrastructure metrics are equally specific — a Semrush Domain Authority score of 100, up from 95, and server-log telemetry that captures every inbound request from a model. It is unusual for a mid-cap to have the receipts to make an AI claim like this rather than a slide-scale assertion, and Trustpilot leans on both external research (Seer Interactive) and its own internal logs.

The flywheel, now spinning for robots as well as people

The framing is that the same loop — more reviews create more impact, which motivates more reviews — now runs through AI search. Users submitted 20% more reviews year-on-year, the dataset crossed 400 million, and new product reviews pages are the direct monetization hook: in beta across 400+ businesses they produced an average 15x lift in AI citation growth. That is the kind of number that turns a feature into a priced add-on module.

When it comes to brand metrics, trust is at the absolute top of the pyramid. It has to be earned, not given. I can't say, "Trust me."

That customer testimonial — Navy Federal Credit Union's CMO — is the whole commercial argument in one line. If a brand cannot self-certify its own trustworthiness to a model, it needs an independent third party. When asked directly whether answer engine optimization converts into pricing power, Blair was unambiguous: “Is it supporting willingness to pay? Absolutely. It's helping in those conversations. It's helping with our GRR.” — Adrian Blair, Chief Executive Officer · 2026-09-15

The financials back the narrative rather than contradict it. Bookings rose 18% in constant currency to $171m, adjusted EBITDA climbed 46% to a record $26.3m, and the margin reached 17.4%, up 2.8 points, keeping management on its stated path to a 25% margin in 2028 and 30% by 2030. Revenue was $151.4m at an 83% gross margin. The enterprise mix continues to compound: customers paying more than $20,000 a year now contribute 47% of bookings, up from 27% in H1 2023, and that cohort's ARR is up 35% year-on-year. This is the large enterprise shift that turns a high-churn SMB product into a sticky, expanding one.

Where the story gets less flattering

Two things complicate the clean narrative. First, the net dollar retention rate slipped to 101% from 103% as previously flagged package migrations annualized, and North America's headline 27% growth was flattered by multinational accounts re-billing out of the U.S. rather than genuine incremental penetration — Blair conceded this was a 1-2 point drag on reported U.K. growth. Second, and more awkwardly for a business whose product is trust, the company disclosed a technical error in how it demonstrated distributable reserves underpinning prior buybacks. Hanno Damm's framing: “It is a technical issue that we need to cure, but it does not change any of the numbers or any of the cash balances or whatever.” — Hanno Damm, Chief Financial Officer · 2026-09-15 The remediation lands in Q4 and blocks a fresh buyback announcement until completed.

On cash, Hanno was explicit about the modeling heuristic — “using adjusted EBITDA margin and then taking a point off of that as a proxy for free cash flow for the full-year is a good starting point” — Hanno Damm, Chief Financial Officer · 2026-09-15 — which implies a materially stronger H2 given the H1 bonus drag and tax outflow. That is a sensible way to read a 5% adjusted free cash flow print against 46% EBITDA growth.

Is Trustpilot riding a wave or catching one?

Scanning the broader market context, the AI-infrastructure theme is loud and crowded — global momentum keywords skew toward open weight model chatter and the compute build-out, and recent reporters like Adobe (agentic software) and Oracle are selling the same AI-transformation story. Trustpilot's angle is subtly different: it is not selling compute or agents, it is selling the independent signal that those systems consume. That is a picks-and-shovels position on answer engine optimization, and it is genuinely company-unique in this dataset — no other recent reporter is claiming the AI-citation tollbooth.

The stock's own price history was not supplied, so the tape cannot confirm or deny whether the market has already voted on this narrative — a caveat worth flagging given the market cap of roughly $833m. What the transcript does show is a business deliberately rerouting its go-to-market for scale: a sales reorganization that outsourced top-of-funnel business development, added a starter plan for digital acquisition, and split customer success from commercial account management. Management was candid that this carried opportunity cost during H1, and the 76% North America new-business leap is the rebuttal. The one number to watch into FY27 is whether group NDR trends back above the current 101% as the enterprise mix, with its superior retention economics, continues to compound.

For now, the story has moved from "reviews subscription" to "the independent trust layer AI recommends." That is a better story. The reserves disclosure is the kind of footnote that only matters if it recurs.