What is Citation rate?

Citation rate is the percentage of AI answers, across a set of questions you track, where your brand or page gets cited as a source. The formula is dead simple: answers that cite you, divided by total answers you checked, times 100. Run 100 buyer questions through ChatGPT, get cited in 22 of the answers, and your citation rate is 22%.

That's the whole thing. It's the headline number for how often AI engines are pulling in your content and giving you credit. Now the useful details.

Why citation rate matters

Old SEO had one big number everyone watched: ranking. Where do I sit on page one? AI search broke that, because people ask a question and read an answer instead of scrolling a list of links. So the new question became, am I in the answer at all?

Citation rate answers that. It's the percentage of buyer-relevant prompts where you show up as a cited source, and it's about the cleanest signal of AI search authority you can get. A low rate means the engines basically don't know you exist. A high rate means they trust your content enough to lean on it again and again. And there's money attached: pages cited in AI answers tend to pull in more clicks and far more brand exposure than the ones sitting one slot away that never get named.

How to calculate citation rate

The core formula:

Citation rate = (answers that cite you / total answers checked) x 100

A few things make or break the number:

• Your prompt set. Citation rate only means something against a fixed list of questions. Pick the 50, 100, or 200 prompts your buyers actually ask, and keep that list steady so you can compare week to week.
• What counts as a citation. Most teams count a real source link or a clear named mention. Stricter setups only count clickable links to your pages. Pick one rule and stick with it.
• The engine. ChatGPT, Perplexity, Google AI Overviews, Claude, and Gemini all cite differently, so you usually track a separate rate per engine, not one blended figure.

What's a good citation rate

It depends wildly on the engine, which trips people up. A 2026 study of 34,234 AI responses found a 46x gap between platforms: ChatGPT cited brands just 0.59% of the time while Perplexity sat around 13%. So a "5% citation rate" can be amazing on ChatGPT and weak on Perplexity. Always read the number next to the engine it came from.

Within a single engine and a focused prompt set, rough benchmarks people use in 2026 look like this:

• Under 10%: basically invisible. The engine isn't reaching for you.
• 20 to 30%: solid, competitive territory for most B2B and SaaS categories.
• Above 40%: category-leading. You're close to a default source.

One more reason engine matters: Perplexity cites roughly 22 sources per answer while ChatGPT cites closer to 8. More citation slots per answer means more room to get in, which is why Perplexity rates run higher across the board. And only about 11% of domains get cited by both, so winning one engine doesn't hand you the others.

Citation rate vs mention rate vs share of voice

These three get mixed up constantly, so here's the clean split:

• Mention rate counts any time your brand name appears in an answer, link or no link. It's the loosest of the three.
• Citation rate is stricter. It counts when you're cited as a source, usually meaning a real reference or link, not just a passing name-drop.
• Share of voice is relative. It's your citations or mentions compared against your competitors' for the same prompts. Citation rate is your own absolute number; share of voice tells you who's beating you.

You can have a healthy citation rate and still be losing share of voice if a rival is getting cited twice as often. Watch both.

How to raise your citation rate

No magic switch here, but the levers are known. Answer the question directly in the first 40 to 60 words so engines can lift a clean chunk. Write self-contained passages that still make sense pulled out of context. Back claims with real stats, original data, and clear expertise, because engines favor sourced content. And earn mentions on sites the models already trust, like Reddit, Wikipedia, and reputable media.

The honest catch

Citation rate is noisy. The same prompt can cite you today and skip you next week, since engines tweak behavior with zero warning and add randomness to their answers. So a single snapshot lies. Run each prompt a few times, track the average, and watch the trend over weeks instead of panicking over one bad day. The direction of the line tells you more than any single reading.

FAQ