How do I know if my competitors are getting more visibility than me in AI-generated answers? Our client method

When a client suspects a competitor is winning in AI answers, they are usually right about something and wrong about what. Our job in the first few weeks is to replace the hunch with a comparison that holds up. Here is the process.

Pick the right comparison set

We ask the client who they think the competitors are, then we add the brands that models actually name for the category. The two lists rarely match. A regional player, a marketplace or a software suite that does one feature well can show up in AI answers more than the rival the sales team worries about. In Promptwatch, auto-discovered competitors help us catch those, and brand merging lets us combine name variants so a brand is not counted twice.

Use one panel for everyone

The comparison only works if every brand is measured against the same prompts. We build a panel from buyer questions: category, comparison, alternatives and brand prompts. We tag by topic, persona and country. Then every brand, the client and the competitors, is read from the same responses. We keep the panel fixed so trends are meaningful.

The four numbers we compare

1. Share of Voice. In what share of tracked responses is each brand named?
2. Position in Answer. When named, where does the brand appear?
3. Sentiment. How does the answer talk about each brand?
4. Citations. Which pages and domains does the engine cite for each?

We read them per engine. A competitor that leads in ChatGPT but trails in Google AI Overviews is a different problem from one that leads everywhere.

Where we look at the gap

We run client programs on Promptwatch. A few features do most of the work for this question:

• Competitor dashboards and the heatmap. The heatmap compares any two dimensions, for example brands against models or topics against countries. It shows in one picture where a competitor leads and where the client does.
• A/B period comparison. Useful for checking whether a gap is widening after a competitor launches something.
• Citation analytics. We list the pages and domains behind each competitor's mentions, including Reddit, YouTube and LinkedIn reports and offsite mentions.
• Page Tracker. We can track specific URLs, including a competitor's, to see how they get cited over time.
• Agent Analytics. IP-verified crawler logs for the client's site. Sometimes the reason a competitor looks better is that the client's pages are failing for a bot.

What the diff usually shows

We hear three patterns again and again.

The competitor is cited from third-party pages the client is absent from. A comparison article, a forum discussion or a review listing carries the competitor and not the client. The fix is earning presence there, not publishing more on the client's own site.

The competitor has a page for a sub-question the client does not answer. Content gap analysis with coverage scores helps us find these, and the Action Board puts the fix on the to-do list.

The client has a crawler problem. It is the cheapest to fix and often the most overlooked.

What we avoid

We do not report a single week as a verdict. We do not compare on branded prompts only. We do not assume that the biggest brand in the category will always win, because we have seen smaller brands named often when they have clear, well-cited content. And we do not promise to close the gap in a set time. Models change, and competitors respond.

What we give the client

A short summary: where each brand stands per engine and topic, which three or four gaps matter commercially, which sources drive the competitor's lead and which fixes we recommend first. Then we track it weekly and report the trend with scheduled PDF reports.

If you want to know whether competitors are ahead of you, the quickest way to find out is to put them on the same prompt panel as you and read the share of voice side by side. Everything useful follows from that.