The AI recommendation gap template is a spreadsheet for logging where AI assistants name competitors instead of you, and what you'll publish to fix it.
Short answer: one row per question per assistant, with columns for the question, intent, assistant, date, what the answer says, sources cited, gap severity, the fix, an owner and a status. Start with 10–15 buying questions, fill it in honestly, sort by severity and work from the top. Below are the columns, a filled-in example and how to keep it current.
What columns does the template need?
Column | What to put in it |
|---|---|
Question | The exact wording a buyer would use: problems, comparisons, "best X for Y", "[competitor] alternatives". |
Intent | Informational, comparison or transactional. Work on comparison and transactional first. |
Assistant | ChatGPT, Gemini, Perplexity, Claude, Google AI Overviews or AI Mode – one row each, because results differ. |
Date checked | Answers change, so every result needs a date. |
Current answer | Who's named, in what order, and whether you appear ("Competitor A, B; us not mentioned"). |
Sources cited | The URLs the answer links to, if any. This tells you what to match or where to get listed. |
Gap severity | High: a buying question where a direct competitor is named and you aren't. Medium: you're named but low down or described wrongly. Low: informational, or you already appear. |
Fix | The specific page or change, e.g. "comparison page: us vs Competitor A for agencies". |
Owner | One named person. |
Status | Not started, in progress, published, re-checked. |
How do you fill it in?
Pick 10–15 buying questions. "Best software for…", "[category] pricing", "[competitor] alternatives". If you haven't done a first pass yet, the one-afternoon AI recommendation audit walks through it.
Ask each assistant in a fresh chat. Turn off memory or use a clean account so past chats don't steer the answer. Save the full response.
Log what you see, including bad news. If the answer praises a competitor or describes you wrongly, write that down.
Sort by severity. High-severity gaps first. If you have many, score them with the gap scorecard to break ties.
Assign a fix, an owner and a date. A gap with no owner stays open.
Re-check after publishing. Re-run the question two to four weeks later and add a new dated row rather than overwriting the old one.
What does a filled-in template look like?
A fictional example for a project management tool called FlowState:
Question | Assistant | Current answer | Sources cited | Severity | Fix |
|---|---|---|---|---|---|
best project management tool for remote-first startups | Gemini | CompetitorTask; FlowState not mentioned | 2 review sites, CompetitorTask blog | High | Guide to project management for remote-first startups, with async features and pricing |
best project tool for agencies | ChatGPT | CompetitorTask, Competitor B; FlowState not mentioned | None shown | High | Fair comparison: FlowState vs CompetitorTask for agencies |
how to manage creative workflows | Perplexity | FlowState named first | FlowState guide | Low | Keep the guide up to date |
FlowState alternatives | Gemini | FlowState plus Competitors B and C; FlowState pricing out of date | Old review | Medium | Update pricing page; ask the reviewer to refresh |
Two things stand out. FlowState is missing from both high-intent questions, and one answer repeats an old price from a review. Neither is fixed by more general blog posts. The first needs two specific pages; the second needs a pricing update and an email.
A note on the fixes: titles like "The Ultimate Guide to… in 2026" age quickly and read as filler. A plain, specific title and a first paragraph that answers the question work better for readers and for being quoted.
Should you keep it manual or use a tool?
The spreadsheet is a good start: it costs nothing and forces you to read the answers. The work grows quickly, though – 15 questions across four assistants is 60 checks per round, every few weeks. Trackers take over the checking: Otterly.AI and Wellows start at $29 a month, Semrush's AI Visibility Toolkit is $99 a month, and blogawesome ($49 a month) adds drafting and publishing for the gaps it finds (prices checked 5 Oct 2026). Keep the "Fix", "Owner" and "Status" columns either way. Those are the parts no tool does for you.
If you're not sure why this matters yet, start with what AI brand visibility is.
To fill in your first rows faster, run the free AI visibility audit. It takes no signup and about 30 seconds, and shows whether AI models recommend you, who they name instead and which gaps to close first.