One score for AI visibility,
tied to your revenue.

An AI visibility score measures how often large language models like ChatGPT, Claude, Gemini, and Perplexity reference your domain when users ask buying-stage questions in your category. Attrifast rescans your seed prompts every 10 days on Pro (every 30 days on Starter), parses every answer for your domain plus competitor domains, and rolls the appearance rate into a single 0-100 number. A score of 38 means your domain showed up in 38 of every 100 sampled answers. Unlike a typical AI visibility tracker, the score is reported alongside the revenue those citation-driven sessions actually produced, so you can see whether the score correlates with paying customers.

Profound and Loamly stop at the citation count: they tell you that you appeared in N answers across M prompts. Attrifast takes the same scan output and joins it back to Stripe via the same session-stitching pipeline that powers our revenue attribution product. The result is a single dashboard row per AI engine showing citation share, referral sessions, paid trials, and revenue. That join is the part GA4 cannot do because chatgpt.com referrers get bucketed as Direct, and the part Profound and Loamly do not do because they have no read into your Stripe account.

ChatGPT (with web search), Claude, Google Gemini and Perplexity — all four are scanned on every run. Each gets its own sub-score because, as the Princeton GEO research shows, the levers that win a citation are not identical across engines: what wins a citation on one does not always work on another. Treating them as one bucket hides the signal you actually need to act on. One caveat specific to Perplexity — it is scanned and scored like the rest, but it is the one engine that does not report the background searches it ran, so it stays blank on the Fanouts tab.

On the $49/mo Pro plan you get up to 30 prompts monitored, rescanned every 10 days across ChatGPT, Claude, Gemini and Perplexity — Attrifast runs the scheduled scan and you can trigger a rescan from the dashboard at most once every 7 days. The $9.99/mo Starter plan covers up to 10 prompts, rescanned every 30 days. You bring the seed prompt list (we suggest a starter set based on your category) and Attrifast handles the scan, parsing, deduplication, and longitudinal storage. Every prompt comes with a presence trend chart so you can see when a model retrain dropped you from an answer or when a competitor displaced you. The scan output feeds the score, the share of voice number, and the revenue join.

Because a citation that sends no buyers is vanity. I have watched DR-80 sites get cited heavily by ChatGPT on competitive prompts and produce zero trials, because the visitors that did click were research-stage. I have watched DR-20 niche pages get cited rarely but convert at 2-3x because the citation arrived inside a buying-shaped query. The only way to tell those apart is to join the citation event to the eventual Stripe payment, which is the thing Attrifast was built to do. Without that join you optimize a score that does not pay the bills.

Yes — it is built for that case. The setup is one tracking script (4 kb) plus a restricted Stripe API key you paste into the dashboard, then you add the prompts you want monitored. There is no consultant onboarding, no annual contract, and no per-seat upsell. Pricing starts at $9.99/mo, and the Pro plan — 30 prompts rescanned every 10 days — is $49/mo, both with a 7-day free trial. Compared to Profound, which is sold via demo into Fortune 500 marketing teams, Attrifast is sized for the founder or growth lead who needs the data to make a budget decision this week, not next quarter.

Accuracy depends on two things: prompt coverage and parse fidelity. Prompt coverage is on you — if your seed list misses the queries your buyers actually ask, the score reflects the wrong universe. We help with that by suggesting starter prompts from your category and from your existing GSC and keyword data. Parse fidelity is on us: we extract every cited URL and bare brand mention from each engine's structured response (where available) and fall back to regex on the rendered answer. For engines that return structured source lists (ChatGPT with web search, Perplexity), parsing is effectively 100%; for Claude and Gemini answers, we hit 90%+ in our own validation.

Partly. The Gaps tab exports to CSV — every source, its citation and competitor-presence rates with the absolute counts behind them, the gap score, and the change against the previous window, so the numbers can be recomputed outside the dashboard. The other tabs are read-in-app for now, and there is no JSON API yet. Both are on the roadmap; if a specific export blocks your workflow, tell us, because roadmap order follows user demand.