Why a capability matrix is useful right now

Job titles in the AI SEO category drifted faster than the underlying skills. The number of LinkedIn profiles using “AI SEO expert” or “GEO specialist” in their headline rose roughly six-fold between January 2024 and January 2026 in the UK alone, based on our search data. The number of practitioners who can actually demonstrate the underlying skills did not rise at the same rate. That asymmetry creates a hiring and procurement problem: buyers cannot tell, from a profile or a 30-minute screening call, whether a candidate owns the capabilities they advertise.

A capability matrix solves the asymmetry by listing the discrete skills the role actually needs and assessing each against an evidence bar. Buyers get a defensible filter; marketing leaders get a roadmap for upskilling existing SEO staff into the GEO category; candidates get a clear view of what good looks like.

Methodology in one paragraph

We identified 100 UK-based LinkedIn profiles using “AI SEO expert”, “GEO specialist”, “generative search consultant” or equivalent in their headline as of February 2026. Each respondent was approached for a paid 20-minute interview structured around the 8 capabilities. Each capability was marked pass or fail on a defined evidence bar. Self-reporting did not count: schema implementation required describing or showing a real JSON-LD block on a named client page; citation tracking required walking through a live dashboard or named methodology; entity strategy required a worked example of how the respondent would mark up a fictional UK firm. We logged binary scores, time-in-role and current employer (agency, in-house, freelance). The full anonymised dataset is available on request.

The 8 capabilities

The eight capabilities cover the discrete skills an AI SEO expert needs to deliver in a 2026 engagement. They are not exhaustive but they are the eight we use ourselves when hiring or assessing partners.

Entity strategy: ability to describe how a brand and its products are represented as named entities in body content and structured data, and how to strengthen weak entity signals. Schema implementation: hands-on knowledge of FAQPage, Article, Organization and Product schema, plus a monthly validation routine. Citation tracking: live or recently-live experience of measuring which prompts on which engines surfaced which pages, with prompt-level granularity. Technical SEO: classic crawlability, indexability, page speed and Core Web Vitals, which remain prerequisites even in GEO. Content production: ability to write or commission a 2,000-word original-research piece, not just optimise existing copy. Vertical knowledge: meaningful command of at least one industry’s vocabulary, regulators and competitive landscape. Measurement and reporting: design of monthly reports that tie AI visibility to pipeline rather than vanity metrics. Ethical practice: awareness of ICO data-protection rules, knowledge of which optimisation tactics are penalisable, and a stated practice on AI-generated content disclosure.

The distribution: what AI SEO expertise actually looks like in 2026

The distribution is sharply asymmetric. The top decile of the 100-respondent sample averaged 6.4 capabilities. The bottom decile averaged 1.8. The median respondent demonstrated 3.7 of 8 capabilities, meaning the typical self-described AI SEO expert in early 2026 owns fewer than half the skills the category requires.

Two observations matter for buyers. First, the long left tail (24 respondents scoring 2 or fewer capabilities) represents practitioners who have rebadged from classic SEO or content marketing without acquiring the new skills. Second, the right tail is thin: only 10 respondents scored 6 or more capabilities, which is roughly the bar we would treat as a genuine AI SEO expert. That scarcity is itself useful information for compensation negotiation.

Which capabilities are common and which are rare

Schema implementation was the most common, present in 71 of 100 respondents. But only 32 of those 71 could describe a monthly validation routine, which we treat as the practical bar. Passive schema knowledge is widespread; operational schema practice is not. Citation tracking was the rarest at 23 of 100, despite being the most discriminating skill in our parallel agency assessment. Entity strategy followed at 27 of 100. Vertical knowledge varied: 84 of 100 claimed at least one vertical but only 41 could pass the evidence bar in 20 minutes.

The rarest capabilities are also the most predictive of results. Citation tracking and entity strategy together are present in only 11 of 100 respondents; those 11 also account for half the top decile. A buyer optimising for results rather than for hire speed should weight these two capabilities heavily.

The pattern by tenure and employer type

The capability score rises with tenure inside the AI SEO category specifically, not with total SEO tenure. Respondents with 18-plus months in an explicitly AI SEO or GEO role averaged 5.6 capabilities. Respondents with under 6 months averaged 2.3, regardless of how long they had been in classic SEO before. The implication is that classic SEO experience does not directly translate to the AI SEO role; the new skills have to be built deliberately. In-house respondents averaged 4.1 capabilities and agency respondents 3.4, with freelancers at 3.6, which is a smaller spread than industry stereotypes would predict.

What buyers should ask in interviews

Three interview techniques surfaced reliable evidence in our survey and translate to standard hiring practice. First, request a 5-minute walk-through of a live citation-tracking setup, named client or sandbox, prompt-level data visible. This is the single highest-discrimination check. Second, ask the candidate to mark up a fictional UK firm with FAQPage, Article and Organization schema on the call. The pace and accuracy of the response separates operational practitioners from theorists. Third, ask for one example of a measurement they would not include in a monthly client report and why. The answer reveals whether the candidate distinguishes vanity metrics from pipeline-relevant signal.

Limitations

Our sample is LinkedIn-based and skews towards practitioners with public presence. Genuinely senior AI SEO experts often have less public footprint because their work is in-house or under NDA, and may be under-represented. The 20-minute structured interview is a snapshot; a longer assessment would reveal capabilities that the format compresses. The 8 capabilities reflect 2026 market needs; the matrix will need refresh as the discipline matures. We will re-run the survey in February 2027 with a wider sample and track which capabilities have shifted in prevalence.

Frequently asked questions

What are the 8 capabilities of an AI SEO expert in 2026?

Entity strategy, schema implementation, citation tracking, technical SEO, content production, vertical knowledge, measurement and reporting, and ethical practice. The first three are AI SEO-specific and rarely transferable directly from classic SEO. The latter five overlap with strong SEO practice but require sharper edges to deliver in a GEO context. Citation tracking and entity strategy are the most discriminating: present in only 23% and 27% of self-described AI SEO experts in our 100-respondent sample, but in 80% and 70% respectively of the top decile.

How does the median UK AI SEO expert score?

3.7 of 8 capabilities, in our February-April 2026 LinkedIn survey. The bottom decile scored 1.8 and the top decile scored 6.4. The median respondent owns fewer than half the skills the role requires, which means buyers cannot rely on the job-title shorthand. A structured assessment against the 8 capabilities surfaces meaningful variation; a 30-minute screening call does not. Freelancers and agency staff scored similarly (3.6 and 3.4); in-house averaged slightly higher at 4.1.

Which capability is rarest?

Citation tracking, present in only 23 of 100 respondents. This is also the most discriminating skill: it correlates with seniority, tenure in the AI SEO category, and overall capability score. Citation tracking requires hands-on experience with a live dashboard (Profound, Authoritas, or an internally-built tool), prompt-level data and at least 90 days of historical tracking against named prompts. The 23 respondents who passed this bar dominated the top decile.

Does classic SEO experience translate to AI SEO?

Partially. Technical SEO, content production and measurement transfer cleanly. Entity strategy, schema operationalisation and citation tracking do not transfer directly and have to be built deliberately. Respondents with 18-plus months specifically in an AI SEO or GEO role averaged 5.6 capabilities; respondents with under 6 months averaged 2.3 regardless of how many years they had spent in classic SEO. Tenure inside the new category is the predictor, not total SEO tenure.

How should I interview an AI SEO candidate using this matrix?

Three live tests give the most signal in a single hour. First, ask for a 5-minute walk-through of a live citation-tracking setup, named client or sandbox, prompt-level data visible. Second, ask the candidate to mark up a fictional UK firm with FAQPage, Article and Organization schema on the call, watching pace and accuracy. Third, ask for one example of a measurement they would deliberately exclude from a monthly client report and why. Together these three checks surface 5 of the 8 capabilities directly and indirectly probe the other 3.

What should I pay a top-decile AI SEO expert in the UK in 2026?

Compensation data from our network and Glassdoor 2025-2026 UK SEO salary data suggests a top-decile AI SEO expert commands £75k-£100k base in-house, or a daily rate of £700-£950 freelance, with another 20-30% premium for vertical depth in a regulated sector. The premium reflects scarcity: only 10 of our 100 respondents cleared the 6-capability bar. Buyers who anchor on classic SEO compensation benchmarks will systematically underpay top AI SEO talent and lose the offer to firms that have updated their pay bands.

Will the 8 capabilities change in 2027?

Yes. We expect citation tracking to become more widely held as Profound, Authoritas and similar tools mature and become standard. Two new capabilities are likely candidates for 2027: causal attribution (proving which prompts caused which sales pipeline) and cross-engine arbitrage (managing visibility across ChatGPT, Perplexity, Gemini and Bing Chat as separate surfaces with separate optimisation patterns). We will publish a refreshed 2027 matrix in February next year.

By Beata Nowak, Strategy Lead, AiBoost | Published 21 May 2026 | 10 min read