Most SEO agencies only optimise for Google, don't understand entity SEO, have no AI citation monitoring capability, still prioritise backlinks over brand mentions, don't implement AI-specific schema like Speakable, and have never tested their clients' visibility across ChatGPT, Gemini, or Perplexity. The gap between "we do AI SEO" on the website and actual delivery capability is enormous.

The "we do AI SEO" problem

Search any UK SEO agency's website. A growing number now mention "AI SEO" or "AI search optimisation" in their services. But ask them these questions:

Which search index does ChatGPT use? (Answer: Bing, not Google)
What's the difference between GPTBot and OAI-SearchBot? (Answer: training vs real-time search)
What schema types does Gemini use to verify claims? (Answer: Organisation, Person, sameAs)
What's the crawl-to-referral ratio for Perplexity? (Answer: under 200:1)
How do you monitor AI citations? (Answer: most have no tooling at all)

Most agencies cannot answer a single one of these questions. Yet they've already added "AI SEO" to their services page.

The six blind spots

1. Google-only optimisation

Most SEO agencies optimise exclusively for Google and ignore Bing entirely. Since ChatGPT — which holds 78% of the AI search market and generates 87.4% of all AI referral traffic — indexes exclusively from Bing, these agencies are optimising for the wrong search engine for the dominant AI platform.

ChatGPT uses Bing. Copilot uses Bing. If an agency isn't submitting sitemaps to Bing Webmaster Tools, implementing IndexNow, or checking Bing rankings, they're missing the single largest AI search platform.

2. No understanding of entity SEO

AI platforms don't just read your content — they verify your entity. They cross-reference your Organisation schema with your Google Business Profile, LinkedIn, Wikidata, and other sources. If these don't match, AI platforms have lower confidence in your content.

Most agencies have never:

Implemented Organisation schema with @id and sameAs
Ensured entity consistency across platforms
Created or claimed a Wikidata entry
Connected Person schema for authors to LinkedIn profiles
Used the @graph pattern for compound entity signals

3. No AI citation monitoring

How do you know if ChatGPT mentions your client? How do you track whether Perplexity cites their content? Most agencies have no monitoring capability because the tooling is new and they haven't invested in it.

Without monitoring, agencies are:

Unable to report on AI visibility to clients
Unable to measure the impact of their work
Flying blind on which platforms mention the brand
Missing citation changes — positive or negative

4. Backlinks over brand mentions

Ahrefs' study of 75,000 brands found that brand web mentions (0.664 correlation) and YouTube mentions (0.737) are far stronger drivers of AI citations than backlinks (0.218). Most SEO agencies still prioritise link building over brand building because their entire methodology was built for the pre-AI era.

The correlation data is clear: brand mentions matter 3x more than backlinks for AI visibility. Yet most agencies still sell link building packages as their core offering.

5. No AI-specific schema implementation

Standard SEO agencies implement basic schema — maybe Organisation and Article. But AI-specific schema types are largely unknown to them:

Schema type	AI impact	Typical agency adoption
FAQPage	+350% citation increase	Sometimes
Speakable	+127% citation increase	Almost never
Organisation with sameAs	Entity verification	Rarely complete
Person for authors	3x citation lift	Rarely
@graph compound pattern	Connected entity signal	Almost never

Speakable schema has under 10% adoption but delivers a 127% increase. This is a competitive gap that specialist agencies exploit.

6. No cross-platform testing

When did your agency last test your visibility in Claude? Or Perplexity? Or Gemini's standalone chat? Most agencies only check Google rankings and call it done.

Each platform cites differently:

ChatGPT — Bing index, brand mentions, YouTube
Perplexity — Reddit (46.7%), real-time, anti-promotional
Gemini — Google index, schema verification, entity consistency
Claude — training data, depth, bullet formatting

A strategy that works for ChatGPT may fail completely for Perplexity. Each platform needs its own approach.

What genuine AI search capability looks like

An agency that genuinely delivers AI search optimisation should be able to demonstrate:

Multi-index optimisation — Google AND Bing submission, monitoring, and reporting
Entity SEO expertise — Organisation schema, sameAs, Wikidata, Knowledge Graph
AI citation monitoring — tools that track mentions across ChatGPT, Gemini, Perplexity, Claude
Brand mention strategy — beyond backlinks, focused on web mentions and YouTube
AI-specific schema — FAQPage, Speakable, @graph, Person with sameAs
Cross-platform testing — regular audits across all AI platforms
Technical AI architecture — website structure, robots.txt, llms.txt, IndexNow

This isn't about bashing agencies

AI search is genuinely new. The tooling is immature. The data is still emerging. It's understandable that most agencies haven't caught up yet.

But there's a difference between "we're learning about AI search" and "we already claim to offer AI SEO services." If you're a business hiring an agency for AI search optimisation, ask the questions above. Their answers will tell you everything.

Why 90% of SEO Agencies Aren't Ready for AI Search