Get recommended by AI.
Prove it with a score.
We track how often ChatGPT, Gemini, Perplexity and Google AI Overviews name your brand for the questions your buyers ask. Then we ship the content, citations and technical fixes that move that number. One score, reported every month.
- Month to month
- Senior team only
- Score you can check yourself
Monthly score report
Example64
Visibility score
Named in 64% of tracked buyer prompts
Target marker: 99
ChatGPT71
Gemini66
Perplexity58
AI Overviews49
- “best project tool for small agencies”Named
- “alternatives to [your competitor]”Named
- “which tool has the best client reporting”Gap
Scored across the engines your buyers ask
ChatGPTPerplexityGoogle AI OverviewsGemini
Claude
Copilot
Grok
Meta AIChatGPTPerplexityGoogle AI OverviewsGeminiClaudeCopilotGrokMeta AIWhy your score is low
Four reasons AI recommends someone else.
Most brands that come to us have some mix of these. Every programme is built as the direct answer to each one.
- 01
AI already has a favourite in your category
Ask ChatGPT or Perplexity for the best tools in your space and a few names come back every time. If yours isn't one of them, that gap widens each month the answers get repeated.
- 02
Your best pages aren't shaped like answers
Long brand pages and top-of-funnel blogs rank, but AI engines lift short, specific, comparable claims. Most sites have very few pages an answer can quote.
- 03
You're missing from the sources AI cites
Engines lean on listicles, review sites, Reddit threads and YouTube videos. If you're absent there, your own site has to carry the whole argument alone.
- 04
Nobody can tell you if it's working
Rank trackers don't see AI answers, and screenshots of one lucky prompt prove nothing. Without a fixed prompt set measured every month, it's guesswork.
Why 99 and not 100? Because no honest agency promises a perfect score. 99 is the direction. The monthly number is the proof.
The score
One number that tells you if AI recommends you.
Your visibility score is the share of your agreed buyer prompts in which AI answers name your brand, measured per engine, every month.
50
buyer prompts tracked
20
answers that name you
40
your score on that engine
Example calculation
The prompts matter more than the number, so we agree them with you in week one: by market, by persona, by buying stage. Then they stay fixed, which is what makes the trend worth reporting.
What we ship every month
Three workstreams. One scoreboard.
Each one targets a reason engines skip you, and each one is measured against the same prompt set.
Answer-ready content
Comparison, alternatives, use-case and FAQ pages built around the prompts in your score, structured so an engine can quote them. New pages and refreshes every month.
Mapped to the prompts you're not named in yet.
Earned citations
Placements on the listicles, review sites, communities and publications the engines actually cite for your prompts. We pick targets from citation data, not from a link vendor's list.
Editorial placements, not link swaps.
Technical access
AI crawler logs reviewed every month, so we know whether ChatGPT's and Claude's bots can reach your pages. Then robots.txt, rendering, schema and llms.txt fixes where they block you.
Fixes shipped, not just listed in an audit.
Every programme also includes
- Prompt set agreed with you in week one
- Baseline score per engine
- Benchmark against your top competitors
- Monthly score report
- AI crawler log review
- Content gap analysis
- Monthly strategy call
- One-page roadmap for next month
Programmes
Start with a score. Scale from there.
Every engagement starts from your baseline, so the scope is set by your gaps, not by a package sheet.
Find out where you stand
Score audit
A one-off baseline: your visibility score per engine on an agreed prompt set, benchmarked against competitors, with a prioritised fix list.
- Baseline score across the major engines
- Competitor benchmark
- Crawler access and technical review
- Prioritised fix and content plan
Scoped and quoted after your free score
Raise the score every month
Growth programme
All three workstreams running monthly on one market, with the score report as the scoreboard. The usual next step after the audit.
- Everything in the score audit
- Answer-ready pages every month
- Earned citations every month
- Technical fixes shipped continuously
Scoped and quoted after your free score
Own the answers in your category
Category programme
For brands competing across several markets, languages or product lines, with a larger prompt set and more output per month.
- Everything in the growth programme
- Multiple markets and languages
- Larger prompt set and more output
- Shopping and sponsored-answer tracking
Scoped and quoted after your free score
Commitments
What we promise. And what we won’t.
11
AI engines in every score
Every engine that matters
ChatGPT, Gemini, Claude, Perplexity, Grok, Llama, DeepSeek, Mistral, Copilot, Google AI Overviews and AI Mode.
1
number to report upward
One score, every month
The same prompts, measured the same way, so the trend means something. Per-engine detail sits underneath.
0
long-term contracts
Month to month
After the audit, you stay because the score moves, not because a contract says so.
100%
yours
You own everything
Content, accounts, data and the prompt set. If we part ways, nothing leaves with us.
What you’ll see, and when
“Is it working?” answered before you sign.
These are the reports you get at each stage. Not guarantees about the numbers in them: commitments about what you’ll be able to check.
Week 1
Baseline
- Prompt set agreed with you, by market and persona
- Baseline score per engine and per competitor
- Crawler access checked against your server logs
Month 1
Fix and ship
- Blocking technical issues fixed
- First answer-ready pages published
- Citation targets picked from the sources engines use
Month 3
First read
- Score trend against your baseline, per engine
- Which prompts moved, and which pages got cited
- Crawl-to-citation path for new pages
Month 6
Prove it
- Six-month score trend benchmarked against competitors
- AI-referred visits and the conversions they produced
- Plan for the next prompt set
The platform behind every score
We measure on Promptwatch.
A mention count tells you very little. To raise a score we need the whole chain: which prompts matter, which pages and domains the engines cite, whether AI crawlers can even reach your site, and what the resulting visits do. Promptwatch puts all of that in one place, which is why our reports can connect the score to revenue instead of stopping at “you were mentioned”.
See Promptwatch- 01Prompt tracking per engine, persona and market, with query fan-outs
- 02Citation analytics down to the page, including Reddit and YouTube
- 03AI crawler logs with the path from crawl to citation
- 04AI-referred visitors and the conversions they produce
Who we work best with
Mutual fit matters.
A good fit
- Brands whose buyers research with ChatGPT, Perplexity or AI Overviews before they buy
- Teams that want one clear number to report upward, with detail underneath
- Companies that can publish the pages we deliver, or give us CMS access
Not for
- Anyone who wants a guaranteed score or guaranteed lead counts
- Teams looking for cheap article volume without measurement
- Brands that only care about classic Google rankings
From the blog
How AI visibility actually works.
What exactly is a visibility score?
The share of your agreed buyer prompts in which AI answers name your brand, measured per engine every month. Say we track 50 prompts and ChatGPT names you in 20 of them: your ChatGPT score is 40. We agree the prompt set with you up front, because the prompts matter more than the number.
Can you guarantee a score of 99?
No, and nobody honest can. AI answers change, competitors move, and engines differ. 99 is the direction we work toward. What we commit to is the work every month and a score you can check yourself.
Which AI engines do you track?
ChatGPT, Gemini, Claude, Perplexity, Grok, Llama, DeepSeek, Mistral, Copilot, Google AI Overviews and AI Mode. The headline score usually weights the engines your buyers actually use.
How long before the score moves?
Technical fixes and new pages can show up within weeks, but a trend you can trust needs a few months of monthly measurement. We report the first read at month three and a benchmarked trend at month six.
Do you still do classic SEO?
Where it feeds AI visibility, yes: crawlability, rendering, schema and pages that rank are often what engines retrieve. We don't sell standalone rank-tracking retainers.
How do you measure it?
We run every client score on Promptwatch. It tracks the prompts across engines, records which pages and domains get cited, reads AI crawler logs, and attributes AI-referred visits and conversions, so one report covers the whole chain from crawl to citation to revenue.
Free baseline
What’s your score today?
Send us your domain and your top three competitors. We’ll come back with your baseline visibility score across the major engines and what we’d fix first. No pitch deck, no obligation.