AI Buyer-Query Library Template

AI Buyer-Query Library (fill-in)

When to use: One row per buyer question, with the role, purchase task, prompt, geography, language and rationale, plus an owner and a completion check.

AI buyer-query library

One row per question you want an AI answer to show you for. Keep prompts as real buyer questions, not keywords.

Library owner: [name] - Engines to run: [ChatGPT / Gemini / Claude / Perplexity / Google AI Overviews] - Review cadence: [weekly / monthly]

Per-row detail (fill for every question)

• Query ID: [short code, e.g. Q-012]
• Branded or non-branded: [names your brand / names a category only]
• Priority: [high / medium / low, and why]
• Target page: [the URL an engine should cite for this answer]
• Source of the question: [sales call / support ticket / Search Console / site search / review]
• Status: [queued / running / cited / not cited / page missing]

Next action

• Library owner: [name] - task: keep rows current and set each priority by [date]
• Prompt runner: [name] - task: run the prompts and log where you appear by [date]
• Content owner: [name] - task: brief or fix the target page for each gap by [date]

Completion check

• Every row has all six fields: role, task, prompt, geography, language and rationale.
• Each prompt reads as a real buyer question, not a keyword string.
• Each row names one engine set, one target page and one owner.
• Geography and language are set, because answers change by both.
• No row is invented in a meeting. Each one traces to a real source.

Where the Questions Come From

When to use: Source real buyer questions from your own data instead of guessing, so the library reflects demand you can act on.

Where the questions come from

A good library is not a list the SEO team made up in a meeting. It is the questions your buyers already ask, in their words.

Rules for sourcing

• Write each question the way a buyer would ask it, not as a keyword.
• Tag the source on the row, so you can defend the priority later.
• Keep a representative sample. You do not need every wording of the same question.
• Pull pricing, comparison and switching questions first. Those sit closest to a sale.

Pull question ideas for a seed from the Questions tab in Keyword Research, then keep only the ones a buyer near a decision would ask. The Buyer Question FAQ Brief helps turn those into answers.

Buyer-Task and Intent Picker (BOFU)

When to use: Choose the purchase task behind each question, so the library covers the moments closest to a decision.

Buyer-task and intent picker

Label every row with one purchase task. These are the moments where an AI answer can win or lose you the shortlist.

How to use it

• One task per row. If a question covers two, split it into two rows.
• Keep more rows on the tasks nearest a decision: compare, cost and switch.
• Pair each task with the page an engine should cite. No page means a gap to brief.

Group the keywords behind these questions into clusters with the Keyword Cluster Tool, then map each cluster to one page using the BOFU Keyword to Page Map.

Prompt-Writing Rules

When to use: Write prompts that match how buyers ask, with role, geography and language built in, so your runs reflect real answers.

Prompt-writing rules

The prompt is the test. Write it the way a real buyer would, then keep it fixed so you can compare runs over time.

Build the buyer into the prompt

• Name the role or context when it changes the answer: team size, industry, budget or the system they already use.
• Keep geography and language consistent for a direct comparison. Answers differ by both.
• One task per prompt. Do not ask for a comparison and a price in the same line.
• Keep a fixed benchmark set you repeat, plus a discovery set for new questions.

What makes a page eligible to be shown

Google says a page must be indexed and eligible to show in Search with a snippet to appear as a supporting link in AI Overviews and AI Mode. There are no extra requirements and no special markup to add.

Meeting this does not guarantee a page is shown.

Source: Google Search Central, AI features and your website, checked 13 November 2026.

Common mistakes

• Pasting a keyword instead of a question. Engines answer questions.
• Changing the wording between runs, so you cannot tell what moved.
• Only tracking branded prompts, which hides whether new buyers find you.

Run the finished prompts yourself and keep the full wording on the row, so every run uses the same prompt. To see which recorded AI answers already name or link your domain, use the AI Visibility Checker.

Owner, Cadence and Run Log

When to use: Record each run the same way, because AI answers change often and a single check is not enough.

Owner, cadence and run log

One check is a snapshot. Answers move, so log each run with the conditions, and repeat on a set cadence.

Cadence rules

• Pick one cadence per set: weekly for high-priority prompts, monthly for the rest.
• Keep the location and language the same each run, or note why they changed.
• Save the answer text or a screenshot. Answers are hard to reproduce later.
• One owner runs the set and logs it. One owner acts on the gaps.

Keep each run as its own line so you can see trends. The AI Search Visibility Template tracks where you appear across engines over time.

Spreadsheet Header (CSV)

When to use: A paste-ready header row and allowed values to run the library in Google Sheets or Excel.

Paste-ready spreadsheet header

Paste this row into cell A1 of Google Sheets or Excel, then split text to columns by comma. One row per question.

query_id,buyer_role,purchase_task,prompt,geography,language,rationale,branded,priority,engines,target_page,source,status,owner,last_run_date,brand_shown,notes

Allowed values

• purchase_task: compare, alternatives, cost, integrations, trust, switch
• branded: branded, non_branded
• priority: high, medium, low
• status: queued, running, cited, not_cited, page_missing
• brand_shown: yes, no

Sheet tips

• Highlight rows where status is page_missing. Those need a brief before you run again.
• Highlight rows where brand_shown is no after two runs. Those are your real gaps.
• Keep one tab per engine if answers differ a lot by engine.
• Never delete a row. Mark it retired, so you keep the history of what you tracked.

When a row has no page an engine can cite, plan one with the Content Brief generator, then set the status back to queued.

Worked Example (fictional Crateloom)

When to use: Four filled rows for an invented warehouse inventory tool: one cited, one not cited, one page missing and one branded.

Worked example: Crateloom inventory software

Example only. Crateloom and Stowvern are invented warehouse inventory tools. The roles, prompts, locations and notes are made up to show a filled-in library.

Library owner: growth lead - Engines: ChatGPT, Perplexity, Google AI Overviews - Cadence: weekly for high priority

How the rows were handled

• Row 1: Perplexity cited the comparison page. Status cited. Keep running weekly.
• Row 2: ChatGPT named three others, not Crateloom. Status not cited. Alternatives page briefed.
• Row 3: No pricing page an engine could quote. Status page missing. Content owner assigned.
• Row 4: AI Overviews showed the integration, but with an old screenshot. Note added to update the page.

Why this works

• Each row traces to a real sales or support moment, so the priority is easy to defend.
• Geography and language are set, so the growth lead compares like with like.
• Gaps turn into briefs, not opinions, because every row names a target page.

Want help turning this library into pages AI engines cite? See SEO and GEO services, or book a free consultation with Anatolii Ulitovskyi.