AI citation gap analysis prompt for finding the evidence competitors own
Compare cited answers across AI assistants, find which source types competitors own and turn the evidence gap into a practical publishing plan.
- Works in
- ChatGPT, Claude, Gemini
- You need
- A fixed set of buyer questions · Answers and citations from at least three assistants · Your existing relevant URLs · Competitor names
- Written for
- ai citation gap analysis prompt
Traditional content gap reports compare keyword lists. An AI citation gap report compares the evidence used to construct answers. The difference matters because the missing asset may be documentation, a dataset, a review or an independent reference rather than another blog post.
Run the AI crawler checker before commissioning content when a relevant page already exists but never appears in a source list.
Analyze citation gaps from the answer records I provide. Do not ask the
assistants new questions and do not infer citations that are not in the
records.
Normalize each cited URL to its domain and classify the source as official
documentation, editorial article, comparison, review, forum, video, dataset,
academic source, public body or other.
For every buyer question report:
Question | Brands named | Cited domains | Source types | Our domain present |
Competitor domain present | Evidence the answer appears to reward | Gap
Separate three gaps:
Reachability gap, where our relevant page exists but is not retrievable.
Answer gap, where no page answers the buyer question directly.
Corroboration gap, where our site makes the claim but no independent source
confirms it.
Never recommend creating a page merely because a competitor has one. Propose
an asset only when the records show a distinct unanswered question or missing
evidence type. Merge recommendations that would target the same intent.
End with a plan ordered by dependency. Technical access comes before content.
First party factual pages come before outreach for independent corroboration.
For each action include the observed evidence, proposed asset, owner, effort
and the question it should improve.
Our brand and domain: [BRAND AND DOMAIN]
Competitors: [COMPETITORS]
Existing relevant URLs: [OUR URLS]
Answer records with citations: [PASTE RECORDS]What to change
Everything in square brackets is yours to replace. Nothing else needs editing.
[BRAND AND DOMAIN]- Needed to separate citations you own from independent corroboration.
[COMPETITORS]- Use known commercial alternatives rather than letting the model invent the comparison set.
[OUR URLS]- Existing pages prevent the output from recommending content you already have.
[PASTE RECORDS]- Use answers captured from fresh conversations with visible citations and dates.
How to run it
- 01Run a fixed question set
Ask each question in a fresh conversation across the same assistants and save every visible citation.
- 02List your relevant pages
Include the pages that should answer each question, even if assistants did not cite them.
- 03Classify the gap
A blocked crawler, missing answer and missing independent reference require different work.
- 04Repeat after publishing
Use the same questions and assistants so the second run can be compared with the first.
Questions people ask
What is an AI citation gap?
It is evidence or a source type that assistants repeatedly use to answer a buyer question but do not retrieve from or about your brand.
Does copying a competitor page close the gap?
No. The useful finding is the question or evidence type the answer rewards. Copying the page creates a weaker duplicate and may target the same intent as a page you already own.
Unsubscribe in one click. We never pass your address on.