I'm Gen Furukawa, founder of SuperMarketers and an AI visibility strategist for B2B SaaS companies. I build AI visibility systems and run AI search audits across dozens of clients. Citation rate is the first number I baseline on every engagement, because it answers the only question that matters in AI search: when your buyer asks, does the engine say your name? Most of the companies I audit have never measured it. Their competitors are already in the answer.
What is citation rate?
Citation rate is how often an AI engine names your company across the set of queries your buyers actually run. If you build a list of 20 queries your buyers would ask while researching your category, and you appear in 6 of the generated answers, your citation rate is 30%. It is a presence metric. It does not measure traffic, clicks, or rank position. It measures whether you exist inside the answer the buyer reads.
This is a different question than SEO ever asked. Rankings assume the buyer sees a list and chooses. AI search collapses the list into a single recommendation, so the only thing that matters is whether you are named in it. That is why answer engine optimization tracks citation rate instead of position. The surface changed, so the metric changed with it.
How to calculate citation rate
The formula is deliberately simple, because the point is to track it consistently, not to make it complicated.
Two definitions decide whether the number means anything. First, a relevant query is one your buyer would genuinely run while researching your category or their problem, not a branded search for your own name. Second, a cited response is one where the engine names your company, not one where it merely discusses the topic. Score each response cited (1) or not cited (0), sum the citations, divide by the query count, multiply by 100.
Run the same query bank across each engine you care about. The engines weight sources differently, so score citation rate per engine and then blended. A company can be cited consistently in ChatGPT and invisible in Google AI Overviews. A single blended number hides that. Per-engine numbers tell you which surface to fix first.
Citation rate vs share of voice vs mentions
These three get used interchangeably and they are not the same. Citation rate is presence. Share of voice is dominance. Mentions is raw count. Confusing them leads to the wrong fix.
| Metric | What it measures | The question it answers |
|---|---|---|
| Citation rate | The percentage of relevant queries where AI names you at all | Do I show up in the answer? |
| Share of voice | How much of the named space you own versus competitors in the same answers | When companies get named, how often is it me? |
| Mentions | The raw count of times you are named, with no denominator | How many times was I named? |
You can have a high citation rate and low share of voice. The engine names you in most answers, but always last, behind three competitors. You are present but not dominant. Mentions is the weakest of the three on its own, because a raw count with no denominator tells you nothing about how often you were eligible to appear. Track citation rate first. It is the cleanest signal of whether you are in the game at all, and share of voice tells you how well you are winning it.
What's a good citation rate for B2B SaaS?
Here is the honest answer: there is no published market benchmark for citation rate. Anyone quoting you a specific industry-average percentage invented it. The metric is too new and the engines vary too much for a credible cross-industry number to exist yet. So do not anchor to an external figure. Judge citation rate against your own category and your own competitors.
What you can use is a directional frame from your own audits. If you appear in fewer than one in five of your primary category answers, treat it as a gap that is actively costing you pipeline, because your buyers are reading answers that name a competitor and not you. The practical target is appearing in the majority of the answers your buyers generate while researching your space. That threshold is a working read from the audits we run, not a published industry average, and you read it alongside the 9-dimension AI Visibility Score rather than as a number in isolation.
Citation rate is one output of a system. We read it alongside the 9-dimension AI Visibility Score, which grades technical readiness, content architecture, entity clarity, authority signals, and cross-engine performance. In our audits, most companies score 2-3/10 on the first run, with a single-digit citation rate to match. The benchmark target on the score is 7+. A citation rate that climbs while the underlying score stays flat is noise, not progress.
How to improve your citation rate
Citation rate moves when you change the structure of the pages an engine reads, not when you publish more posts. The factors that lift it are the same ones that decide whether you get cited at all.
- 01Answer-first structureA page that opens with a direct, 40-60 word definition gets extracted. A page that opens with a setup paragraph does not. Put the answer in the first 100 words.
- 02Entity clarityOne unambiguous description of your company, used verbatim across your site, LinkedIn, and external mentions. "An AI visibility system for B2B SaaS founders" is citable. "A full-service growth partner" is not.
- 03Third-party validationAI engines weight mentions on credible external sources. A company named by three independent sources is cited more often than one with a great website and nothing else pointing at it.
- 04Topical depthFive well-structured pages that each answer one specific buyer question build more citation-worthy authority than 20 thin posts on scattered topics.
The research supports the direction. The GEO study (Aggarwal et al., KDD 2024) measured how structural content changes affect AI engine visibility, and adding citations and authoritative source references improved visibility scores by up to 40% in the benchmark. The step-by-step version of this work lives in the guide on how to get cited by ChatGPT, which covers the structural changes that move citation rate within 90 days.
How to measure it
You do not need a paid tool to baseline citation rate. You need a fixed query bank, the four engines, and 30 minutes a month.
- Build a query bank.List 20 queries your buyers actually run: "best [category] tools for [use case]," "what is [your category] software," "how to solve [your buyers' top problem]." Branded searches for your own name do not count.
- Run every query across the engines.ChatGPT, Perplexity, Claude, and Google AI Overviews. Same query bank, every engine, so the numbers are comparable.
- Score cited or not cited.For each response, mark 1 if the engine names your company and 0 if it does not. Note which competitors appear when you do not.
- Calculate the rate.Citation rate equals cited responses divided by total relevant queries, times 100. Record it per engine and blended.
- Re-run it monthly.Same query bank, same scoring, once a month. This replaces "posts published" as your visibility metric. The trend line is the point.
The full walkthrough, with the exact queries and scoring sheet, is in the guide on how to run an AI visibility audit in 30 minutes. Run that once and you have your baseline citation rate and a list of the gaps costing you pipeline.