the citation gap · 2026 edition

The Discoverability Gap · 2026

Name: The Discoverability Gap · 2026
Creator: Legit.Show
Published: 2026-07-22T17:00:00+00:00

AI answer engines now send the traffic search used to. We checked whether 6639 launched web apps, SaaS and AI tools are even built to be found and cited by them — most aren’t.

33%

ship with no structured data

across 6639 web apps we measured · according to Legit.Show · 2026

The findings

No structured data — 33%
No JSON-LD / schema.org markup — AI answer engines can’t parse what the page *is*, so they can’t cite it as an entity.
No canonical URL — 24%
No canonical link — engines may index a duplicate or query-string variant and split the citation signal.
No OpenGraph card — 24%
No og:image — when an AI or human shares the link it unfurls blank, losing the citation hook.
No sitemap — 21%
No sitemap.xml — crawlers discover pages only by following links, so deep pages get missed.
No meta description — 6%
No meta description — engines fall back to scraping random body text for the answer snippet.

How citable is each page

Citable — 6–7 signals — 71%
Discoverable — 3–5 — 21%
Invisible — 0–2 — 8%

By category

Productivity — 30% no structured data
AI & Agents — 33% no structured data
Business & Finance — 27% no structured data
Developer Tools — 39% no structured data
Lifestyle & Other — 28% no structured data
Other — 49% no structured data

Why this is the new SEO

Answer engines — ChatGPT, Claude, Perplexity, Google AI Overviews — increasingly answer *for* the user instead of sending a list of blue links. To be the source they cite, a page has to be machine-readable: structured data that says what it is, a canonical URL, a clean OpenGraph card, a real meta description. We call building for that AEO (answer-engine optimization) and GEO (generative-engine optimization). This measures how ready the catalog is for it.

What this measures

Public-surface discoverability signals any crawler sees: schema.org structured data, canonical link, OpenGraph image, meta description, and a sitemap — a share of 6639 tested web services as of 2026-08-13. Hygiene for being cited, not a ranking guarantee.

The invisible majority

8% carry two or fewer of the seven signals — effectively invisible to an AI that’s deciding what to cite. 33% ship with no structured data at all, the single signal that tells an engine what the page *is*. It isn’t hard to fix — it’s the unglamorous last 10% a demo never forces you to add.

Sample composition

Not a random sample — this is what we measured. The mix below is the caveat; judge it for yourself.

Productivity: 1630
AI & Agents: 1256
Business & Finance: 928
Developer Tools: 795
Lifestyle & Other: 432
Other: 408
Largest single maker: 0% (modelcontextprotocol, 3)

What we measured (6639)

The full list, so anyone can spot-check. Every item links to its public benchmark.

How this was measured →