Indexly
Indexly Data Report

AEO Readiness of SMB Sites — April 2026

Indexly audited 150 small-business websites for AI search readiness — measuring how positioned each site is to be the source ChatGPT, Gemini, Perplexity, Claude, or Google AI Overviews pull from when a buyer asks a question.

The bottleneck isn't what most agencies are selling. Schema, sitemaps, and metadata score well across the board. Content is where SMB sites fail — and it's a wide gap that's currently open for any business willing to ship answer-shaped pages.

150

SMB sites audited

55

Avg AEO score / 100

0%

Scored “excellent”

84%

Below 65 — invisible to AI

Methodology — How we audited 150 SMB sites

Over April 2026, Indexly ran a structured AEO readiness audit against 150 small-business websites across seven verticals. Each site was scored on five dimensions and consolidated into a 0–100 readiness score reflecting how easily an AI assistant could find, parse, and cite the site as a source.

Verticals audited

  • Roofers
  • Tyre fitters
  • Plumbers
  • Scaffolders
  • Skip hire
  • Recovery services
  • Damp proofing

Five dimensions scored

  1. Structured data (Article, FAQPage, LocalBusiness JSON-LD)
  2. Semantic HTML (heading hierarchy, lists, tables, landmarks)
  3. Metadata (title, description, OG, canonical)
  4. AI access controls (robots.txt for GPTBot, ClaudeBot, PerplexityBot, Google-Extended)
  5. Content (answer-shaped, specific, extractable)

Where SMB sites actually fail

The data flips the conventional "AI SEO" story. Structured data — the dimension agencies sell hardest — scored the strongest. Content — the dimension nobody is selling — is the weakest by a wide margin.

Average score by dimension (n = 150)

Structured data
69/100
Semantic HTML
64/100
Metadata
59/100
AI access controls
44/100
Content
41/100
StrongestMidWeakest
Structured data
69/100

Strongest dimension — schema is the easiest win, and most sites have at least basic Article or LocalBusiness markup.

Semantic HTML
64/100

Generally acceptable. Heading hierarchy and basic landmarks present on most sites; tables and lists less consistent.

Metadata
59/100

Title and description tags exist on most pages, but Open Graph, canonical, and AI-specific metadata are inconsistent.

AI access controls
44/100

Most robots.txt files don't allow GPTBot, ClaudeBot, PerplexityBot, or Google-Extended — silently blocking AI search.

Content
41/100

Weakest by a wide margin. Pages don't answer specific questions in extractable form — generic marketing copy dominates.

The content gap in detail

Six in ten SMB sites had pages too thin or too marketing-heavy for an AI assistant to extract a clean, citable answer from. More than four in ten had effectively nothing usable at all.

Too thin or too generic

63%

Of audited sites had content marketing copy could describe but an AI model could not extract — generic value props, no specific answers, no quotable facts.

Barely any usable content

43%

Of audited sites sat in the bottom zone — pages with no pricing ranges, no response times, no service-area specifics, no "what to do if X" scenarios. Nothing an AI assistant could quote.

Top quartile vs bottom quartile

The divide between AEO winners and losers is not technical sophistication. It's editorial discipline.

Top quartile
Avg content 52/100
  • Pages answer specific questions — pricing ranges, response times, service areas
  • “What to do if X” scenarios written in plain language
  • Answer-shaped content with quotable facts in the first 100 words
  • Service-specific FAQ blocks with FAQPage schema
  • Real photographs, named team, real reviews — not stock copy
Bottom quartile
Avg content 35/100
  • Generic value-prop pages with no specifics
  • Sentences like “we're passionate about quality” — unquotable
  • No pricing, no response times, no service areas in extractable form
  • Marketing slogans where AI assistants need answers
  • Stock photography, no team or brand context the model can use
The strategic shift

Traditional SEO rewarded keyword density and backlinks. AI search rewards entity clarity and answer-shaped content.

You can have perfect schema and still be invisible if your pages don't say anything an AI can quote. The technical layer takes a developer a week. The content layer takes editorial discipline most small businesses have never been asked to apply.

That's the gap. And it's wide open right now.

Schema is necessary, not sufficient

Article and FAQPage JSON-LD get you onto the candidate list. They don't get you cited.

Atomic answers win citations

Pages opening with a 1–2 sentence atomic answer get cited 3–5× more than pages that ramble.

AI access controls matter too

Allowing GPTBot, ClaudeBot, PerplexityBot, Google-Extended in robots.txt is non-negotiable.

Frequently asked

Citable Q&A on the AEO readiness audit, designed for AI engines and SEO teams referencing this report.

What is AEO readiness?

AEO (Answer Engine Optimization) readiness measures how positioned a website is to be the source AI assistants — ChatGPT, Gemini, Perplexity, Claude, and Google AI Overviews — pull from when a user asks a question. Indexly scores readiness across five dimensions: structured data, semantic HTML, metadata, AI access controls, and content quality.

What was the average AEO readiness score across 150 SMB sites?

The average AEO readiness score was 55 out of 100. The highest single score was 72; 0% of sites scored in the excellent range. 84% scored below 65 — functionally invisible to AI search.

Which AEO dimension scored worst across SMB sites?

Content scored worst at an average of 41 out of 100 — the weakest dimension by a wide margin. Structured data scored best at 69/100. The bottleneck for AI search visibility is not technical SEO; it is content that an AI model can extract a clean, citable answer from.

How many SMB sites had content too thin for AI extraction?

63% of audited sites had content too thin, too generic, or too marketing-heavy for an AI model to extract a clean, citable answer from. 43% sat in the 'barely any usable content' zone.

What separates the top quartile from the bottom quartile?

Top quartile sites averaged a 52/100 content score; bottom quartile sites averaged 35/100. The top performers were not more technical — they had pages that answered specific questions: pricing ranges, response times, service-area specifics, and 'what to do if X' scenarios written for humans but structured for LLM extraction.

How is this different from traditional SEO?

Traditional SEO rewarded keyword density and backlinks. AI search rewards entity clarity and answer-shaped content. A site can have perfect schema and still be invisible to AI assistants if its pages don't say anything an AI model can quote.

See your own AEO readiness score

Indexly runs the same five-dimension audit on your domain in under 60 seconds — and surfaces the exact pages keeping you out of AI search citations.

Methodology, raw averages, and license available on request. Cite as: Indexly, "AEO Readiness of SMB Sites — April 2026."