The prompt set
40 to 60 real buyer questions in your category: category questions, head-to-head comparisons against 3 named rivals, alternatives questions, attribute checks, and problem-led questions with no vendor named at all.
The AI Shortlist Audit
See exactly where you're missing when buyers ask AI who to buy from.
40 to 60 buyer questions, sampled 5 to 10 times each across 4 engines. 3 rivals tracked. Every wrong fact found with evidence. 90-day roadmap, priced. 5 working days.
£899 flat. Credited back in full if you hire us for the fix within 30 days.
40 to 60
Buyer questions built from your exact category and segment
5 to 10
Sample runs per question, so one lucky answer can't hide the truth
4
Engines tracked: ChatGPT, Perplexity, Gemini and Google AI Mode
5 days
From kickoff call to the finished report in your inbox
What's Included
Ten things land in your inbox. Nothing vague.
This is the exact scope, every time. No tiered "basic audit" that turns out to be five prompts and a screenshot.
The sampling
Each question run 5 to 10 times. One AI answer is a coin flip. We report the average and the range, never a single screenshot presented as evidence.
The engines
ChatGPT, Perplexity and Gemini through their APIs. Google AI Mode has no public API, so it's sampled by hand and labelled separately in every chart. Never presented as an API number.
The scorecard
Seven numbers, explained in plain words: category visibility, recommendation rate, share of model, share of voice, attribute accuracy, sentiment and citation authority.
The losing list
The exact questions where a named rival wins and you don't appear at all, ranked by how often it happens.
Wrong facts, with evidence
Every place AI describes you incorrectly: stale pricing, the wrong company size band, countries you don't operate in, features you don't have. Dated and quoted.
Root causes
The top three reasons you're losing, with the evidence behind each one. Usually one of: missing entity data, no third-party mentions, or pages that don't answer the actual question being asked.
The 90-day roadmap
A priced, week-by-week plan to fix what the audit found: entity corrections, page rewrites, comparison pages, and the off-site work that actually moves the needle.
The delivery call
60 minutes, walking through the full report live, so nothing lands as a PDF you have to interpret alone.
A credit, not a sunk cost
Hire us for the fix within 30 days and the full £899 comes off your first invoice. Don't, and you still keep the report, the roadmap and every finding in it.
The Report
One PDF. Never more than 18 pages.
The same shape every time, because a fixed structure is how findings stay comparable audit to audit, and how you can hand it to your board without a glossary.
The summary and the method
One page on the headline number and three sentences of context. One page on exactly how we measured it: prompt count, run count, engines, date range. One page for the scorecard itself.
Where you lose, and why AI gets you wrong
Each losing question, who won, how often. Then every place AI misdescribes you, with dated quotes as evidence.
Competitor share of voice and root causes
A chart of where you stand against the 3 rivals you named, then the top three reasons behind the gap, with the evidence for each.
The roadmap and what happens next
A week-by-week, priced 90-day plan, and one page on exactly what the next step looks like if you want us to run it.
Who This Is For
One niche. Not a general AI visibility check.
B2B software companies in the UK and Europe, £3 to £50 million in revenue, in a category crowded enough that buyers actually ask AI to shortlist it for them. We work HR and people tech most closely today: core HR, payroll and EOR, ATS, performance management, learning, and workforce management.
You already have someone doing marketing
A founder or head of marketing who can approve copy and get things published. The audit is worthless if nobody can ship the fix.
You've noticed the traffic drop
Organic traffic down double digits year on year, or a sales team starting to hear "we asked ChatGPT and it didn't mention you" on calls.
Enterprise sales cycles or pre-revenue
Nine-month enterprise sales cycles and pre-revenue startups don't see the same shortlist behaviour from buyers. Come back once you have a category to defend.
Price
WHAT WE WON'T DO
WHAT WE WILL DO
IF YOU GO FURTHER
£899 flat. Paid up front. Credited back if you hire us.
Named agencies in this space start at £3,000 to £4,000 a month before you've seen a single finding. This is a flat fee, paid once, and it comes straight off your first invoice if you hire us for the fix within 30 days. Say no, and you still keep the report.
Guarantee a recommendation. No one can promise what a model will say. Anyone who does is lying or about to spam.
Report real, sampled data with a range, every time. Never a single screenshot presented as proof of anything.
The 90-Day Sprint (from £5,000) or a monthly retainer (from £2,500) carries the same share-of-model guarantee: no movement in 90 days, we work free until it does.
Questions
What people ask before they book the call.
Why does this cost money if the point is to show me I'm invisible?
Because a person who paid £899 shows up to the delivery call and actually reads the report. Free reports get skimmed and ignored. And it comes back off your bill in full if you hire us within 30 days, so if you go ahead, the audit ends up costing nothing.
What exactly do I need to give you to start?
Your company URL, the names of 3 competitors you actually lose deals to, your approximate company size band, and the countries you sell into. That's it. We confirm all of it on a short kickoff call before anything runs.
How is this different from just asking ChatGPT myself a few times?
One question asked once is a coin flip. We run 40 to 60 real buyer questions, 5 to 10 times each, across 4 engines, which is the only way to tell a genuine pattern from noise. We also check the facts AI states about you against what's actually true, which takes a defined process, not a chat window.
Will you tell me what to do, or just what's wrong?
Both. The roadmap in the report is priced and sequenced week by week. You can hand it to your own team, hire us to run it, or take it to another agency. It's yours either way.
Do you use my real company name and rivals in the prompts?
Yes. Generic prompts produce generic findings. We use your actual category phrase, your actual competitor names, and the size and country bands your real buyers fall into.
What happens on the delivery call?
60 minutes. We walk through the scorecard, the losing questions, the wrong facts, and the roadmap together, and answer anything that isn't clear. No slide deck, no pitch beyond the roadmap itself.
Ready
Book the call. We confirm your category and 3 rivals, then we run it.
30 minutes. You leave with a start date and a delivery date, 5 working days out.
£899 flat. Credited back in full if you hire us within 30 days.