Use when assistants answer "best tool for X" with competitors and never name you.
shortlist-inclusion-criteria.md
You are reverse engineering the criteria behind shortlist answers. You are not writing a sales argument.
Assistant answers naming a shortlist, with the reasons each gave: {{SHORTLIST_ANSWERS}}
The qualifier buyers use, for example "for small teams" or "for regulated industries": {{SEGMENT}}
Vendors named in those answers: {{NAMED_COMPETITORS}}
What is true about us and where it is published: {{OUR_FACTS}}
Produce:
1. A table: | Vendor from {{NAMED_COMPETITORS}} | Reason the answer gave | Criterion that implies | Evidence type behind it (published price, review volume, documented integration, named customer, certification, third-party test) |
2. A ranked list of criteria by how often they recur across {{SHORTLIST_ANSWERS}}.
3. A pass table against {{OUR_FACTS}}: | Criterion | Do we meet it | Is it published anywhere | Gap type (not met / met but unpublished / met and published but hard to find) |
4. Ranked actions, taking "met but unpublished" first, each naming the page and the sentence it needs.
Constraints:
- Judge every criterion against {{SEGMENT}}, not the category in general.
- Separate not qualifying from not being visible. Never recommend claiming a criterion we do not meet.
- Where an answer names a vendor with no reason attached, write "no reason given" rather than inferring one.
Replace each placeholder with your own detail. The more specific you are, the less the model invents.
When do I use this rather than the competitor citation gap?
When you are absent from best tool for X answers entirely, rather than losing one page comparison. The citation gap prompt compares two pages for one question. This works backwards from several shortlist answers to the rules those answers imply, which is the right level when the problem is exclusion.
What do I need in front of me?
Several shortlist answers captured verbatim, with the reason given for each vendor, plus the qualifier buyers use. The reason clause is the whole input, so answers that merely list names produce no reason given rows. Pick one segment, since criteria for enterprise buyers and solo users rarely overlap.
What comes back, and which column is the payoff?
Criteria reverse engineered from each vendor mention with the evidence type behind them, a frequency ranking, and a table scoring you against each. The gap type column is the payoff. It separates criteria you do not meet from criteria you meet but never published, and the second kind is a page edit.
What is the mistake that costs me?
Closing a gap by claiming a criterion you do not meet. The prompt refuses to recommend it, and it fails twice over: the claim gets checked, and it puts you on shortlists you then lose at the demo. Work the met but unpublished rows first, since each names a page and a sentence.