Short version: Tracking competitor visibility in ChatGPT means running a consistent set of buyer-intent prompts repeatedly, logging which brands get named and in what position, and turning that into a share-of-voice number you can watch over time — since there's no public rank tracker for ChatGPT the way there is for Google. I built a rough version of this by hand before moving most of it over to Obsurfable, mainly because "by hand" and "repeatedly, over time" turned out to pull against each other pretty quickly.
This isn't the same shape of problem as SEO rank tracking
The instinct is to look for a ChatGPT equivalent of a rank tracker — type in a query, get back a clean position 1 through 10. That tool doesn't really exist, for a reasonable reason: ChatGPT isn't returning a ranked list, it's generating a single answer that may or may not name any given brand, and the same exact prompt can produce a different answer on a second attempt. Competitor visibility here is closer to "share of answers" than "share of rank" — how often a brand gets named across a representative set of questions, not where it lands on a fixed list.
It's also worth knowing upfront that ChatGPT doesn't always search the web before answering. Plenty of responses come from what the model already learned during training, with no live retrieval involved at all — which means a brand can be genuinely absent from an answer not because it lost a competitive comparison, but because the question never triggered a search in the first place. That distinction matters when you're trying to figure out whether a gap is a content problem or just a query that stayed in the model's static knowledge.
Step 1: Build a prompt set worth tracking
Buyer-intent questions work better than broad category questions: "best CRM for a five-person sales team" tells you more than "what is a CRM." A useful mix usually includes direct comparisons ("[your brand] vs [competitor]"), alternative-seeking questions ("alternatives to [competitor]"), and intent-modified questions using words like "cheapest," "for startups," or "enterprise," since those tend to be the ones closest to an actual decision. I'd originally built mine around generic category terms and gotten thin, unhelpful results — the buyer-intent versions surfaced far more useful signal. A prompt explorer is what eventually replaced my guesswork here with the actual range of questions buyers ask.
Step 2: Run each prompt more than once
This is the part that surprised me most. The same prompt, asked twice in separate sessions, can come back with a different set of brands or a different order entirely. Treating a single run as the answer is misleading — you're better off running each prompt a handful of times and looking at how often a brand shows up across those runs, not whether it happened to show up once.
Step 3: Turn raw results into something comparable
A simple log is enough to start. Here's roughly what a handful of rows might look like:
| Prompt | Your Brand Mentioned? | Competitor Mentioned? | Position | Notes |
|---|---|---|---|---|
| Best CRM for small sales teams | No | Yes (Competitor A) | 1st of 3 named | No citation, just named |
| Alternatives to Competitor A | Yes | Yes (Competitor A) | 2nd of 3 named | Cited with link |
| CRM comparison for startups | No | Yes (Competitor A, B) | — | Neither position clear |
| Cheapest CRM with automation | Yes | No | 1st of 1 named | Only brand mentioned |
Once you've got enough rows, rolling it up into a scorecard makes the trend visible instead of buried in individual entries:
| Metric | You | Competitor A | Competitor B |
|---|---|---|---|
| Mention rate (of 40 prompts) | 45% | 78% | 30% |
| Appears in top position | 20% | 55% | 12% |
| Cited with a link | 15% | 40% | 8% |
| Mentioned alongside you | — | 25% | 10% |
Both of these are illustrative, not real numbers from anywhere specific — the point is the shape: a raw log for detail, a rolled-up scorecard for the trend you'd actually act on.
Step 4: Figure out why they're showing up
Once you know who's winning, it's worth understanding why. Competitor visibility in ChatGPT tends to trace back to things you can actually go inspect: clear category positioning, a strong volume of third-party mentions (review sites, comparison blogs, industry directories), and detailed public content answering the exact questions you're testing. Retrieval readiness analysis covers whether a competitor's — or your own — content is even structurally set up to be pulled from in the first place, which is often part of the explanation when one brand consistently outperforms another on nearly identical content quality.
It's worth resisting the urge to copy a competitor's exact content once you spot a gap. What usually matters more is the underlying reason they're winning a given prompt — heavier third-party coverage, a clearer comparison page, more consistent entity information — since matching their specific wording rarely closes a gap that's actually rooted in one of those.
Step 5: Decide how much of this you'll actually keep doing by hand
Here's where the math stops working in favor of manual tracking. Forty prompts, run three times each, checked monthly, across two or three competitors, is a genuinely large number of individual checks to do by hand on a recurring basis — and that's before accounting for the fact that a one-off skip during a busy month quietly turns into a six-week gap with no data. I did this manually for a while and the actual failure wasn't the method, it was consistency — the checks that didn't happen because nobody had time that week.
Prompt monitoring is what replaced the manual version for me — running the same prompt set against a live model on an actual schedule, rather than whenever I remembered to. If you want a quick read on where you currently stand against a competitor before committing to a full tracking setup, Obsurfable's free AI visibility checker gives a fast first look, no account required.
What ongoing tracking actually needs to show you
A single scorecard tells you where things stand today. The more useful version shows the trend, and flags it specifically when something changes — a competitor who used to lose a prompt to you and suddenly doesn't, for instance. A rolled-up AI Brand Health score gives me that trend line instead of a pile of disconnected monthly snapshots, and incident alerts are what caught the first time a competitor overtook me on a prompt I'd been reliably winning, within days rather than at the next scheduled check.
For tracking this across a defined competitor set on an ongoing basis rather than a one-time comparison, Obsurfable's plans cover what that setup actually looks like.
FAQ
How can I track competitor visibility in ChatGPT without a big tooling investment?
Start manually: 20 to 30 buyer-intent prompts, run three times each, logged in a spreadsheet with columns for brand mentioned, position, and whether a citation was included. It's a real first pass and costs nothing but time.
Why does the same prompt give different answers each time I ask it?
ChatGPT's responses can vary between runs of an identical prompt, which is exactly why a single check is unreliable — running each prompt multiple times and looking at the pattern across runs gives a more honest picture than any one answer does.
Is being mentioned the same as being recommended?
No. Being named alongside several other brands is different from being the one specifically recommended for the stated need — worth tracking as separate columns rather than collapsing them into one mention count.
How often should I re-run this tracking?
Monthly is a reasonable default for competitive categories. The bigger risk isn't the exact cadence, it's inconsistency — checks that quietly stop happening during busy stretches are more damaging than a slightly-too-infrequent schedule.
Does ChatGPT visibility correlate with classic SEO rankings?
Often, but not perfectly. Strong SEO authority, review site presence, and public web mentions tend to overlap with ChatGPT visibility, since a lot of what ChatGPT retrieves traces back to the same web ecosystem — but the overlap isn't complete, and a brand can rank well in Google while barely showing up in ChatGPT's answers, or the reverse.
Should I track every competitor, or just the main ones?
Just the ones that actually matter to a buying decision — usually two to four. Tracking a longer list dilutes attention without adding much useful signal, and it's easier to expand later than to sustain a bloated list from the start.
The honest version of this: the tracking method above isn't complicated, it's just repetitive in a way that's easy to underestimate until you're three weeks into "I'll get to it" and realize you have no idea whether you gained or lost ground. The method works. What actually determines whether you keep doing it is whether the repetition survives contact with a busy month.
Top comments (0)