How to track competitors in AI answers

The useful question is rarely “how visible are we?”. It is “who is getting recommended instead of us, and what do they have that we do not?”. AI answers are unusually good at exposing this, because unlike a ranked list of ten results, an assistant names a short set of businesses and often explains why.

This guide covers how to track that properly with Visibility AI’s Monitor layer, and how to turn it into work rather than anxiety.

Why competitor tracking works differently here

In search, your competitor is whoever ranks above you for a keyword, and the gap is measured in positions. In AI answers there are no positions to speak of. There is a shortlist of two or three names and everyone else, which makes the competitive picture blunter and easier to act on: either you are in the answer or your rival is there alone.

It also surfaces a different competitor set. The businesses that appear in AI answers are the ones well represented in the sources engines read, which is not the same as the ones that outrank you or outspend you. It is common to find a competitor you have never tracked appearing in half your answers because they are all over a review platform and two roundups.

The other shift is that you can see the reasoning. An assistant asked for the best option in a category will often say why it picked one, and cite where it learned it. That is competitive intelligence you could not get from a rankings report.

Step 1: Fix a competitor set, and base it on the answers

Start from data, not from memory. Run your tracked questions, collect every business that gets named alongside or instead of you, and count how often each appears. That frequency list is your real competitive set.

Then fix it. A competitor list that changes every month makes trend comparison impossible, which is the same discipline that applies to your prompt set. Add deliberately when a newcomer keeps appearing, and note when you do, so a step change in the chart has an explanation attached.

Visibility AI tracks a competitor set alongside your brand across the same questions and engines, so every comparison is like for like rather than a screenshot of one assistant on one day.

Step 2: Measure share, not just presence

Presence tells you whether you were named. Share tells you how much of the answer space you own relative to the set, which is the number that moves when a competitor invests.

Watch it per engine as well as overall. Sharp differences between engines are normal and diagnostic: a rival who dominates Perplexity but not ChatGPT is usually strong in a source one engine leans on. Measuring AI share of voice covers the mechanics and the traps, including why a rising share can hide a shrinking answer set.

Step 3: Read where they win, not just that they win

An aggregate share number tells you the score. The useful detail is in the breakdown:

• Which questions do they win and you lose? Losing every comparison question but holding the category ones is a completely different problem from the reverse.
• Which engines? A gap on one engine points at a source, not a strategy.
• Are they recommended or just mentioned? Being listed among options is weaker than being the one the assistant picks, and the difference tells you how much ground you actually have to make up.

Grouping your prompts into topics makes this legible quickly, because the pattern is usually concentrated in one part of your question set rather than spread evenly.

Step 4: Work backwards to the sources behind their answers

This is the step that turns tracking into a plan. For the questions where a competitor wins, look at what the engines cited or drew on. Across enough answers a pattern appears: the same review platform, the same “best X in Y” roundup, the same community thread.

That list is more valuable than the competitor’s own website, because those sources are reachable for you. A roundup that names three of your rivals and not you can be pitched. A review platform where they have forty reviews and you have four is a straightforward, if unglamorous, project.

Tracking citations and sources covers how to collect this, and earning the citations AI trusts covers how to work the shortlist once you have it.

Step 5: Set a cadence and react to real moves

Check on a schedule, not on impulse. Weekly is enough for most categories, and it separates a genuine competitive move from the normal wobble that comes with model updates and rephrasing.

React to changes that persist across two readings and appear on more than one engine. A single-engine, single-week jump is usually noise. A competitor climbing steadily on three engines has done something specific, and the sources behind their new answers will usually tell you what.

See who is being recommended instead of you

The quickest version of this takes a minute: run the free Visibility Check and read the list of businesses the assistants name for your category. Some of them will be expected. The ones that are not are the reason this is worth tracking properly. From there, Monitor turns it into a weekly line rather than a one-off, and finding your optimization opportunities turns the gaps into a ranked list of work.

Frequently asked questions

How many competitors should I track?

Five to eight is the practical range. Fewer and you miss the challenger that keeps appearing; more and the comparison gets noisy and every report needs a paragraph of explanation. Pick the ones that actually show up in your answers, which is often not the same list as the one in your board deck.

What if a company I have never heard of keeps appearing?

Take it seriously and add it. AI answers regularly surface businesses that are invisible in your usual competitive research, because they are strong in exactly the sources engines read - a review platform, a category roundup, an active community thread. Those are the most instructive competitors you have.

Can I see why a competitor is being recommended?

Usually, yes. Answers cite or paraphrase sources, and the pattern across many answers is readable: the same review site, the same roundup, the same forum thread keeps appearing behind their mentions. That pattern is a target list, because those sources are reachable for you too.