Tracking a Competitor's URL: How to Find Out Why Their Page Wins the Citation
Use page-level tracking on a rival's URL to see which prompts it wins, how its answer is built, and what your own page needs to change to take the slot.
When an AI answer cites your competitor instead of you, the useful question is not "how do we beat them" in the abstract. It is "what does that specific page do that ours does not, for that specific question". You can answer that, but only if you track the rival's page the way you track your own.
Page-level tracking turns a vague competitive worry into a short comparison you can finish in an afternoon.
Start with the losses, not the rivals
Do not begin with a list of competitors. Begin with a list of prompts where you wanted to appear and did not. Pull ten from your monitoring, ideally ones with real commercial value and a repeatable pattern: the same competitor page cited again and again.
Repetition is the filter. A source that shows up once may be an accident. One that appears across several runs and several related prompts is the page engines have chosen as the answer for that topic, and that is the page to study.
Add the URL to a page tracker
Promptwatch's Page Tracker accepts specific URLs, yours or a competitor's. Once a page is tracked, you can see the responses and the prompts where it appears. That reverses the usual view. Instead of asking "which sources did this prompt cite", you ask "which prompts cite this page".
The reversed view is better for comparison. You see the full set of questions the rival page serves. Often it is broader than you assumed. A pricing explainer may be winning for comparison prompts, how-to prompts and definition prompts alike, which tells you it is working as a hub, not as a single answer.
Read the winning page like an editor
With the prompt set in hand, open the page and look at it next to the answer. Check five things.
The opening. Does the first screen answer the question directly, in plain words? Pages that win citations frequently state the answer early and elaborate later.
Structure. Are there headings that match the questions people ask? A heading phrased like the prompt makes the page easy to match.
Specifics. Does it have concrete detail, such as steps, ranges, named options or dates, where yours has general claims?
Freshness. Is there a visible update date, and does the content reflect recent changes in the topic?
Format. Is it a list, a table, a guide, a comparison? Compare the format with the page types being cited for the other prompts in the group.
Write down differences as a plain list. Resist the urge to copy anything. You are looking for gaps in your coverage, not text to imitate.
Check the source type
Sometimes the winning page is not a competitor's blog. It might be a forum thread, a video, a review site or a third-party article that mentions the competitor. In that case, no rewrite of your page fixes the problem, because the engine is choosing a different kind of source.
Promptwatch's citation analytics break this down with Reddit, YouTube and LinkedIn citation reports, and offsite mentions that show where a brand is named in third-party pages without a link. Citation Trends classifies each citation by content type and source type. If the pattern says "third-party review", your answer is outreach and presence, not on-site edits.
Check whether your page had a fair chance
Before blaming the content, verify access. Look at crawler activity for your own page: did GPTBot, OAI-SearchBot or ChatGPT-User fetch it, and with what status code? Agent Analytics reports fetches and errors with failure explanations, and a citation rate per page. A fair comparison needs both pages to have been readable.
This step has saved many teams from rewriting a perfectly good page. The loser was never fetched properly. Crawler logs are included from the Professional plan upward, and on all agency plans. The Essential plan has none.
Build the fix list
Combine the evidence into a short brief per prompt group:
- Which page wins and for which prompts.
- What source type it is.
- Whether your page was fetched without errors.
- The three biggest content gaps, from your editor's read.
- The action: refresh, new page, outreach, or access fix.
For refresh actions, Content Refresh gives diff-style suggestions for existing pages. For new pages, content gap analysis with coverage scores shows which sub-questions you do not yet answer. The Action Board collects these as prioritized, evidence-backed tasks, and its Discovery agent can suggest up to five per weekly run, which is a decent way to keep a competitive backlog moving without a meeting.
Do not over-read one rival
A word of caution. A competitor page that wins today may not win in a month. Engines rotate sources, and your own changes feed back into the answer. Track the page for several weeks before you treat its win as structural, and record the date of any change you make so you can tell whether it worked.
It is also worth tracking more than one rival page per topic. If three competitors alternate, your problem is the topic, not one company.
What a good outcome looks like
Success is modest and measurable. After a refresh, your page begins to appear for some of the prompts that previously cited the rival, shown in the same tracker. You will not take every slot. Taking two of five is a result worth reporting.
The recommendation
Start with one prompt group you keep losing. Add the winning URL and your own to the Page Tracker in Promptwatch, read both next to the answer, check your crawl access, and write the brief. That single exercise tends to teach more than a month of share-of-voice charts. You can try the tracker through promptwatch.com, and the free Explore plan is enough to run the first comparison on ChatGPT.