Do AI assistants recommend your SaaS when a buyer asks for your category?
We ask ChatGPT, Gemini and Claude the questions your buyers ask ("best survey tool for a startup", "Typeform alternatives", "X vs Y"), several times each. The report shows how often each one names you, who it names instead and which sites it cites. Then it gives you 5 fixes written as developer tickets.
Three full sample reports
These are complete Standard audits of three public self-serve SaaS sites (25 buyer prompts × 3 samples per engine), run on 2026-10-07. They were made from public pages only, without the companies' involvement, and we are not affiliated with them.
Formbricks (survey software)
Named in 8% of ChatGPT answers, 35% Gemini, 79% Claude.
HTML report · PDF · tickets
Papermark (document sharing / data room)
Named in 23% of ChatGPT answers, 35% Gemini, 77% Claude.
HTML report · PDF · tickets
Plausible (privacy-friendly analytics)
Named in 68% of ChatGPT answers, 79% Gemini, 77% Claude.
HTML report · PDF · tickets
What the report covers
- Share of answer: how often each engine names you and your competitors, shown as mean and min–max over samples. Answers vary from run to run, so every number shows its spread.
- Prompt-level grid: each buyer question, which brands were named, and the rank.
- Crawlability for AI bots: robots.txt rules for GPTBot, OAI-SearchBot, ChatGPT-User, PerplexityBot, Google-Extended and ClaudeBot, plus meta robots and sitemap.
- Schema and entity: JSON-LD (SoftwareApplication, Organization, FAQPage, Offer), titles, consistent naming of brand and category.
- Third-party footprint: the review sites, listicles and forums the engines cite, counted for you and your competitors.
- Pricing and landing page: whether plans and prices are readable without JavaScript, and whether the persona, trial and call to action are clear. Every finding quotes your page.
- 5 developer tickets, ranked by impact × confidence ÷ effort, each with evidence, the change, acceptance criteria and effort (S/M/L). Delivered as Markdown and CSV too.
Packages
| Package | Buyer prompts × samples | Pages crawled | Competitors | Delivery | Price (USD) |
|---|---|---|---|---|---|
| Starter | 15 × 3 | up to 15 | 3 | HTML + PDF report, 3 days | 199 |
| Standard | 25 × 3 | up to 25 | 5 | HTML + PDF + tickets (Markdown/CSV), 3 days | 249 |
| Direct | 25 × 5, plus a re-measure 30 days later | up to 40 | 5 | Standard + 30-day re-measure + 30-minute call | 750 Buy |
Starter and Standard will be sold through Fiverr and Upwork; links appear here when the listings go live. The Direct package is paid by card through Stripe: Buy the Direct audit (USD 750). Enter the site to audit at checkout; after the report is delivered we book the 30-minute call.
Method and limits
Engines measured today: ChatGPT (via the Codex CLI with web search on), Gemini (via the Antigravity CLI, no search requested) and Claude (via the Claude Code CLI with web search on). Each report names the exact model and tool behind every engine.
Perplexity is not measured in current audits. It shows as "not measured" in the report, with no numbers. Google AI Overviews are not measured either.
These are the assistants' answers through their developer tools, not screenshots of chatgpt.com or gemini.google.com. Consumer apps can answer differently (personalization, location, model version), so treat the numbers as a measured proxy, not what every buyer sees.
Public data only: we never ask for Search Console, analytics, CMS or repository access. The crawler obeys robots.txt, stays on your domain and sends at most one request per second. We do not scrape G2, Capterra or marketplaces. Referring domains are not measured.