You added FAQPage schema. You wrote llms.txt. You rewrote your headings as questions. Then you sat there wondering if any of it actually worked, because there is no dashboard that just tells you. That gap between doing AEO work and knowing whether it paid off is where most site owners get stuck, and it costs them momentum on the exact project that was starting to pay off.
AI citation rate is the percentage of tracked, category-relevant questions where an AI engine names your site as a source in its answer. AEO Excellence, an Agentic Engine Optimization service, builds the audits and templates that move this number, so the question comes up in nearly every engagement: what number should you actually be aiming for?
What Counts as an AI Citation?
A citation happens when ChatGPT, Perplexity, Claude, or Google AI Overviews names your brand or links your domain while answering a question a real customer would ask. It does not count if the engine answers correctly but names a competitor, and it does not count if your page merely ranks in a linked source list nobody reads aloud.
To track it, you pick a fixed set of questions your buyers actually ask, run each one against each engine, and log which domain got named. Google Search Console shows some AI Overview referrals, but it will not show you what happened inside a ChatGPT or Perplexity conversation, so manual or tool-based tracking across engines is the only complete method.
Two citations that look identical are not always worth the same. A citation on a high-intent question a paying customer would actually ask is worth more than a citation on a vague, low-traffic phrasing nobody types. When you build your tracked question list, weight it toward the phrasing your sales team actually hears from prospects, not the keyword you assume ranks well.
What Is a Healthy AI Citation Rate?
The honest answer depends on how established your brand already is in the category. Category-leading brands typically get cited in 30 to 60 percent of relevant AI answers across major engines. Newer or smaller brands should expect a citation rate closer to 5 to 15 percent during the first six months of AEO work.
| Brand stage | Typical citation rate | What to expect |
|---|---|---|
| New or small brand, first 6 months | 5-15% | Early wins on long-tail, low-competition questions |
| Established mid-market brand | 15-30% | Consistent citations on core category questions |
| Category leader | 30-60% | Cited across most major engines for head-term queries |
The trendline matters more than the absolute number. Most sites see a measurable citation lift within 14 to 30 days of shipping a complete AEO foundation, because AI crawlers refresh their indexes far more often than Google refreshes its search index.
Do not treat these ranges as a pass or fail grade. A brand-new site with a 6 percent citation rate after two months is on a normal curve, not failing. A ten-year-old category leader stuck at 8 percent almost certainly has a structural problem, since its age and authority should already be pulling that number toward the higher band.
Why Does Citation Rate Vary by Engine and Query Type?
Each engine weighs structural signals differently, so your rate on ChatGPT will not match your rate on Perplexity for the same question. GPTBot typically revisits a mid-size site every 2 to 7 days, while Perplexity refreshes more aggressively for topics that are trending, which means a timely post can get cited on one engine days before another catches up.
Query type matters just as much as engine choice. Branded questions such as whether a company offers a certain service get cited far more easily than broad category questions like naming the best provider in a category, because branded queries have less competition for the citation slot. Track both separately, or your blended number will hide which category you actually need to fix.
A blended average also hides which engine is dragging your score down. A site cited in 40 percent of Google AI Overviews answers but only 5 percent of ChatGPT answers has a ChatGPT-specific problem, most often missing or malformed JSON-LD, since GPTBot leans harder on structured data than on page layout. Split your weekly log by engine from day one so this kind of gap shows up immediately instead of six months later.
Once your rate is low and you cannot tell why, guessing at fixes wastes the weeks you have left before a competitor closes the gap. An AEO audit checks the specific structural signals an engine actually reads: schema type by page, whether every FAQ block has a matching visible answer, and where question-style headings are missing entirely. That diagnostic step is what turns a flat citation rate into a plan with a fix attached to each gap. Our AEO FAQ page covers the exact schema types and crawler rules an audit checks.
How Do You Track Your Citation Rate Every Week?
You do not need expensive software to start. A spreadsheet and thirty minutes a week gets you a real trendline within a month.
- List your top 10 category questions, phrased the way a customer would type or say them.
- Run each question in ChatGPT, Perplexity, Claude, and Google AI Overviews.
- Log which domain got named for each question, including competitors.
- Calculate your citation rate: citations for your domain divided by total questions asked.
- Repeat the same 10 questions every week so the trend is comparable.
- Note which engine and query type moved, not just the blended average.
Tools like AthenaHQ, Profound, Otterly, and SE Ranking AI Visibility automate this same process once you outgrow a spreadsheet, but the manual version is enough to know if you are moving in the right direction.
Keep the question list itself stable for at least eight weeks before changing it. Swapping questions every time one looks discouraging destroys the one thing that makes this measurement useful: a comparable trendline. If a question genuinely stops being relevant to your category, retire it and note the date, so you can explain the jump in your log later.
What If You Do Not Have Time to Track This Weekly?
That is the most common reason sites give up on measuring AEO, and it is a fair concern if you are doing it by hand across four engines. The honest fix is not to skip measurement, it is to shrink the question set to your top 5 questions and check monthly instead of weekly, or hand the tracking to one of the automated tools above. A partial trendline beats no trendline, and a monthly check still catches whether a schema or heading change moved the number.
The other honest concern is cost of the automated tools once your question list grows past what a spreadsheet can track cleanly. Most AI visibility platforms price by tracked query volume, so start with the manual spreadsheet on your top 5 to 10 questions, prove the trendline moves when you ship a fix, and only pay for automation once you are tracking enough categories that manual logging becomes the bottleneck rather than the fix itself.
Get a Baseline Instead of Guessing
You cannot improve a number you have never measured. If you want a starting point instead of building a tracking spreadsheet from scratch, contact AEO Excellence for a homepage audit, which comes back with your current structural gaps and a prioritized fix list inside 24 hours.