AI SEO Agency UK: an independent buyer's guide
I'm not an SEO agency. I built TurboAudit, an AI search visibility tool. I have no agency to sell you. Here's how I evaluated twelve UK AI SEO agencies on disclosed criteria — including the ones that scored badly.
Most UK 'AI SEO agencies' are traditional SEO agencies with AI page veneer. The six things that distinguish a real one are verifiable in thirty minutes. Below is the framework, a scored shortlist of twelve UK agencies, and the red flags that should disqualify one immediately.
I am not an SEO agency. I built TurboAudit, a tool that measures AI search visibility. I have no agency to sell you and no commercial relationship with any agency named below. Some of these agencies are TurboAudit prospects in theory; none currently are. Scoring is based on what is publicly verifiable on each agency's own site and through public searches. Agencies that scored poorly may improve and would do so visibly — the scores reflect the state of public evidence as of June 2026.
Six things a real AI SEO agency actually delivers
Strip the marketing language and a credible AI SEO agency ships six workstreams. Use this list as a comprehension check during discovery. If a prospective agency cannot discuss at least four of these specifically, with examples from past clients, you are looking at a traditional SEO agency that added an 'AI' page to its site in 2025.
- 01
LLM citation tracking across the engines that matter
Monitoring of brand mentions, citations, and source-links across ChatGPT, Perplexity, Google AI Overviews, Google AI Mode, Gemini, Claude, and Copilot. Each engine returns different answers; an agency that tracks only one or two is not running real GEO. They should be able to name their monitoring stack (Profound, Peec, Otterly, TurboAudit, in-house) and discuss how they handle LLM non-determinism.
- 02
AI crawler accessibility audit and llms.txt
robots.txt rules for GPTBot, ClaudeBot, PerplexityBot, ChatGPT-User, OAI-SearchBot, Google-Extended. If the agency hasn't audited a client's crawler access, the rest of their work is wasted on whichever engines can't see the site. They should also have shipped llms.txt for at least one client (proposed late 2024, common by 2026).
- 03
Entity authority engineering
Consistent brand naming across properties, knowledge graph coverage, Wikipedia or Wikidata presence where editorially defensible, schema markup for the brand entity. Generative engines disambiguate entities before answering questions about them; the agency's job is to make that easy.
- 04
AEO content restructuring
FAQPage schema on relevant pages, question-format H2s, answer capsules in the first thirty percent of every important page (Kevin Indig's analysis of 1.2M ChatGPT responses found 44.2% of citations come from that window), definitional pages for category-defining terms.
- 05
Cited-source citation engineering
Earned mentions on sources LLMs already cite. Chen et al. (2025) documented a systematic preference in generative engines for earned third-party sources over brand-owned content. A real AI SEO agency targets specific publications and review sites; generic link building is not the same workstream.
- 06
Reporting against citation share, not keyword rank
The deliverable that proves the work happened. A real agency reports on mention share, citation share, and source-link share across the engines they monitor. If their report is a Google rank tracker with 'AI Overview triggered: Y/N' added as a column, the work being measured isn't AI SEO.
The 30-minute vetting test
Run this before any discovery call. Each step takes five to ten minutes. By the end you'll know whether the agency is worth the hour-long sales meeting.
- 01
Search the agency's own brand in Perplexity
Type the agency name. Do they appear? If they show up in a relevant context (cited as a source on a GEO query, mentioned in industry coverage), they have visible AI search presence. If they don't appear at all on their own brand name, ask yourself why — they sell this skill.
- 02
Verify three case-study clients
Pick three named clients from their case studies. Run prompts that should surface each client in ChatGPT or Perplexity (their main product category, their brand name, their major competitor comparison). If the clients don't appear in answers where the agency claims to have moved them, the case study is theatre.
- 03
Check the agency's robots.txt
Visit https://[agency-domain]/robots.txt. Look for explicit rules covering GPTBot, ClaudeBot, PerplexityBot, ChatGPT-User, OAI-SearchBot, Google-Extended. No rules at all is a yellow flag (they may have an opinion they haven't shipped). Blocking the AI crawlers is a real red flag — you are about to hire someone to do work they have explicitly opted out of for themselves.
- 04
Audit their last five blog posts
Open the last five posts on their blog. Each should have a visible author byline with credentials, a published date and updated date, FAQPage schema where appropriate, answer capsules in the first thirty percent of the post, and external citations to authoritative sources. An agency selling AEO content whose own content doesn't follow AEO conventions is selling something they don't practice.
- 05
On the discovery call, ask them to live-search
First five minutes: 'Pull up Perplexity and search [a relevant prompt] for me right now.' Their response — willing or unwilling, fast or stumbling, confident with the data or surprised by it — tells you more than the rest of the deck. An agency that lives in AI search dashboards does this in fifteen seconds.
Twelve UK AI SEO agencies, scored
The shortlist below covers twelve UK-based agencies that publicly position around AI SEO, GEO, or AEO. Inclusion is not endorsement; exclusion is not condemnation. The score reflects publicly verifiable evidence on the four criteria — deliverable count (0–6 from the deliverables list above), pricing transparency, verifiability of AI-specific claims, and overall fit — as of June 2026. Some agencies score badly. Read the notes for the reason, not the number alone.
Each agency was assessed against four signals: (a) how many of the six deliverables their own site discusses with specifics, (b) whether their pricing is publicly disclosed, (c) whether their AI claims are verifiable (case studies with verifiable outcomes, named tools, screenshots), (d) overall coherence of their AI SEO positioning. The score is a weighted composite. None of these agencies were contacted for the audit. Findings reflect public evidence only and may be incomplete; agencies that improve their public surfacing would visibly move up.
- 01
Found 8/10 · London / New York
What works. Discusses GEO, AEO, entity optimisation, LLM performance tracking, and llms.txt explicitly. Named case study (AllClear: 292% visibility, 50% YoY transactions). Proprietary internal tooling (Luminr) suggests real measurement infrastructure.
Watch-outs. No pricing disclosed; positioning is enterprise. Whether they actually deliver this scope at mid-market budgets is unverifiable without a sales conversation.
- 02
Builtvisible 8/10 · London
What works. Long-standing technical SEO reputation; published technical research; team includes recognised individual contributors in the AI search space. Has not yet shipped a dedicated /ai-seo-agency/ page targeting the head term — a structural soft flank that says they don't need it to win clients.
Watch-outs. AI SEO offering is integrated into broader technical SEO retainers rather than a separate product. May be over-skilled (and over-priced) for SMBs that need a focused AI SEO sprint rather than full technical SEO.
- 03
Reposition 6/10 · UK
What works. Published one of the more cited UK AI SEO agency listicles (the one whose ordering many AI Overviews currently synthesise from). Demonstrates GEO/AEO vocabulary fluency.
Watch-outs. Ranks itself #2 on its own 'best UK AI SEO agencies' listicle without disclosing the conflict prominently. That pattern is exactly what your own essay on the listicle penalty argues against.
- 04
Charle 6/10 · London / Manchester / NY
What works. Published a substantive AI SEO agency comparison piece (last updated Feb 2026). Real founder byline. Active in the AI SEO conversation.
Watch-outs. Ranks itself #1 on its own listicle. Self-promotional pattern dampens credibility on the comparative claims.
- 05
Blue Array 7/10 · Reading, UK
What works. Strong technical SEO foundation; explicit AI search content offerings; recognised in industry coverage as a credible specialist. Active in industry conversations including AI search.
Watch-outs. Mid-to-enterprise positioning; less suited to SMB budgets. AI SEO is one offering among many traditional ones.
- 06
Rise at Seven 6/10 · Sheffield / London
What works. Strong digital PR and earned-media practice — directly relevant to the citation engineering deliverable. Brand recognition in UK search marketing.
Watch-outs. Digital PR DNA dominates positioning; whether they've built dedicated AI search monitoring infrastructure is not visibly documented on their site.
- 07
Impression 7/10 · Nottingham / London
What works. Substantive technical SEO depth; team has published on AI search topics; B-Corp positioning suggests editorial discipline.
Watch-outs. Broader full-service agency. AI SEO is a service line within a larger portfolio, not a focused practice.
- 08
Hallam 6/10 · Nottingham
What works. Well-established UK agency with B2B SaaS specialism; published thinking on AI search. Some pricing guidance disclosed.
Watch-outs. AI SEO positioning is recent. Practice depth in the discipline harder to verify than in traditional SEO.
- 09
PN Digital 4/10 · London
What works. Has a dedicated /ai-seo-agency/ page — explicit targeting of the category. Mentions GEO and AI platforms.
Watch-outs. Page is heavy on marketing language ('intelligent algorithms', 'supercharging') and light on specific deliverables. No FAQ, no pricing, no author, no date. Closer to a category capture page than an evidence of AI SEO practice.
- 10
Digital Search Group 5/10 · UK
What works. Pricing disclosed (campaigns from £750/mo) — rare transparency. Long-established UK agency with documented methodology.
Watch-outs. AI positioning is retrofit on a general SEO services site. Mentions LLM citation building briefly but doesn't show specific AI-search deliverables.
- 11
Propeller 5/10 · London
What works. Dedicated AI SEO service page. Mentions AI platforms by name. Reasonable structural framing.
Watch-outs. Service page mostly describes what AI SEO is rather than what they specifically do. Limited public evidence of executed work.
- 12
ClickSlice 5/10 · London
What works. Active in the AEO conversation; positions specifically around answer engine work. Published thinking on the AEO/GEO space.
Watch-outs. Smaller boutique; capacity and case-study depth less visible than enterprise-tier comparators.
UK pricing reality
UK AI SEO agency pricing is mostly opaque. The bands below are synthesised from agencies that publish pricing (Digital Search Group, some Hallam guidance), industry surveys, and inference from comparable traditional SEO retainers with a documented 'AI' premium.
SME / Boutique
£800–£1,500/moLimited scope — usually one deliverable per month (audit OR content sprint OR monitoring setup), not all three. Best suited to brands with under £2M revenue testing whether AI SEO is worth fuller investment.
Mid-market
£1,500–£3,000/moTwo or three workstreams concurrently (monitoring + audit + monthly content optimisation). Most common tier for UK SaaS and DTC brands. The pricing sweet spot for an agency with genuine AI specialism but not enterprise capacity.
Mid-enterprise
£3,000–£6,000/moFour or more workstreams, dedicated account manager, weekly reporting. Citation engineering and digital PR included. Fits brands with £5M+ revenue and a content team the agency can plug into.
Enterprise
£6,000–£12,000+/moFull GEO programme integrated with traditional SEO and PR. Dedicated team, custom dashboards, multi-region monitoring. Found, Builtvisible, Reposition (enterprise tier), Blue Array all land here.
AI-specific deliverables typically add £500–£1,500/month to a comparable traditional SEO retainer. If an agency is charging double for the same deliverable list with 'AI' prepended, the premium is a label, not a scope change.
Five red flags that should disqualify an agency
If two or more apply, walk away. Each one individually has explainable edge cases; the combination does not.
- 01
'AI SEO' or 'GEO' used twenty-plus times on the site with no concrete deliverables
Marketing-language density without a scope-of-work attached is the signature of a traditional SEO agency that added a category page in 2025. Look for specific verbs (audit, monitor, restructure, build, report) attached to specific objects (robots.txt, FAQPage schema, citation share, llms.txt). If the verbs and objects are missing, the deliverables are missing.
- 02
Self-ranks #1 on their own 'best UK AI SEO agencies' listicle with no disclosure
This is the pattern Lily Ray documented in February 2026 as the trigger for visibility losses across multiple B2B SaaS sites. An agency that runs this play on their own marketing is running the playbook Google is downweighting — and is the playbook they will run for your brand.
- 03
Claims to get clients cited in ChatGPT but their own brand doesn't appear on relevant queries
Eat-your-own-dog-food test. Search the agency's own brand name and the category queries that should surface them in Perplexity and ChatGPT. If they don't appear, ask them why on the discovery call. The answer is revealing.
- 04
Conflates backlinks with AI citations
Backlinks and AI citations are different mechanisms. Backlinks are HTML href elements; citations are brand mentions inside generated answers. An agency that says 'we get you cited in AI by building backlinks' is selling 2018 SEO with a new label. Real citation engineering targets sources LLMs already pull from — not the same as link building in general.
- 05
No first-party measurement tooling — they're reselling vendor dashboards with no methodology view
Most agencies use third-party monitoring tools (which is fine). An agency that has no opinion about the tool's methodology, no view on its limitations, no internal QA when data looks wrong, is renting a dashboard, not delivering insight. Ask which tool they use, what they think of it, and what they do when the numbers don't match a manual check.
Ten questions for the discovery call
Each question is designed as a discriminator — one that a real AI SEO agency answers cleanly and a bluffer cannot.
- 01
Pull up Perplexity right now and search [a relevant prompt for a current client]. What appears?
Real agencies do this in 30 seconds. Bluffers offer to send a report.
- 02
What's in your own robots.txt for GPTBot, ClaudeBot, and PerplexityBot?
Real agencies have a position and can defend it. Bluffers don't know what's there.
- 03
Walk me through your monitoring stack and how you handle LLM non-determinism.
Real agencies name specific tools and discuss statistical sampling. Bluffers say 'proprietary AI dashboard'.
- 04
Show me an llms.txt file you've shipped for a client.
Real agencies show one. Bluffers either don't know what llms.txt is or call it unnecessary without explaining why.
- 05
What's your typical scope-of-work — the bullet list of monthly deliverables and cadence?
Real agencies have a tunable standard SoW. Bluffers describe 'flexible engagement' or 'whatever you need'.
- 06
What does success look like at month 3, month 6, month 12? Leading versus lagging indicators?
Real agencies separate leading (schema deployed, crawl access granted, content restructured) from lagging (citation share). Bluffers describe success in revenue terms at month 3.
- 07
What percentage of your clients have measurable citation share growth, and how do you prove it?
Honest answer is a fraction with a methodology. 'All of them' is a lie.
- 08
How do you handle a client whose product genuinely isn't surfacing in AI answers because the product category isn't AI-queried much?
Real agencies tell you when AI SEO isn't the right investment. Bluffers sell you anyway.
- 09
What's your view on the GEO agency / AI SEO agency / AEO agency taxonomy? Which are you?
Real agencies have an opinion and explain why they picked their label. Bluffers say all three.
- 10
Have you ever fired a client? Why?
Tests whether the agency has standards. Agencies that have never fired anyone take everyone — including clients they cannot serve.
Frequently asked questions
What does an AI SEO agency actually do differently from a traditional SEO agency?
Six things specifically: cross-engine citation tracking (ChatGPT, Perplexity, AI Overviews, Gemini, Claude, Copilot); AI crawler accessibility audits including llms.txt; entity authority engineering with brand schema and knowledge graph work; AEO content restructuring with FAQPage schema and answer capsules; citation engineering targeting sources LLMs already pull from; and reporting against citation share rather than keyword rank. If a UK agency doesn't discuss at least four of these in discovery, they are a traditional SEO agency with an AI marketing page.
How much does an AI SEO agency cost in the UK?
Realistic 2026 bands: SME / boutique £800–£1,500/mo (one workstream at a time); mid-market £1,500–£3,000/mo (most common tier); mid-enterprise £3,000–£6,000/mo (dedicated account management, weekly reporting); enterprise £6,000–£12,000+/mo (full programme integration). AI-specific deliverables typically add £500–£1,500/mo to a comparable traditional SEO retainer.
Are AI SEO agencies worth it for a UK SME?
Conditionally. They are worth it when (a) your annual customer LTV is high enough to ROI on a £1,500+ monthly retainer, (b) you've already shipped basic GEO foundations (FAQPage schema, AI crawler access, author bios), and (c) you have someone internally who can act on the agency's reporting. They are not worth it when your category isn't being heavily queried in AI answers, when your CMS doesn't allow the schema work the agency needs, or when you'd be better served by a £200/mo monitoring tool plus a freelancer.
How do I tell if a UK agency's AI claims are real or just marketing?
Run the 30-minute vetting test in this guide: search the agency's own brand in Perplexity, verify three case-study clients' AI search presence, check their robots.txt for AI crawler rules, audit their last five blog posts for AEO conventions, then on the discovery call ask them to live-search a relevant prompt in front of you. The combination of those five checks separates real practice from veneer in well under an hour.
Should I hire a UK AI SEO agency or do AI SEO in-house?
In-house is cheaper long-term if you have a content team and an SEO lead willing to learn. Agency makes sense when you need execution capacity in addition to strategy and have budget for £1,500+/month. The mistake most UK SMEs make is hiring an agency without an in-house counterpart who can interpret the reporting — in which case the agency invoice is for a PDF nobody reads.
Can a traditional UK SEO agency adapt to AI search, or do I need a specialist?
Some traditional agencies have adapted credibly (Builtvisible, Blue Array, Hallam, Impression, Found) by adding genuine AI deliverables alongside their existing practice. Others have added an 'AI SEO' page to their site without changing what they ship. The discovery questions in this guide separate the two. A traditional agency that ships the six deliverables is fine; a specialist is unnecessary if your incumbent already does the work.
- Aggarwal et al. — GEO: Generative Engine Optimization (KDD 2024)arxiv.org →
- Lily Ray — Is Google Finally Cracking Down on Self-Promotional Listicles? (Feb 2026)lilyraynyc.substack.com →
- Surfer — Scraped AI answers vs API results: a 1,000-prompt study (Dec 2025)surferseo.com →
- Chen et al. — Earned-media bias in generative engines (arXiv 2509.08919)arxiv.org →
- Kevin Indig — ChatGPT citations: 44% from the first third of contentalmcorp.com →
- OpenAI — GPTBot crawler documentationplatform.openai.com →
- Anthropic — ClaudeBot crawler documentationanthropic.com →
- Wikipedia — Generative engine optimizationwikipedia.org →
- Search Engine Land — What is Generative Engine Optimization (GEO)?searchengineland.com →
Audit your own site before any agency call.
TurboAudit runs the audit step of the AI SEO agency workflow on every page of your site — schema coverage, AI crawler accessibility, content extractability, citation density, author attribution. Walking into an agency discovery call with your own audit changes the conversation. From $49/mo.