Unit tests for AI search.
AI engines cite passages, not pages. We test the questions your buyers ask against the pages that win them today, show the exact passage you lose with and why, and keep testing every week so a redesign or rewrite never quietly costs you an answer.
The problem
AI visibility dashboards tell you, weeks later, that you lost. They do not say which paragraph lost, what the winning page has that yours lacks, or which deploy caused it. So teams rewrite blind and regressions go unnoticed.
Who it is for
- Teams that already invest in content and want to know which pages AI actually cites
- SaaS, fintech and legal sites where one AI answer drives real pipeline
- Developers and SEO leads who want AI visibility checked in CI like any other test
What we test, every week
Each explainer is 15 seconds. They play as you scroll.
Who gets cited, and why
Your passages against your competitors’ for each buyer question, with the reason you lose.
Tests that block bad releases
A pull request fails when a change would cost you a citation.
Alerts with a reason
The week you lose an answer, you know which passage dropped and who took it.
Plans
Start free with the open-source tool, or have us run it for you. Monthly plans cancel any time.
Open source
Free
Run getcited yourself
- Citation arena, X-ray and audit
- Citation tests and GitHub Action
- Regression history in CI
- MCP server for your AI agents
Test Suite Setup
$490one-time
Baseline in 5 business days
- 25 buyer questions and your real competitors mapped
- Citation share baseline with the passages engines read
- Passage-level fix plan, ranked by impact
- getcited.yml and GitHub Action installed for you
- Credited to Citation Guard within 30 days
Citation Guard
$190/month
Most teams start here
- 25 questions tested every week
- Regression alerts by email: what changed and why
- Monthly live check of ChatGPT, Perplexity and Gemini
- Monthly 20-minute review call
- Cancel any time
Citation Guard Pro
$490/month
Up to 3 sites
- 100 questions across up to 3 sites
- Weekly regression alerts in email or Slack
- Two losing pages rewritten for you every month
- Monthly brand truth check: what AI gets wrong
- Priority support
More sites, agencies or a custom question set? Talk to us. Agencies can also resell it through white-label AI visibility.
What you get
Question map
25 to 100 buyer questions with the competitor pages that win them today.
Citation share baseline
Your share of each answer’s context against every competitor, with the passages engines would read.
Passage-level fix plan
Why each losing page loses: missing words, buried answers, no numbers, dead-weight sentences.
Tests in your pipeline
A getcited.yml and GitHub Action, so a pull request fails before it costs you a citation.
Weekly regression alerts
What changed since last week and why: lost passages, new competitors, blocked crawlers.
Live engine checks
Monthly checks of what ChatGPT, Perplexity and Gemini actually cite, compared with the tests.
Tested on the real web
We ran getcited on 27 pages that rank for 5 real questions, our own sites included. The raw data is public.
100%
citation share for one focused page
A big-brand article on the same question ranked #36 at passage level.
2 of 22
ranking pages turn AI crawlers away
robots.txt said allowed. The firewall returned 403 to GPTBot, ClaudeBot or PerplexityBot.
20%
quotable sentences on the median page
Four in five sentences on pages that already rank are weak or dead weight to an answer engine.
Common questions
The engine is the same open-source getcited. The service is the work around it: choosing the right questions and competitors, reading the results, writing the fixes, running it every week and telling you when something regresses.
Find out which answers you are losing, and why.
Get a free AI Answer Snapshot: five of your buyer prompts, tested across six engines, with the competitors named instead of you.