The questions buyers ask AI before Googling you

Keywords told you what people typed. Prompts tell you what people decide. We surface the decision-stage queries where AI quietly compare brands — and place yours in front of them.

4+1 beta

Engines tracked

15.1words avg

Real-prompt length

78.9% intent

Decision-stage focus

Prompt

“Best CRM that small teams can set up in a day without a sales engineer — under $40/seat, with HubSpot-grade automation.”

1. Pipedrive

64 %

2. Your brand

47 %

3. Folk

38 %

4. Attio

31 %

5. Close

22 %

1 / 3 · click to advance

01 · The shift

From keywords to prompts

Search used to be a list of strings. Now it's a conversation. The unit of measurement changed — and so did everything downstream: the data sources, the metrics, the optimisation loop

The dimensionWhat changed

What we used to trackKeyword research

What we track nowPrompt research

Unit of measurement

Keywords, pages, ranking positions on a SERP.

Conversational prompts, brand mentions, recommendation context.

Popularity data

Years of search volume, CPC, seasonal trends.

No direct prompt volume — only estimates via search proxies and surveys.

Output behaviour

Relatively stable, predictable ranking positions.

Volatile, personalised, run-by-run answers from each LLM.

What you optimise for

Clicks and SERP positions.

Mentions, citations and the framing AI gives your brand.

Success metrics

Rank, CTR, organic sessions.

Mention rate, citation frequency, sentiment, Share of Voice.

Why personas matter

Useful for targeting — not critical for ranking.

Critical. Persona constraints decide whether AI recommends anything at all.

02 · Workflow

A four-step loop, run on a schedule you set

Prompt research isn't a one-off audit. It's a loop you keep tightening: identify the prompts that matter, simulate them across engines, track who gets named, optimise the gaps

Identify

persona briefs · seed prompts

Define personas and the constraints that push an LLM from "explain mode" into "recommend mode" — budget, role, region, ingredient, stack

Simulate

engine-by-engine snapshots

Run each prompt in parallel through ChatGPT, Gemini, Claude and Perplexity. Record every brand mentioned, every link cited, every ranked recommendation

Track

historical visibility record

Watch visibility, citation rate and competitive overlap drift across days, weeks and engine updates. Catch the moment a competitor takes your slot

Optimise

prioritised content · PR briefs

See which content the AI cites when it does mention you — and which pages competitors own that you don't. Brief them, publish, re-simulate

03 · Inputs

Hunting down the prompts your buyers actually use

No engine ships a "Prompt Console" yet. So we triangulate — combining what your customers say with what your site, your competitors and the public web already reveal

01/05

Google Search Console

Pipe in your GSC and we surface the long-tail interrogatives your site already ranks for — then rewrite them into conversational prompts.

Auto-sync via OAuthRegex isolation of questionsKeyword → prompt rewriter

Your own analytics

GA4 plus server logs tell us which of your pages are already pulling AI traffic (ChatGPT-User, Perplexity-User). We auto-generate prompts around them.

GA4 connectorLog parsing for AI agentsTop-AI-traffic prompt seeds

Internal data — sales calls, support tickets, help docs

Upload CSVs and transcripts. NLP extracts recurring pain points, objections and constraints — then turns them into persona-driven prompts in the language your buyers actually use.

CSV / TXT / JSON uploadNLP pain-point miningAuto persona generation

Competitor visibility gaps — where they show up and you don't

We surface every prompt where a competitor is recommended but you aren't, ranked by reach. These are the highest-leverage gaps in your AI presence.

Automatic gap detectionShare-of-Voice per promptCited-page analysis for outreach

04 · Study

What real prompts look like vs. what we estimate

We compared hundreds of prompts users said they'd type into ChatGPT against the prompts most tools guess they'd type. The gap is wider than you'd expect — and it changes the content you should be making

Key Insight

Users write with conversational constraints and problem statements, not just search keywords.

Content Strategy Takeaway

Focus on tool comparisons and "help me decide" guides

Real / Estimated promptsThe shape of a real ChatGPT query

Avg prompt lengthwords per prompt

15.1

8.8

+71%

Use personal pronounsi / my / me

52.1 %

18.8 %

2.8×

Tool-finding intentwant a specific product

78.9 %

62.5 %

+16pp

Problem-oriented"help me…" "I need…"

21.1 %

7.1 %

3.0×

Conversational tonenatural sentence flow

23.9 %

5.8 %

4.1×

Start with "best"commercial framing

11.3 %

18.8 %

−7.5pp

05 · Output

Track your visibility trends, platform breakdowns, and competitive overlap over time

Primary metric · 28-day window

47%

AI Visibility Score% of prompts in this set that mention Your brand

vs. previous period+12 pp

47 %

61 %

38 %

22 %

Secondary signalslast 7 days

Citation rate

31%

▲ +4 pp

Avg rank when named

2.4of 5

▲ +0.6

Sentiment

+0.42net

flat

Gap prompts

38open

▲ +12

Competitive overlap

Brands AI most often mentions alongside you

Pipedrive68%

Folk54%

Your brand47%

Attio41%

Close29%

Start with one persona. See who AI names instead of you

Tell us about your category in a 30-minute call. We'll show you, live, which prompts your buyers ask AI — and which brands AI names instead of yours

Book a demo See pricing

Treat your Search Console like a prompt mine

Standard queries are statements. Use regex to isolate the interrogatives — the questions people actually ask LLMs.

Filter for questions

Go to Search results

^(how|who|what|where|when|why|which|can|could|do|does|did|is|are|was|were|should|would|will|may|might)\b

6/12

Queries match

54%

Clicks captured

Sample GSC export · top queries, last 28 days

howto track ai search visibility

1,240prompt-able

best ai search monitoring tool 2026

980filtered

whatis generative engine optimization

870prompt-able

chatgpt mention tracker

760filtered

howdo i appear in chatgpt answers

710prompt-able

ai brand visibility platform

650filtered

whydoes perplexity cite some sites and not others

540prompt-able

geo vs seo difference

510filtered

canchatgpt see my website

470prompt-able

share of voice on ai search

420filtered

whichai tools track brand mentions

380prompt-able

ai prompt research process

290filtered