Our SEO Agency Said We're "Fully Optimised." So Why Did We Score 34 Out of 100 on the Agentic Readiness Audit?
This is one of the most common questions we encounter. A clinic owner runs the Iris agentic readiness audit expecting a reasonable score — they're paying for SEO, their Google rankings are good, their agency sends monthly reports showing keyword position improvements. The score comes back at 34 out of 100. They call us confused and not a little defensive on behalf of the agency they're paying. The answer isn't that the agency is doing poor work. It's that the agency is doing work for a different system than the one the audit measures.
What Your SEO Agency Is Actually Optimising For
A competent medical SEO agency in 2026 is optimising your website for Google's traditional ranking algorithm. This means keyword research to identify the queries most likely to send patients to your type of clinic. On-page optimisation to ensure your service pages use those keywords in titles, headings, and body copy at appropriate density. Technical SEO to ensure your website loads quickly, is mobile-responsive, passes Core Web Vitals, has no broken links, and is correctly indexed by Googlebot. Content strategy to produce blog posts and FAQ pages that rank for informational queries and build topical authority. Link building to acquire backlinks from authoritative healthcare directories and local websites that signal your clinic's authority to Google's ranking algorithm. GBP management to maintain your local search pack visibility.
All of this is real, valuable work. It drives patients to your clinic through traditional Google search — patients who type a query, see a list of ranked links, and click yours. This patient acquisition channel is not disappearing. Traditional Google search remains significant and will remain so for years. Your SEO investment has genuine return and should not be abandoned.
But none of the work above — keyword density, backlinks, Core Web Vitals, content authority, GBP management — has more than an indirect and partial effect on AI search citation. Google AI Overviews and AI Mode cite based on schema markup and content structure, not keyword rankings. ChatGPT cites based on what PerplexityBot can find and extract from indexed content, not on your domain authority. Voice assistants cite based on Speakable markup and GBP data, not on your page speed score. The deliverables of traditional SEO and the requirements of AI search citation overlap in some areas (well-indexed content is useful for both) but diverge significantly in the specific technical infrastructure layer.
The Five Things Your SEO Agency Is Not Doing
Not building llms.txt. A llms.txt file at your website root is a new standard — it did not exist in any meaningful way before 2024. Traditional SEO practitioners do not build it because it has no effect on Google's traditional ranking algorithm. It does not improve keyword rankings or link equity. But it is the primary mechanism through which AI crawlers navigate to your most important content efficiently. Without it, AI systems must independently discover and prioritise your content — an unreliable process that frequently misses key service pages.
Not implementing AI-specific schema. Most SEO-optimised medical clinic websites have some schema markup — typically Organization or LocalBusiness, sometimes Article schema on blog posts, occasionally BreadcrumbList. What they typically lack is the full medical schema stack: MedicalBusiness with the correct medicalSpecialty property, MedicalProcedure for each offered service, FAQPage on patient Q&A content, Speakable on sections suitable for voice delivery, and Person schema with verified physician credentials. SEO agencies implement schema primarily for rich result eligibility in traditional SERP display. The full schema stack required for AI citation goes significantly beyond what traditional SEO schema implementations cover.
Not configuring AI crawler access. robots.txt management is a traditional SEO task — SEO agencies configure robots.txt to permit Googlebot and Bingbot, manage crawl budget for large sites, and handle specific disallow rules for admin pages. What they do not routinely configure is explicit permissions for AI crawlers: ChatGPT-User, PerplexityBot, ClaudeBot, Google-Extended, OAI-SearchBot. These crawlers either did not exist or were negligible traffic sources when most robots.txt configurations were last updated. An SEO agency managing your robots.txt may have a perfectly valid configuration for traditional search crawling that simultaneously blocks all AI search crawlers.
Not writing content for LLM extraction. SEO content is written for keyword ranking — it uses target keywords at appropriate density, includes semantically related terms, builds to a conclusion that satisfies search intent. AI-readable content is written for extraction — it opens with a direct, standalone answer sentence, structures information in an inverted pyramid so the most important fact appears first, and uses FAQ format wherever natural question-answer structures exist. These are different writing disciplines with different structural requirements. The same piece of content can be well-optimised for SEO and poorly structured for AI citation simultaneously.
Not deploying ai-index.json. ai-index.json is a root-level file that provides AI retrieval systems with a machine-readable entity profile for the clinic — name, services, practitioners, location, hours, usage permissions. It does not exist in traditional SEO methodology because it has no effect on traditional search ranking. It is a GEO-specific deliverable that supplements schema markup and strengthens entity resolution for AI systems cross-referencing multiple data sources.
Why Your SEO Agency Is Not at Fault
This is genuinely important to understand: the gap is not a failure of your SEO agency. It is a limitation of their scope. Traditional SEO is a well-defined discipline with established best practices, reliable measurement, and real return on investment. Your SEO agency is delivering what they were hired to deliver. The AI search layer is a new discipline — GEO — that most SEO agencies have not yet incorporated into their service offering. Those that have are typically in early stages of building the capability. Asking your traditional SEO agency to also deliver GEO is like asking a GBP management specialist to also manage your paid social campaigns — it is adjacent, but it is not the same work, and most specialists are not equipped for both.
The right approach is to maintain your SEO agency relationship and the work they are doing — it continues to have value for traditional organic traffic — and to layer GEO infrastructure on top of the foundation they have built. The SEO work actually improves the baseline of the GEO layer: well-indexed content is more likely to be retrieved by AI systems than non-indexed content, and strong domain authority provides some signal of trustworthiness to AI systems. The two layers are complementary, not competing.
If you want to understand exactly which GEO infrastructure components are missing from what your SEO agency has built, the Iris agentic readiness audit gives you a scored breakdown across all six AI visibility dimensions in 60 seconds. The score of 34 out of 100 is not a verdict on your SEO. It is a map of the AI infrastructure that has not been built yet — and that map is exactly what a GEO engagement resolves.