# Buyer question research worksheet

Turn a documented customer question into a reviewed prompt and a useful page requirement. Keep the original wording and source beside the adapted question so the next content decision stays tied to a real buyer task.

Worksheet version 1.0 · 21 September 2026. Complete and repeat records as needed. Blank fields request your evidence; this manual file does not collect answers, import records or establish search demand.

Complete guide: https://rankecho.io/resources/find/buyer-intent-prompt-research

## 1 Capture the buyer question

Complete one record per question. Use sources your team may access and retain a retrievable reference without copying unnecessary personal details.

**Research scope and owner** — Record the product, audience, research objective, locale, time window, researcher and review date.

[Enter your response]

**Original question and source** — Retain exact wording, source type, source pointer and capture date. Distinguish interviews, sales and support records from Google Search queries and generated suggestions.

[Enter your response]

**Buyer task and stage** — Describe the decision the person is trying to make and classify discovery, comparison or purchase. Explain the classification; do not infer buying intent from a keyword alone.

[Enter your response]

## 2 Write and review the prompt

Write a natural, complete question. Add only context that materially changes the answer, and keep the original wording for comparison.

**Natural prompt and necessary context** — Record the adapted prompt and relevant use case, product category, compatibility need, location or budget. Explain every context detail added to the original question.

[Enter your response]

**Evidence and unknowns** — Record why the source supports this buyer task, any conflicting questions, product facts needing verification and information still missing. Search Console queries are an incomplete Google Search input, not a full assistant-demand inventory.

[Enter your response]

**Review decision** — Mark accept, revise, exclude or needs research, with rationale, reviewer and date. Assign a prompt ID only under your team’s registration process; a generated idea is not observed demand.

[Enter your response]

## 3 Map the content requirement

Several related questions can use the same page. Inspect existing owners before requesting new content, then specify the evidence the reader needs.

**Candidate page and owner** — Record existing destination and intended canonical owner or unresolved status. Identify related wording that belongs on the same page.

[Enter your response]

**Required information** — State the answer, specification, condition, comparison or policy link the buyer needs. Name its source and a visible locator if already present.

[Enter your response]

**Content decision** — Choose reuse, improve, deeper linked material or further research. Explain why the proposed scope is useful and who owns the decision.

[Enter your response]

| Fictional TrailBottle question | Stage | Content needed |
| --- | --- | --- |
| What is it and who is it for | Discovery | Product overview |
| How long does water stay cold | Comparison | Tested condition and limitation |
| Will it fit my pack pocket | Purchase | Dimensions and fit |

## 4 Register the reviewed set

Keep rejected, unresolved and accepted candidates distinguishable. The six-record fictional source example contains two questions at each stage and maps to three pages; it is not a demand estimate.

**Candidate register** — Retain question IDs, source references, stages, page mappings, dispositions and reasons. Record counts and exclusions without treating them as market volume.

[Enter your response]

**Selected panel and settings** — Link the approved prompt portfolio, version, scope, exact wording and intended pages. Define any separate answer-collection settings before a later audit.

[Enter your response]

**Review and follow up** — Assign missing-evidence tasks and the next review date. Evaluate page coverage, answer observations, traffic and business outcomes separately; a mapping is not proof of discovery or conversion.

[Enter your response]

## Complete guide

Prompt research for AI search starts with real buyer questions, not guessed keyword variations. Collect questions from sales, support, interviews, and search data; group them by buyer task; turn each task into a small set of natural prompts; then map every prompt to the page that should answer it. Review the list with customer-facing teammates before publishing.

Original guide update: 2026-09-07. Worksheet prepared 21 September 2026. The original guide and its evidence records retain their dates. The worksheet is an original manual form, not a provider-approved measurement standard.

## Step by step

1. **Collect real questions** Gather wording and question ideas from sales, support, interviews, site search, Search Console, reviews, community discussions, keyword research, and competitor comparisons. Keep the source and date beside every idea; do not present guessed volume as measured demand.

2. **Group each buyer task** Sort questions by what the person needs to discover, compare, or decide before purchase. Use discovery, comparison, and purchase intent as a simple starting set, then add detail only when it helps your team act.

3. **Write and map natural prompts** Turn each task into one clear question a buyer might ask an AI assistant, preserve context that changes the answer, and choose the product, guide, comparison, or support page that should answer it.

4. **Review with your team** Ask sales, support, product, and legal owners to remove invented demand, merge duplicates, and flag answers that need evidence before you use the list.

## Where should buyer-intent questions come from?

Start with language customers already use: discovery calls, demos, objections, support requests, interviews, site search, reviews, community discussions, and Search Console queries. A short list with a traceable source is more useful than hundreds of generated variations.

Keyword research and competitor comparisons can reveal ideas you missed, but keep their source clear and verify them against customer evidence. Search Console is useful but incomplete because privacy omissions and table limits mean not every query is shown.


## How do you turn a buyer question into an AI search prompt?

Write the prompt as a complete, natural question. Add only context that changes the answer: the buyer, use case, product category, location, budget, compatibility need, or decision stage.

Do not create a separate page for every wording variation. Google recommends useful, people-first content and says its systems understand synonyms and meaning; one strong page can answer several closely related phrasings.


## Which prompts should you keep?

Keep prompts that represent a real decision and can be answered with information you can support. Give priority to questions that expose a missing specification, unclear comparison, unsupported claim, buried policy, or weak next step.

Drop duplicates, prompts outside your market, and questions your business cannot answer responsibly. Record uncertain ideas separately until customer evidence or search data gives you a reason to include them.


## What does a simple buyer example look like?

This fictional example uses TrailBottle, an imaginary insulated bottle from Aster & Vale. A support question, interview, and sales conversation are turned into plain buyer prompts, then mapped to the information each page needs.

| Buyer stage | Natural prompt | Content needed |
| --- | --- | --- |
| Discovery | What is an insulated trail bottle and who is it for? | Product overview |
| Comparison | How long does TrailBottle keep water cold? | Tested specification |
| Purchase | Does TrailBottle fit a standard backpack bottle pocket? | Dimensions and fit |


## How should prompts map to your website?

Send product identity, fit, specifications, proof, price, and purchase questions to a useful product page. Use comparisons for genuine alternatives, support pages for care or troubleshooting, and policies for delivery, returns, privacy, or warranties.

A prompt map is a content plan, not a reason to publish thin pages. If one page can answer a related group clearly, improve that page and link to deeper material where the buyer genuinely needs it.

- [Use the product-page checklist](https://rankecho.io/resources/fix/product-page-for-ai-search)

## What this can't tell you

Prompt research does not reveal a provider's hidden query volume, ranking rules, retrieval process, or likelihood of showing your site. Customer questions and search data also reflect different populations, so do not turn the list into a demand forecast.

A mapped prompt is not proof that an answer system used, cited, or recommended the page. Test published pages separately and keep traffic and business outcomes in their own analytics views.


## How can RankEcho help after the manual review?

Once your team has a small prompt list tied to real sources, RankEcho's AI Visibility Audit can help you organize repeatable checks across selected AI tools. Keep the original buyer source and page map beside any tool output so the work remains tied to a real customer task.

Use the prompt portfolio template if you want a reusable place to record audience, intent, owner, and target page before you run an audit.

- [Review the AI Visibility Audit](https://rankecho.io/product/ai-visibility-audit)
- [Use the prompt portfolio template](https://rankecho.io/templates/prompt-portfolio)
- [Track AI referrals in GA4](https://rankecho.io/resources/prove/ga4-ai-referrals)

## What is AI search prompt research?

It is the process of collecting real audience questions, grouping them by task or buying stage, writing natural prompts, and mapping each prompt to the page that should answer it.

## Is GEO prompt research different from keyword research?

They overlap, but prompt research often preserves a complete question, audience, constraint, and decision context. Keyword data can inform the work, but it should not replace customer language and page-level usefulness.

## How many AI search prompts should I start with?

Start with a small set your team can explain, map, and review. Expand only when a distinct buyer task or supported question would change the content or measurement plan.

## Should every prompt have its own page?

No. Closely related questions often belong on one useful page. Create a separate page only when the search intent and answer deserve their own complete resource.

## Can an AI tool generate the prompt list for me?

It can help rephrase or organize questions, but a person should verify that every retained prompt reflects a real audience need and that the mapped answer is accurate.

## Preserved dated evidence

- [Google Search: optimizing for generative AI features](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide) — checked 2026-09-07. Google says established SEO practices still apply to its generative AI features, recommends useful people-first and non-commodity content, and says there is no special AI markup or required writing format for inclusion. Current Google Search Central generative-AI guidance This first-party guidance supports useful buyer-facing content, clear technical access, and ordinary SEO foundations for Google Search. It does not disclose ranking systems, prescribe this prompt-research worksheet, or promise discovery, inclusion, citation, traffic, or sales. Confidence: High

- [Google Search Console: Performance report overview](https://support.google.com/webmasters/answer/7576553?hl=en) — checked 2026-09-07. Google Search Console documents that its Performance report can show the Google Search queries most likely to show a site, the queries bringing traffic, and query-level clicks, impressions, click-through rate, and average position. Current Google Search Console Performance-report documentation This first-party documentation supports using recorded Google Search queries as one prompt-research input. It does not cover questions asked in other search or assistant products, represent all buyer demand, or predict a future ranking, impression, click, citation, referral, or sale. Confidence: High

- [Google Search Console: dimensions and data groupings](https://support.google.com/webmasters/answer/17011259?hl=en) — checked 2026-09-07. Google Search Console says some queries are omitted to protect privacy and that its tables are truncated to important rows, so the visible query table does not contain every query associated with the property. Current Google Search Console query-dimension documentation This first-party limitation supports treating Search Console as a useful but incomplete input. It does not provide hidden query rows, AI-assistant demand, a total-market estimate, or evidence that an omitted query had no impressions, clicks, buyer relevance, or commercial value. Confidence: High

- [OpenAI Help: searching the web with ChatGPT](https://help.openai.com/articles/9237897-chatgpt-search) — checked 2026-09-07. OpenAI says publishers that want their sites eligible for ChatGPT search should allow OAI-SearchBot and its published IP ranges, while placement depends on multiple factors and is not assured. Current ChatGPT Search help documentation This first-party page supports checking technical eligibility for ChatGPT search. It does not publish a prompt-demand tool, selection formula, product-page template, ranking position, or expected referral outcome. Confidence: High

## Complete fictional source records

These are the original fictional TrailBottle records, not real customer observations.

```json
[
  {
    "id": "Q01",
    "source": "customer-interview",
    "sourcePointer": "synthetic-research/interviews/01",
    "capturedAt": "2026-08-17T09:00:00Z",
    "buyerStage": "discovery",
    "buyerTask": "understand the product category",
    "originalQuestion": "What is an insulated trail bottle?",
    "naturalPrompt": "What is an insulated trail bottle and who is it for?",
    "targetPath": "/products/trailbottle",
    "contentNeed": "A plain-language category and audience answer."
  },
  {
    "id": "Q02",
    "source": "support-ticket",
    "sourcePointer": "synthetic-research/support/12",
    "capturedAt": "2026-08-17T09:05:00Z",
    "buyerStage": "discovery",
    "buyerTask": "check care requirements",
    "originalQuestion": "Can the lid go in a dishwasher?",
    "naturalPrompt": "How do I clean an insulated trail bottle?",
    "targetPath": "/help/trailbottle-cleaning",
    "contentNeed": "Care steps plus the lid exception."
  },
  {
    "id": "Q03",
    "source": "sales-call",
    "sourcePointer": "synthetic-research/sales/07",
    "capturedAt": "2026-08-17T09:10:00Z",
    "buyerStage": "comparison",
    "buyerTask": "compare materials",
    "originalQuestion": "Is this better than a steel flask for hiking?",
    "naturalPrompt": "TrailBottle or a steel flask: which is better for a day hike?",
    "targetPath": "/compare/trailbottle-vs-steel-flask",
    "contentNeed": "A balanced comparison by weight, insulation, capacity, and use case."
  },
  {
    "id": "Q04",
    "source": "customer-interview",
    "sourcePointer": "synthetic-research/interviews/04",
    "capturedAt": "2026-08-17T09:15:00Z",
    "buyerStage": "comparison",
    "buyerTask": "check a key specification",
    "originalQuestion": "How long does water stay cold?",
    "naturalPrompt": "How long does TrailBottle keep water cold?",
    "targetPath": "/products/trailbottle",
    "contentNeed": "The tested condition, duration, and practical limitation."
  },
  {
    "id": "Q05",
    "source": "sales-call",
    "sourcePointer": "synthetic-research/sales/11",
    "capturedAt": "2026-08-17T09:20:00Z",
    "buyerStage": "purchase",
    "buyerTask": "confirm fit before purchase",
    "originalQuestion": "Will it fit a normal backpack pocket?",
    "naturalPrompt": "Does TrailBottle fit a standard backpack bottle pocket?",
    "targetPath": "/products/trailbottle",
    "contentNeed": "Exact dimensions and a reminder to check the buyer's pocket size."
  },
  {
    "id": "Q06",
    "source": "support-ticket",
    "sourcePointer": "synthetic-research/support/19",
    "capturedAt": "2026-08-17T09:25:00Z",
    "buyerStage": "purchase",
    "buyerTask": "understand the return path",
    "originalQuestion": "What if the bottle is the wrong size?",
    "naturalPrompt": "Can I return TrailBottle if it does not fit my pack?",
    "targetPath": "/products/trailbottle",
    "contentNeed": "A direct link to the current return terms and conditions."
  }
]
```

## Related reading

- [Find resources](https://rankecho.io/resources/find)
- [Optimize a product page](https://rankecho.io/resources/fix/product-page-for-ai-search)
- [Track AI referrals in GA4](https://rankecho.io/resources/prove/ga4-ai-referrals)

## Product next step

After review, register selected questions in the prompt portfolio and evaluate the AI Visibility Audit for repeatable checks. This worksheet does not collect answers or demonstrate demand, citations, visits or sales. https://rankecho.io/product/ai-visibility-audit