Competitor analysis in AI answers means recording, every month, which companies ChatGPT, Perplexity, Gemini and Google’s AI Overviews name for your buyers’ questions, which sources they cite for each competitor, and what those sources say. The pattern tells you why a competitor is recommended: a review profile you lack, a comparison page they published, coverage in a publication you have not reached, or simply clearer answers on their own site. Each reason points to a specific action.
What you need
- The question set and monthly test runs described in how to track brand mentions for free
- A list of three to six competitors buyers actually compare you with, from sales, not from guesswork
- An hour or two a month
Step 1: count who gets named
From your runs, count how often each competitor appears and calculate their share of voice the same way as yours: runs where they are named divided by total runs. Split the result by question group (category, problem, comparison), because competitors often dominate one group and are absent from another.
Example: share of voice by question group
Illustrative numbers for three companies across 60 runs per group
- Competitor A, category questions55%
- Competitor A, problem questions12%
- You, category questions15%
- You, problem questions38%
Illustrative example only, to show the kind of pattern the analysis reveals.
In this example, Competitor A wins “who offers” questions, while you win “how do I” questions. The action is different for each.
Step 2: record the sources behind each mention
For every run where a competitor is named, note the cited sources next to their name. After a month you will see where their visibility comes from:
| Source type | What it tells you | Action |
|---|---|---|
| Their own service or product page | Their page answers the question clearly | Compare with your page and improve the answer |
| Review platform or directory | They have a profile or reviews you lack | Build or improve your profile (review sites guide) |
| “Best of” list or industry article | They were included by a publication | Contact the author, or earn coverage elsewhere (digital PR) |
| Comparison page | They wrote the comparison buyers are reading | Publish a fair comparison of your own (template) |
| No source shown | The answer comes from training data | Long-term: consistent presence across the web |
Step 3: read how each company is described
Copy the sentence each assistant uses to describe every competitor and yourself. Compare them. Assistants often repeat a positioning line found on the company’s site or in reviews, such as “best for small teams” or “strong in manufacturing”. If a competitor owns a description you want, look for where it comes from. If your own description is vague or wrong, fix it at the source: your About page, profiles and structured data.
Step 4: turn the findings into a monthly action list
Month | Question group | Competitor winning | Main source | Our gap | Action | Owner | Due
Limit it to three to five actions a month. The aim is steady improvement you can measure next month, not a long wish list.
Step 5: report the trend
Share of voice for you and each competitor, by question group, month by month, is the clearest summary for leadership. Add the actions taken, so changes can be linked to work. A dashboard for this is described in building an AI visibility dashboard in Looker Studio.
Frequently asked questions
What if competitors are named that we did not expect?
Add them to the list. Assistants sometimes reveal competitors your sales team does not see, such as larger international firms or new entrants.
Should we copy what competitors do?
Match what gives them visibility, such as a review profile or a clear comparison, but add something they lack. Information that already exists elsewhere is easy for assistants to credit to someone else.
How many runs make the comparison reliable?
At least three per question per assistant each month. With fewer, one unusual answer changes the result too much.
Sources
- Seer Interactive, “What drives brand mentions in AI answers?”
- Ahrefs, study of brand visibility factors in AI Overviews across 75,000 brands