To compare your AI visibility with competitors, ask the AI engines the same set of category questions your buyers ask, on a schedule, and count how many of those answers name you and how many name each rival. That count, your share of answers, is the headline number. The useful part sits underneath it: which rivals are cited instead of you, on which prompts, and which pages the engines read when they chose them. Cituna runs this benchmark every day across seven engines (ChatGPT, Perplexity, Gemini, Claude, Grok, Google AI Overviews and Google AI Mode), puts you and your tracked rivals on the same answers, and turns each prompt you lose into a fix.
The seven steps below work whether you use a tool or a spreadsheet. Where Cituna does a step for you, the guide says what it shows and where to find it.
The three kinds of competitor in AI answers
A competitor in AI search is not always the company your sales team names. Engines decide who counts, and they do it per question. Before you measure anything, separate three relationships, because each one calls for different work.
| Kind | What it means | Where you see it in Cituna | What to do about it |
|---|---|---|---|
| Co-mentioned | The rival is named in the same answer as you | “Cited alongside you on N prompts” in a rival’s head-to-head | Watch the order: who is named first, and what the answer says about each |
| Cited instead | The rival is named in an answer that leaves you out | “Cited instead of you on N prompts” and the “Where rivals beat you” list | The main gap: find the pages the engine read and get onto them, or publish the page that answers the prompt |
| Compared | The buyer asks about you and the rival by name (“X vs Y”, “X alternatives”) | Read per prompt, under “Brands named, in order” in each answer. There is no separate view for it | Make sure the comparison is accurate; these prompts name you, so they are kept out of share |
The first two kinds come from category prompts, questions like “best X for Y” that do not name anyone. The third comes from branded prompts. Keep them apart in your reading, because an answer to a question that already names you will usually mention you.
Step 1: Pick prompts that make a fair comparison
The benchmark is only as good as the questions in it. Use the questions a buyer types before they have a shortlist: the category question (“best tools for tracking AI citations”), the problem question (“why doesn’t ChatGPT recommend my brand”), and the need-specific question (“best X for a small agency”). Include the two or three questions you expect to lose. A set of prompts you already win flatters you and tells you nothing.
Keep branded prompts out of the share calculation. Cituna does this by default: prompts that name your own brand are left out of the competitor share, and the head-to-head notes how many were excluded and why. Our guide to choosing AI visibility tracking prompts covers how to write the set; ten real questions beat fifty generated variations.
In Cituna’s Prompts tab you can type your own, use Suggest prompts, or pull them From Search Console, and pause any prompt at any time. Plans include 10 prompts on Starter, 30 on Pro and 90 on Max, each checked daily on all seven engines.
Step 2: Choose the rivals to measure
Start with the brands the engines actually name, not your market map. After a scan, Cituna’s competitor manager lists the brands found in your latest scan so you can add the ones that matter. Expect a surprise or two: a publisher, a marketplace or a review site often shows up where you expected a direct rival. A review site that keeps appearing is not a competitor to beat; it is a page to get onto.
Add each rival’s website, not only its name. A website makes the comparison exact: it lets Cituna recognise the rival’s own pages among the sources, and crawl the site for the on-page comparison in Step 6.
How many you can track depends on the plan: 5 rivals per brand on Starter ($39 a month), 10 on Pro ($119) and 15 on Max ($399). The free trial tracks 3.
Step 3: Measure share of answers on the same answers
Share of answers is the percentage of your tracked category prompts whose answer names a given brand, you included. What makes it fair is that every brand is measured on the same answers. Counting your citations on one set of questions and a rival’s on another produces two numbers that cannot be compared.
Measure per engine as well as overall. A rival can lead on one engine while you lead on another, and the fixes differ. And measure more than once: answers change between runs, so a single check is a sample, not a result. Cituna shows each brand’s share, your rank among your rivals, the gap to #1 in points, and a chart of each brand’s share of answers over time, scan by scan. For the definition and its variants, see AI share of voice.
Step 4: Read the gap prompt by prompt
The headline share hides where the gap actually is. Open each rival and read three things:
- The prompts where they are cited instead of you. This is the list that carries the work. In Cituna it sits under “Where rivals beat you”, one row per prompt, with a Gap badge and the option to queue the fix in Articles.
- The prompts where you are both cited. Here the question is position. Cituna’s answer map plots each brand’s share of citations on category prompts against its average position when cited, so you can spot a rival that is named less often but placed first.
- The prompts you own. These are worth protecting. A drop here shows up in the trend before it shows up in your traffic.
Then open the answers themselves. For each prompt and engine, Cituna stores what the engine said, the brands it named in order, and the sources it read. An answer can also read your page and still recommend someone else; Cituna marks that as “Used as a source, not recommended”, which is a positioning problem rather than a visibility one. If ChatGPT keeps naming one rival, our answer on why ChatGPT recommends your competitor lists the usual reasons.
Step 5: Follow the citations to the pages doing the work
A rival is rarely cited because of its own site alone. Engines cite review articles, comparison lists, forum threads and documentation, and the rival is named on them. Find those pages and you have the gap in a form you can act on.
Cituna’s Sources tab lists the pages engines cited for your prompts, how often, which engines cited them, and which competitors they were cited for. It reads each page to check whether you are on it: linked, mentioned without a link, or absent. The To pitch list is the comparison you want here: sites citing rivals but not you. Open a row for the evidence, the play and a pitch you can copy, and set its status to Planned, Pitched, Placed or Declined. Mark a rival’s own site as Competitor and it is hidden from outreach. Our guide to earning citations from third-party websites covers the outreach itself.
Look at the format too. If the cited pages are comparison tables, a better essay will not displace them; a table will.
Step 6: Compare on-page readiness last
It is tempting to start by auditing a rival’s website. Do it last. Cituna’s head-to-head includes two on-page bars, On-page SEO and Generative readiness (GEO), which read the rival’s homepage against a full crawl of your site. They show whether a rival’s pages are easier for crawlers to read and quote. They do not show who the engines recommend, and a large rival can score low here and still be cited more often everywhere. The citation bar is the one that tells you who is winning; the on-page bars help explain why.
Step 7: Turn the gap into work and measure again
Each lost prompt needs one of two things: a page on your site that answers it, or a place on the pages the engine already reads. For the first, queue the prompt from “Where rivals beat you”: it lands in Articles, where Cituna writes the page that answers it once Articles is switched on for your domain. For the second, the To pitch list gives you the targets. Ship one, then watch the trend for that prompt over the following scans rather than the overall score, which moves slowly.
Connect Google Search Console as well, on any paid plan. A share gain that never shows up as clicks or impressions is worth knowing about, and Cituna joins your Search Console data to the same brand.
A benchmark worksheet you can copy
If you run the comparison by hand, keep one row per prompt per engine per check, and these columns:
| Column | What goes in it |
|---|---|
| Date and engine | When you asked, and which engine, with web search on or off |
| Prompt | The exact wording, unchanged between checks |
| Branded? | Yes if it names you; leave these out of share |
| Brands named, in order | Every brand in the answer, first to last |
| You named? Linked? | Whether the answer names you, and whether it links your site |
| Sources cited | The URLs the engine linked or read |
| Kind of competitor | Co-mentioned, cited instead, or compared |
Share for a brand is then the number of category rows naming it divided by the number of category rows. Use fresh chats each time, keep the wording fixed, and repeat at least weekly. By hand, seven engines and ten prompts is seventy answers per check, which is the reason tools exist.
What Cituna does not cover
Cituna measures seven engines: ChatGPT, Perplexity, Gemini, Claude, Grok, Google AI Overviews and Google AI Mode. It does not track Microsoft Copilot or Meta AI. If those matter in your market, compare them by hand with the worksheet above; for your own Copilot citations, see how to get cited by Microsoft Copilot, which covers the AI Performance report in Bing Webmaster Tools.
Plans start at $39 a month with one brand, 10 daily prompts and 5 tracked rivals, and go to $119 (Pro) and $399 (Max) for more prompts, brands and rivals. API keys come with every paid plan, and the MCP server lets you query your Cituna data from Claude: read-only on Starter, read and write from Pro. Our guide to tracking prompts and competitors shows the setup screens.
Drafted with AI assistance from our own research and Search Console data, and reviewed by Rahul A before publishing. Rules and prices change; check the linked official source before you act.