Promvia

Yesterday we switched on something I have wanted to exist for a while: a public page that measures, every Monday, which sources seven AI answer engines actually cite for a fixed panel of 40 buyer questions, and publishes all of it (prompt list, per-engine tables, movers, an engine changelog, JSON under CC BY 4.0). It costs us about EUR 2 a week to run. Page: promvia.app/ai-source-index

Week one, across the answer engines (ChatGPT, Perplexity, Gemini, Claude, Google AI Overviews, AI Mode; Bing organic kept separate):

• youtube.com was cited in 36 of 40 questions, reddit.com in 33. Those two are the spine.

• - Editorial and review publications came next: Forbes 10, TechRadar 7, Tom's Guide 7, RTINGS 5, CNBC 4. Mostly "best X" roundups, not news.

• - Review directories (G2, Capterra, TripAdvisor, idealo, AlternativeTo): under 1% of citations. On consumer questions the "alternatives to X" slot went to blog posts and vendors' own /vs pages, not marketplace grids.

• - About 80% of citations were the long tail nobody has a name for: vendors' own pages, retailers (Best Buy in 14 questions), niche blogs.

The engines eat different diets for the same questions. Gemini leaned on Reddit (26/40) and YouTube (25/40); AI Overviews on YouTube (24) and Reddit (18); ChatGPT's top sources were YouTube (7), Tom's Guide (6) and Forbes (5); Perplexity's were Reddit (6), Forbes (5), CNBC and the NYT; Claude cited almost nothing but review publications. "Reddit dominates AI citations" is true for Google's surfaces and not for ChatGPT or Perplexity.

Two things the first live week caught in our own product, which is the real reason to build in public:

1. Gemini's grounding API does not return the page it read; it returns an opaque redirect URL that 302s to the real source. We had been storing the redirect host as the "source" for weeks. Every Gemini citation in the customer product was effectively blank. Fixed the same day, backfilled the week, and we now log the day the fix landed so it never reads as a platform change later.

2. 2. Google AI Mode emits shopping cards (google.com/search?...) next to real citations. Not sources. Filtered.

Caveats we print on the page: one sample per engine per question, so a move of one or two questions is noise; source types come from a rule list, so an unknown domain lands in "other"; movers and the observed changelog only start with the second week (this Monday).

What I would like from IH: if you run a product site, look up your own category's questions on the page and tell me whether the citation set looks like yours. And if you think the panel is missing a question shape, say which; the panel is versioned, so additions are new ids, never edits.

Business status, since this is IH: launched August, EUR 0 MRR, one paying-tier customer would cover the index for a year. Building in public, honest numbers only.

Hi IH — I'm Anil, solo and bootstrapped, building Promvia: it checks whether ChatGPT, Perplexity, Gemini and Google's AI answers actually name your product when buyers ask for a recommendation, and what those answers send your site. Launched August 2026, €0 MRR so far, building in public. Disclosure up front: the numbers below come from our own tracking.

The thing worth sharing is a finding, not a feature. We track a fixed set of buyer prompts every week across six answer engines and log every URL the answers cite — about 2,150 cited URLs over 14 days in one niche (software tools). Two things fell out:

1. "Alternatives to X" prompts cite comparison-shaped pages (paths like /vs, /alternatives, /compare — vendor pages and blogs included, not just G2 or Capterra) in 34–43% of citations. "Best X" prompts: 2.8%.

2. When we first counted by directory DOMAIN instead of page shape, the effect looked ~10x smaller. The domain list was a bad proxy. What gets cited is the page type, wherever it lives.

The practical version: a well-structured /alternatives or /vs page on your own domain competes for the same answer slot as a G2 grid. "Get listed on directories" and "build comparison pages" are sold as two tactics; at the citation level they look like one. And the cited pages had almost no organic traffic — winning the SERP and being chosen for an answer are different games.

Caveats, honestly: one niche, two weeks, URL-path labeling rather than rendered pages. Direction solid, magnitudes indicative. We pre-registered a follow-up (vendor-authored vs third-party comparison pages) and will publish it on Sep 7 whatever it says.

We eat our own cooking: Promvia's own domain scored 0 on day one, and we publish our own weekly run live at promvia.app/live so the score is a public number rather than a promise.

Question for people running product sites here: do your /vs or /alternatives pages pull AI citations (or AI referral traffic) out of proportion to their organic traffic? That publisher-side number is the one nobody seems to have.