How AI answers your buyers,
on the record
Seenfact sends your buyers' questions to ChatGPT Search and Google AI Mode word for word, every day. It keeps each answer, its citations and the searches behind it, then turns them into rates, sources and a list of what to fix.
Answers · 2026-10-05
Total mentions 4 / 7excl. questions that name your brand
From sign-up to your first weekly report
Set it up once. Sampling runs every day, and you review it every Monday.
- 1
Set up your brand
Enter your brand name, website domain and other spellings, choose one market and language, and add up to three competitors.
- 2
Confirm your questions
We draft candidates from your site and sales FAQs. Confirm them into a fixed set, or write your own.
- 3
Run a baseline
Collect today's answer to every question on every surface. Results usually arrive within 10 minutes, and sampling runs daily after that.
- 4
Review every Monday
A weekly report arrives at 09:00 on Monday, and the team works through the opportunities.
Six pages, one chain of evidence
Every page in the console traces back to the same stored answers.
A fixed set of buyer questions
Each question goes to every surface word for word. Category, use-case and comparison questions are tracked apart, and questions that name your brand only check facts.
- Edit a single word and it becomes a new version, so trends restart cleanly
- Pause, resume or archive any question
- See where each question came from: suggested, sales FAQ or added by you
Best OpenRouter alternatives?
- Your brand
- Mentioned
- Your website
- Not cited
- Competitors in the answer
- Portkey, LiteLLM
Evidence fingerprint SHA-256 9c41e07a…d3b27b2f
Behind every number, a stored answer
The answers page shows each question on each surface, day by day. Open any cell to read the answer exactly as the surface returned it, untranslated.
- Your brand and competitors highlighted in the text
- Every citation marked as your site, a competitor or a third party
- The searches ChatGPT Search ran in the background to write the answer
- The raw response archived with a SHA-256 fingerprint, ready to download
Every rate shows its sample
The overview reports your mention rate and how often answers cite your site, by surface over 7, 30 or 90 days, next to each competitor's share of voice.
- Each rate shows k/n and a Wilson 95% interval
- Failed collections are counted apart, never as "not mentioned"
- A change is reported only when both periods have 30+ valid samples and the difference is significant
See which sites AI cites
The sources page groups every page AI cites for your questions by site, whether it's a third party, a competitor or you, and opens each site to its pages.
- Citations, answers and surfaces for every site
- Which competitors appear alongside it, and how often you're mentioned there
- Counted only on questions that don't name your brand
Gaps become the team's to-do list
The opportunities page pulls gaps from the last 7 days of samples. Each one says why it's there and what to do first, with the answers attached.
- Three kinds: competitors present without you, third-party pages worth winning, and surfaces that never cite your site
- Claim it, assign an owner, keep notes and mark it done
- When newer data no longer supports it, it expires and stays on record
From 09-29 to 10-05, 4 answers to this question named a competitor but not you.
Open the answers below. See which pages the AI cited and what reasons it gave when it named a competitor, then fill in what you're missing on your site or on those pages.
50 valid / 52 planned · 2 missing · 10 excluded (brand questions)
A report every Monday
The weekly report covers the previous seven full UTC days. It's ready at 09:00 on Monday in your time zone and emailed to every teammate who wants it.
- Valid, missing and excluded samples, counted
- Changes only when they're significant; otherwise "no reliable change"
- This week's new opportunities
- Written in Chinese or English
Built for a team
Work as a team
Invite teammates to read results, edit questions and work on opportunities. Only the owner changes workspace settings and billing.
Chinese or English
Use the console and get the weekly report in Chinese or English. Answers stay in the language the surface returned.
Edits never mix cohorts
Changing your brand, competitors or questions only affects later samples. New data is counted separately, and earlier data and answers are kept.
What Seenfact won't do
These limits are deliberate. They're why you can trust the numbers.
Read the methodology- No composite "GEO score"; just rates, sample sizes and intervals
- No model APIs standing in for the surfaces your buyers use
- No old answers or other vendors' data to fill a gap
- No promises about how AI will rank or recommend you
- No crediting a change in the numbers to any single edit
Start with a free scan
of 5 questions
One plan at $99/month: 1 brand, 3 competitors, 50 questions and 2 AI surfaces.
Scan for free