VerticalScope Stops Buying Eyeballs and Starts Selling Its Forums to the Machines
The trade nobody wanted to make
VerticalScope is a ~$29 million market cap owner of roughly 1,300 enthusiast forums — car, home-improvement, hobby verticals where people who actually own the thing argue with other people who actually own the thing. For two quarters, the growth story was Audience Engine: buying traffic on third-party platforms and monetizing it programmatically. In March, the CEO framed it as a free lunch — “it is bringing in MAU, and it is also driving revenue as well. So it's the best of both” — Christopher Goodridge, Chief Executive Officer · 2026-03-04 — and in August the CFO insisted the solution “basically funds itself” — Vincenzo Bellissimo, Chief Financial Officer · 2026-08-14.
It didn't. By the end of Q2 the receivables had grown to $20 million — traffic acquisition costs paid up front, collections from partners later — against users that, by management's own admission on this call, monetized at roughly a tenth of the core audience. So the plug was pulled in the tail end of August. “We pulled back from Audience Engine and used the working capital that freed up to pay down debt.” — Christopher Goodridge, Chief Executive Officer · 2026-11-11 Reported MAU averaged 77 million in Q3 and, with no paid traffic in Q4, will step down again — the company was explicit that the entire quarter-over-quarter drop in MAU came from that one decision, and that the underlying core audience has been stable.
That is the honest version of the story, and it deserves credit for being honest. But note what it concedes: two quarters of the company's incremental strategic energy went into a working-capital-intensive traffic arbitrage that the CEO had to publicly disown. You never say never, he allowed about reviving it — but the investment dollars are now pointed elsewhere.
A data deal that is really two deals
The elsewhere is data licensing. VerticalScope signed its first data agreements with two "Frontier AI Labs" in October — described, without names or amounts, as "real players in the space." The first is a non-exclusive training license over a selection of communities; delivery finished in early November and the fees are being settled on contributed token volume by year-end. The second is variable and usage-based, paying out when VerticalScope content shows up in an AI-generated answer — “it's a calculation on the contribution to an answer” — Christopher Goodridge, Chief Executive Officer · 2026-11-11 — and started producing revenue this month. The CFO says it will likely be itemized as its own P&L line.
Here is the tension. In August, management described this exact negotiation as structurally recurring: “It's more -- it's meant to be an ongoing arrangement, not a onetime event.” — Christopher Goodridge, Chief Executive Officer · 2026-08-14 The deal that actually got signed is half onetime by construction.
One is more recurring in nature. The other has -- because it's mainly -- most of the economics are focused around pretraining against a select segment of our community data, so that will be a lumpier amount that comes in, in Q4.
Analysts tried three separate ways to size it and got three separate non-answers; the closest thing to guidance was that it could be a meaningful part of the quarterly results and that it “flows through almost entirely to the bottom line.” — Christopher Goodridge, Chief Executive Officer · 2026-11-11 That is the whole bull case in one sentence: near-100% incremental margin on an asset that has already been built. It is also the whole risk: a pretraining check is a spot sale, not an annuity, and it will make Q4 look better than the underlying business is.
What the cost cut and the balance sheet actually buy
The quarter itself was soft. Revenue of $13.1 million fell 11% year-over-year and 5% sequentially: programmatic down 13% to $6.1 million on lower search impressions and softer summer CPMs, direct down 14% to $4 million as a handful of custom-content campaigns failed to repeat, e-commerce flat at $3 million (23% of revenue, up 10% excluding Ritual). The one genuinely encouraging operating number is that AI-driven commerce recommendations lifted affiliate commerce revenue on Fora by 29%. Adjusted EBITDA was $4.8 million at a 37% margin versus $6.2 million at 42%, and net income was a thin $0.1 million.
Cash is the reason this equity is interesting at all. Operating cash flow was $6.1 million, free cash flow $3.7 million at 77% conversion, roughly $16 million over the trailing twelve months. Against a ~$29 million market cap, that is a startling multiple — but it is a multiple on a shrinking revenue base, which is precisely why the November realignment matters: an estimated $4.4 million of annualized cost removed, with about two months of savings in Q4 and roughly $1.6 million of one-time termination costs booked outside adjusted results. Net debt sits near $26 million at 1.4x leverage, with the revolver at $31 million and $69 million undrawn.
Read the two together and the strategy is legible: shrink the cost base faster than revenue declines, throw the difference at Applied AI ($1.2 million deployed year-to-date against a $2 million plan), and let the resulting free cash flow de-lever the balance sheet toward 1x. Management is not promising growth from the core — the CEO's own medium-term framing was 5% to 10% organic, "historically," if the initiatives work.
The tape disagrees with the thesis
Here is the part worth flagging. frontier AI labs is a global market keyword, and it has been one of the more brutally sold themes in the last month — down roughly 30% across a basket including CoreWeave-adjacent and neocloud names. The public market has spent the autumn de-rating the exact counterparty class that just validated VerticalScope's content. VerticalScope is therefore selling into a demand signal that is real at the operating level and dislocated at the capital-markets level.
And the operating demand signal genuinely is real: management disclosed that in the last 30 days, AI agents and AI search engines made 115 million attempts to access its communities. None of that shows up in MAU. All of it is the reason a training license has value. Notably, none of the other reporters in the last five days — a cohort consumed by data centers, memory pricing, tariff refunds and de novo facilities — is signing AI data licenses. This theme is company-unique, not sector consensus, which cuts both ways: no confirmation, but no crowding either.
What to watch is not Q4 revenue, which will be flattered by a lumpy pretraining payment and a seasonal ARPU lift (already visible at $0.057 from $0.045). It is whether the usage-based agreement expands beyond its initial subset of communities and whether the "productive discussions" with potential partners convert. The onetime nature of deal one buys optionality, not a growth rate. If the second deal stays small, this is a cheap cash-flow stream with a litigation kicker against OpenAI still grinding through the courts; if it scales, it is a data royalty on the open web. The market, at 1.4x leverage and ~1.8x trailing free cash flow, has priced almost none of the second outcome.