> TomHow / Paid audit

AI search audit. GBP 2,250. Two UK working days.

Find out what your public site and search evidence actually support, how your brand appears in one recorded buyer-question panel, where named competitors are stronger, and which ten fixes matter first. The result is a signed report, a 90-day plan and a findings call.

The TomHow AI search audit costs GBP 2,250 and takes two UK working days. In that time TomHow reviews your public website, your Google Search Console and GA4 evidence where you supply access or exports, dated Semrush evidence where available, and up to 12 priority pages. One fresh 20-question ChatGPT panel is run and recorded, and up to three named competitors are compared against you. You receive a signed report with linked source and evidence rows, ten prioritised fixes, a 90-day action plan and a 45-minute findings call. If you sign a qualifying retainer within 30 days of receiving the audit, the GBP 2,250 fee is credited in full against your first month's invoice. It is a point-in-time diagnostic, not a promise of rankings, citations, leads or regulatory approval.

The audit does not include a multi-provider AI panel, implementation, ongoing monitoring or paid API work. Evidence that could not be checked is labelled unavailable rather than presented as fact.

Pay GBP 2,250 securely - opens in a new tab See what you get Start with the free snapshot
GBP 2,250 one-off
Two UK working days
Linked evidence, signed
Credited back if you retain us
Last reviewed 2026-09-06
> Why the audit is paid

This is not a sales call in PDF form. Atlas records one fresh 20-question ChatGPT panel. Anvil reviews the public site and priority pages. Compass turns the findings into ten prioritised fixes. Thomas Howcroft reviews and signs the final document.

> The credit-back policy

If you sign a qualifying retainer (Citation Engine, AEO Stack, or Search Lab Retainer) within 30 days of receiving the audit, the GBP 2,250 audit fee is credited in full against your first month's retainer invoice. If you do not retain us, you still leave with the baseline, the fixes, and the roadmap.

> Audit method

The audit proves the gap before it sells the fix.

The point is simple: run the questions your buyers use, record who gets named or cited, check why your site is hard to trust, then turn that into a ranked action plan. No vague AI visibility score. No promise of guaranteed citations.

Captured prompt set

The prompts are written down before analysis. The report keeps the exact wording, platform, date, answer state, and outcome for each useful row.

Provider boundary

The included provider is one fresh, recorded 20-question ChatGPT panel. This is not a multi-provider panel. Missing evidence is marked unavailable.

Evidence standard

Every finding needs a source row: prompt, platform, cited brand, cited URL where visible, confidence note, data quality note, and recommended next move.

Action plan

The report turns the evidence into fixes: pages to publish, schema to add, links to build, authority gaps to close, and prompts to keep tracking.

> Prompt set

Twenty buyer questions, captured verbatim.

The audit uses an agreed prompt set, then keeps the exact wording in the report. The aim is not to trick a tool into mentioning you. The aim is to test real buyer questions and see who gets named.

Buyer shortlist prompts

Example intent: who should we hire, which agency is credible, which provider understands this category.

Problem prompts

Example intent: why is my brand not appearing in AI answers, how do I track citations, what does an AEO audit include.

Category prompts

Example intent: GEO agency, AEO consultant, AI search optimisation, LLM citation tracking, AI visibility reporting.

Competitor prompts

Example intent: brand versus named competitors, alternatives, tools, and sources AI answers already lean on.

Audience prompts

Example intent: what a B2B founder, marketing lead, or technical team should fix first.

> Platforms checked

The report separates tools from guesses.

Each platform is recorded separately because AI answers do not behave the same way. Where a platform is unavailable, quota-limited, or does not show a result, the report says that plainly.

Included

One fresh, recorded 20-question ChatGPT panel using the agreed buyer questions.

Evidence

Exact prompt, date, answer state, named brands and visible source URLs are stored in the source ledger.

Excluded

No multi-provider panel, paid API work, implementation or ongoing monitoring is included in this audit.

> Proof standard

The evidence standard changed the offer.

Controlled prompt checks show the honest problem: AI answers can understand a category, mention alternatives, and still leave a brand out. That is why the audit starts with evidence, not a promise.

The category is visible

In captured prompt sets, AI systems can recognise a category and recommend names. The market is not theoretical.

Gaps are measurable

The audit records where a brand is missing, which competitors or sources fill the answer, and which prompts are worth fighting for.

Evidence beats screenshots

Each useful row needs platform, prompt, date, cited brand, cited URL where available, confidence note, and the next fix.

Technical blockers matter

Crawler policy, schema, sitemap hygiene, entity clarity, internal links, and page depth decide whether good content is easy to retrieve.

> Fit check

This is a buying path, not a sales maze.

Good fit

You already have a site, some search demand, and a reason to be cited in AI answers.

Not magic

This does not guarantee AI citations. It shows the current gap and the work needed to improve your odds.

Built to act on

The output is written so a founder, marketing lead, or developer can move without waiting for another sales meeting.

Send the site before you pay.

This prepares an email draft with the information needed for a quick fit check. Nothing is sent automatically. If a call would help, ask for one.

> What you receive

For GBP 2,250, you get exactly this.

No vague score. No hidden dashboard. No extra call funnel.

  • A signed report with linked source and evidence rows, sized to the evidence
  • The captured prompt set and evidence table
  • One fresh 20-question ChatGPT panel
  • Up to 3 competitor comparisons
  • A technical AEO checklist with severity and effort
  • A ranked content and entity roadmap
  • A 45-minute findings call after delivery
  • GBP 2,250 credited back if you retain TomHow within 30 days
01

Recorded ChatGPT baseline

One fresh 20-question panel using agreed buyer questions. The report stores the wording, date, answer state, named brands and visible sources.

02

Website and technical review

The public website and up to 12 priority pages are checked for crawl access, page ownership, schema fit, internal links, source quality and visible buyer answers.

03

Search evidence review

Google Search Console and GA4 evidence are included when you supply usable access or exports. Dated Semrush evidence is used where it is available.

04

Competitor comparison

Up to 3 named competitors are compared against the same buyer questions and page evidence. Missing or unavailable evidence is labelled.

05

Ten prioritised fixes

The findings become ten ordered actions, split by impact, evidence, effort and owner.

06

90-day action plan and findings call

You receive a signed report with linked source and evidence rows, a 90-day action plan and a 45-minute call with Thomas Howcroft.

> Evidence and report shape

Sample sections, with the source rows behind them.

Evidence standard

  • Exact prompt text, not a paraphrase
  • Platform, answer mode, date, and capture status
  • Cited brand and cited URL where the tool exposes one
  • Whether your brand is present, absent, or only weakly mentioned
  • Competitor or source that filled the answer instead
  • Confidence note, data quality note, and next action

Sample report sections

  1. Executive verdict: what is broken, what is fixable, and what to do first
  2. Prompt set: the exact questions used for the audit
  3. Recorded 20-question ChatGPT baseline and source ledger
  4. Competitor citation map: who appears where you do not
  5. Technical AEO findings: schema, llms.txt, crawler policy, sitemap, internal links, and entity clarity
  6. Content and authority gaps: pages, proof assets, and source signals worth building
  7. 90-day action plan: owners, order, evidence and next measurement
  8. Source index: evidence rows and data quality notes

Inspect the public-safe sample first.

The sample uses fabricated data, but the structure, source index, signed narrative, and next-action layer match the reporting format.

Open sample report ->
> The two-working-day timeline

From payment to signed PDF in your inbox.

Start

The clock starts after payment clears, intake is complete, and the agreed evidence is usable.

Working day 1

Atlas records the 20-question ChatGPT panel. Anvil reviews the public site and up to 12 priority pages. Search and competitor evidence is reconciled.

ATAtlasAVAnvilHKHawk
Working day 2

Compass ranks ten fixes. Scribe assembles the report with linked source and evidence rows plus the 90-day action plan. Thomas reviews and signs it, then books the findings call.

CMCompassSCScribe
> Before you buy

Read the pages that explain the offer.

Sample monthly report

A fabricated public-safe report showing the source index, citation section, shipped work, and next actions.

Read resource ->

AEO explainer

How answer-engine work supports clear, useful pages without inventing a separate product.

Read resource ->

GEO explainer

How buyer questions, brand evidence, competitors and source gaps fit the wider service.

Read resource ->

UK AI search agencies and GEO tools compared

How to judge providers by evidence standard, not pitch language.

Read resource ->

What an AI citation tracking report should include

The report structure behind the paid audit and monthly Scribe output.

Read resource ->

How to measure AI search visibility

How to keep provider evidence separate, dated and reproducible.

Read resource ->

Zero traffic recovery plan

The full diagnostic route for proving whether traffic is gone or the tracking setup has broken.

Read resource ->

Measure AI search visibility

How the prompt set is chosen before citation checks, competitor checks, and reporting.

Read resource ->

Search setup guide

How the audit treats entity clarity, service clarity, and machine-readable evidence.

Read resource ->

Search visibility setup

How commercial pages should connect proof, reports, audit pages, and resources.

Read resource ->

Why AI tools name some agencies

The corroboration layer behind AI recommendations: owned pages, third-party evidence, and repeatable source language.

Read resource ->
> Clean audit inquiry

After payment, send the audit details.

  • The site URL you want audited
  • Up to 3 competitors to compare
  • Priority queries if you already know them
  • GA4 and Search Console access, or exports, if you want those included
Send audit details by email
> Buy the audit

One transaction. Two UK working days. A signed plan.

GBP 2,250
One-off | Credited back vs retainer | Pay in advance

Pay through secure Stripe checkout, then send the site URL and competitor details for the audit. The audit starts when payment clears and the intake details and usable evidence are in. The signed report lands within two UK working days.

Pay GBP 2,250 securely - opens in a new tab ->

No AI citation guarantee. The audit shows the current state, the gaps, and the work most likely to improve visibility.

VAT not charged (HOWCROFTS CONSULTING LTD is not VAT registered). Receipt on payment. Audit clock starts when funds clear.