What does a GEO or AEO audit cost in the UK? 2026 benchmarks

Many UK agencies that sell a GEO (generative engine optimisation) or AEO (answer engine optimisation) audit do not publish a price at all: of the 20 UK GEO agencies in Tilio's list, 14 publish none. The prices that are public sit in a fairly narrow band, and the real differences between offers are what is checked, how many times, and who signs the findings off.

This guide sets out the published prices as read on 1 November 2026, how US bands compare, what a sound audit should contain, the red flags to check before you pay, and how to scope the work. Providers may quote differently from what they publish, and we re-check these figures monthly.

The short answer

Where UK providers publish a price for a one-off audit of how a brand appears in AI answers, it runs from £300 to £1,500. One UK pricing guide for B2B SaaS firms puts GEO work at £3,000 to £8,000, but it does not say how it arrived at that range. Monthly GEO retainers are published from £450 a month at the low end. US guides quote higher bands, in dollars, for differently scoped work.

For scale, Whito's June 2026 analysis of 45 published UK SEO retainer prices found a median of £799 a month. A published UK audit price is therefore roughly one to two months of a typical small retainer. What that money buys varies a great deal, which is why the checklist further down matters more than the headline figure.

Published UK prices

The table lists the audit and retainer prices we found published on 1 November 2026. Audit prices are one-off; retainer prices are per month. The ProfileTree and Novi figures come from Whito's round-up rather than the providers' own pages, and the retainer figures from Tilio's list.

Two cautions. These are the prices providers chose to publish, and 14 of the 20 agencies in Tilio's list publish none, so the public figures may not represent what most agencies charge. And a price on its own says nothing about scope, so compare what each quote includes using the checklist below.

For context on retainers generally, the same Whito analysis puts established agencies at £1,500 to £5,000 a month or more.

How US prices compare

US figures are in dollars and describe different packages, so treat them as context rather than as a UK benchmark.

• A US agency-network pricing guide puts GEO audits at $2,000 to $8,000 and retainers at $2,000 to $15,000, with white-label wholesale work at $3,000 to $8,000 a month. It does not state its sample.
• A separate breakdown splits US audits into productised audits at $440 to $1,500, consultant-led audits at $1,500 to $5,000 and sprints at $5,000 to $15,000, and notes that audit fees are often credited against a retainer, as Apex does in the UK.

The useful point in the second breakdown is that it ties price to scope. A productised audit, a consultant's audit and a multi-week sprint are different things, and their UK equivalents should be compared like for like.

What a sound audit should contain

A price means little until you know what was checked. Before comparing quotes, ask each provider for these in writing.

1. The prompt set. The actual buyer questions, written without the brand's name, agreed with you before collection and versioned, so a later re-run asks exactly the same thing.
2. The engines, and how each was asked. Which assistants or models were asked, and whether each answer came from an API or from the consumer app, with or without web search. An analysis of how AI visibility tools collect data argues that disclosure matters more than which method is used. A provider who will not say how answers were collected cannot defend the result.
3. Runs per prompt. AI answers change from one request to the next. In a study of 2,961 runs across 12 prompts, the same list of brands came back in fewer than 1 in 100 runs; the study was co-run with Gumshoe, which sells an AI visibility tool. SparkToro's follow-up argues that a rate measured over many runs is steadier than any single answer. Ask how many runs sit behind every figure.
4. The verbatim answers. Every answer stored word for word, with the date, the model and any sources it cited, so you can check each finding against the text yourself.
5. The cited sources. For engines that search the web, the URLs their search returned. These show which third-party pages are shaping the answer.
6. A named reviewer. Who read the answers and signed the findings off, and when. An automated classification is a starting point, not a finding.
7. Remediation. A prioritised list of work, each item tied to the evidence that prompted it, with an owner and a review date.
8. The limits. What the audit does not cover, such as Google AI Overviews, Copilot or the consumer apps, if they were not checked.

Our AI visibility report template shows how these pieces fit into one client-ready document.

Red flags before you buy

• A single score with no denominator. A visibility percentage that does not say how many questions, engines and runs it was counted over cannot be checked. Ahrefs, which sells its own AI visibility product, puts it plainly: "Think of AI tracking less like rank tracking and more like polling." A London agency makes the same point, calling single percentages false precision.
• Positions sold as rankings. In the study above, the same brands in the same order came back in fewer than 0.1% of runs. A position taken from a single pass tells you very little.
• Collection that is not disclosed. If the provider cannot tell you whether answers came from an API or an app, which models answered, or how many times each question was asked, the figures cannot be reproduced.
• Guaranteed citations, rankings or traffic. Nobody controls what a third-party answer engine says. A promise of placement is a promise that cannot be kept.
• A checklist sold as the main fix. Some audits lead with an llms.txt file or schema mark-up. A study of 300,000 domains found no clear effect of llms.txt on AI citations, and an Ahrefs-sponsored analysis reports schema changes moving citations by +2.2%, +2.4% and −4.6% against controls: small effects, in both directions. Neither belongs at the centre of an audit.

How to scope an audit

Whoever you buy from, agree these points before any collection starts.

• One brand, one market. A second brand or market is a second piece of work; mixing them blurs every count.
• The questions. A fixed set of unbranded buyer questions, written the way a buyer would ask them. Fix the number before collection: more questions give a broader picture but take longer to review.
• The engines and the method. Name the engines, say whether each is asked through an API or read in the app, and decide which consumer surfaces, such as Google AI Overviews, a person will check by hand.
• Runs. Several runs of each question on each engine if you want to see how much the answer moves (AnswerLift's default is three), reported as counts with their denominators and dates.
• Sign-off. Who reviews each record, and whether anything unreviewed can reach the client.
• The deliverable. The evidence behind every figure, the prioritised remediation list and the client report, in a format agreed up front.
• Time and credit. How many working days the work takes, and whether the fee is credited against follow-on work.

If you are comparing software rather than services, our guide to AI visibility tools for agencies lists published UK prices and how each tool collects its data.

AnswerLift's prices, and why the Sprint costs more

For transparency, these are AnswerLift's own prices, not market benchmarks. Answers are collected from the providers' APIs through the OpenRouter gateway; Gemini is asked without search grounding; Google AI Overviews are checked by hand, never collected automatically; and no score is produced anywhere. Each step is set out on the methodology page. No VAT is added, because Nikah AI Ltd is not VAT registered.

• Free Snapshot. One brand and three unbranded buyer questions put to the models behind ChatGPT, Claude, Gemini and Perplexity, with every answer checked by a person. Request one here.
• Verified Baseline, £1,250 by invoice. 10 to 15 prompts across four engines, three runs each, with five prompts also checked by hand, delivered in five working days. The fee is credited against a Visibility Sprint bought within 30 days.
• Visibility Sprint, £2,750 one-time. Each question is asked three times of each engine on the panel by default, every figure is reported as "named in k of n runs", and a named reviewer signs off every record. You receive the verified evidence ledger, a prioritised remediation backlog and one client-report link within 20 business days of kickoff.
• Monthly Verified Snapshot, £345 per brand per month by invoice. 25 prompts across four engines, run weekly, with one reviewer sign-off a month, week-by-week counts on the report and a white-label report link.

£2,750 is above every audit price in the table; only the £3,000 to £8,000 range from the SaaS guide, which does not state its basis, is higher. The provider cost of collecting the answers is a small part of the fee. Most of it pays for a person reading every stored answer and signing each record off by name, for the verbatim evidence behind every count, and for a report you can hand to a client. If you want a quick read of the landscape, the free Snapshot or a cheaper audit may be the better buy. If you need evidence you would put your agency's name to, compare each quote against the checklist above.