State of GEO: July 2026

Score distribution: 75% still below the Good threshold

The average GEO score across 360 unique domains is 54.1 out of 100 — up 0.5 points from June's 53.6. The median is 55, with the bottom quarter of sites scoring at or below 43 and the top quarter at or above 67. The highest score in the dataset is 100 — the first perfect 100 in the benchmark's history, up from a maximum of 90 in June. The lowest is 6.

The band breakdown shows a meaningful shift toward higher tiers compared to June:

The practical reading: three in four audited sites still cannot be described, cited, or recommended by an AI engine with confidence. But the direction of travel is positive. The Good band grew from 19.4% to 22.8% (+3.4 percentage points), Excellent nearly doubled from 4 to 7 domains, and the share of sites below Good dropped from 79.2% to 75.3% (-3.9 pp). The Foundation band shrank from 64.6% to 61.1%, suggesting some sites are climbing out of the middle tier into Good.

Category breakdown: schema is gaining, llms.txt is slipping

The GEO rubric scores eight categories. Each has a different maximum. The table below shows the July 2026 average for each, what the maximum is, and the efficiency percentage — how much of the available points the average site is capturing.

Three categories form a strong foundation. Meta tags lead at 88.6% efficiency — unchanged from June, as title tags, descriptions, and canonical URLs are well understood and broadly implemented. Content quality follows at 80.0%, also flat, driven by the 78.6% of sites that lead pages with a direct, parseable answer (up from 77.1%). Robots and crawler access improved slightly to 76.1% (from 75.6%): most sites allow the major AI bots, with an average of 23.2 bots explicitly permitted.

The middle tier saw the most notable month-over-month movement. Schema markup improved from 38.1% to 42.5% efficiency — a 4.4 percentage-point gain, the largest of any category. This aligns with the adoption data: overall schema adoption rose from 70.1% to 75.6%, WebSite schema from 52.4% to 58.9%, and FAQ schema from 13.2% to 18.1%. Conversely, LLMs.txt efficiency dropped from 38.9% to 36.1% — a 2.8 percentage-point decline, reflecting the adoption regression from 58.3% to 54.2%. Brand & entity remained flat at 38.0%.

The outlier remains AI Discovery at just 10.0% — an average of 0.6 points out of 6. This is up from 8.3% in June, a modest improvement, but AI discovery files are still a concept that almost no one has acted on.

Adoption rates: the machine-readability stack

The GEO rubric checks specific technical signals beyond category scores. These adoption rates show where the default behavior of the web sits in July 2026 — and what changed from June.

Two stories emerge from the month-over-month comparison. Schema is gaining: every schema type we track improved — overall adoption +5.5pp, WebSite +6.5pp, FAQ +4.9pp, Organization +1.2pp. Teams are investing in structured data, and the category efficiency rose from 38.1% to 42.5%. llms.txt is slipping: adoption dropped 4.1pp and full-file adoption dropped 0.9pp. The 72 new domains added in July are the likely driver — they bring traditional SEO signals (schema, canonical, meta tags) but not AI-specific signals (llms.txt, AI discovery endpoints). This is the same skills-transfer gap identified in June, now visible in the data at scale.

Month-over-month: what changed between June and July

This is the first edition with a prior month to compare against. The table below shows the key metrics side by side, with the direction and magnitude of change.

The headline: the cohort grew 25% and the average did not decline. That is a positive signal — it suggests the new domains entering the benchmark are not significantly worse than the existing cohort, which in turn suggests AI readiness awareness is spreading beyond the earliest adopters. However, the median dropped from 57 to 55, indicating that the new entrants cluster slightly below the existing median.

The divergence between schema (up) and llms.txt (down) is the most actionable finding. Teams new to AI readiness are adopting structured data — a mature practice they already understand from traditional SEO — but are not yet adopting llms.txt and AI discovery files, which require a different vocabulary and deliberate action. The skills-transfer gap identified in June is now quantifiable in the month-over-month data.

The biggest gap remains: AI Discovery at 10.0% efficiency

Every category in the GEO rubric has a gap. But AI Discovery is still in a different tier. The average score is 0.6 points out of a maximum of 6 — a 10.0% capture rate, up from 8.3% in June. Only 17.5% of audited sites expose any AI discovery endpoint, up from 16.0%.

AI discovery files — like /.well-known/ai.txt and the signals inside a well-structured llms.txt — are designed to make a site explicitly machine-navigable. They go beyond allowing crawlers: they describe the site's purpose, its most important content, and the context an AI needs to answer questions about it with specificity rather than generality.

The gap exists because these standards are new and publishing them requires deliberate action — nothing in a default CMS install or hosting setup creates them automatically. That makes it one of the rarest signals in July 2026, and one of the easiest to capture simply by doing what almost nobody has done yet. The modest improvement from 8.3% to 10.0% efficiency means the category is moving, but at this rate it will take many months before AI discovery becomes mainstream.

By TLD: who is leading AI search readiness in July 2026

The TLD landscape shifted in July. The dataset diversified — .com remains dominant at 199 domains (up from 163), but new extensions entered with strong scores. .fr debuted at the top of the table with a 64.2 average and 100% schema adoption, and .hk followed at 63.3 with 100% schema adoption as well. Italian .it domains improved dramatically from June.

.fr domains debuted strong: 6 domains with a 64.2 average score, 100% schema adoption, and 66.7% llms.txt adoption. This is the highest average of any TLD with more than 3 domains. French web teams appear to be investing heavily in both structured data and AI-specific signals simultaneously — a pattern that stands in contrast to the .it and .org cohorts where schema adoption outpaces llms.txt.

.it improved dramatically month-over-month: average score rose from 54.3 to 61.4 (+7.1 pts), schema adoption from 81% to 100% (+19 pp), and llms.txt from 33.3% to 50.0% (+16.7 pp). While the .it cohort shrank from 21 to 8 domains — a sampling artifact, not a trend — the domains that remain are significantly better prepared. Italian sites now have the highest schema adoption rate of any TLD in the dataset alongside .fr and .hk.

.cn showed the largest absolute score improvement: from 23.2 to 35.6 (+12.4 pts). While .cn remains the lowest-scoring TLD, the gap is narrowing. Schema adoption rose from 22.2% to 40.0% and llms.txt from 33.3% to 60.0% — a notable jump that suggests Chinese domains in this cohort are beginning to adopt AI signals, even if they still lag on overall readiness.

.ai domains remain strong but no longer lead on score: at 57.8 average (down from 59.7 in June, though the cohort grew from 10 to 17 domains), .ai is now surpassed by .fr (64.2), .it (61.4), .hk (63.3), and .co (59.3). However, .ai still has the second-highest llms.txt adoption at 82.4%, behind only .co at 100%. The AI-native extension's lead on llms.txt persists even as the cohort diversifies.

.org sites improved their schema adoption from 77.8% to 91.7% — now the third-highest in the dataset — but llms.txt adoption remained low at 41.7% (down from 44.4%). Non-profit and institutional organizations continue to have strong technical foundations but are slower to adopt emerging AI signals. .de domains, new to the dataset, follow a similar pattern: 80.0% schema adoption but only 30.0% llms.txt.

What July 2026 tells us about AI search readiness

The June-to-July comparison reveals two things at once. First, the floor is rising slowly: the share of sites below Good dropped from 79.2% to 75.3%, and the Good band grew from 19.4% to 22.8%. The first perfect 100-point score appeared. Seven domains now sit in the Excellent band. These are real gains, even if the average moved only 0.5 points.

Second, the skills-transfer gap is now visible in the data. The 72 new domains added in July brought traditional SEO signals with them — schema adoption jumped 5.5 percentage points, the largest single gain — but they did not bring AI-specific signals. llms.txt adoption dropped 4.1 percentage points. AI discovery improved only 1.5 percentage points. The pattern is clear: teams understand structured data (a mature practice) but do not yet understand llms.txt and AI discovery files (a new practice). The vocabulary gap between traditional SEO and AI readiness is the bottleneck.

The correlation data reinforces this. Sites with schema markup score 28.2 points higher than sites without (61.0 vs 32.8) — up from 27.3 points in June. Sites with llms.txt score 20.0 points higher (63.3 vs 43.3) — up from 19.8 points. The correlations are strengthening, meaning the signals that distinguish high-scoring sites from low-scoring sites are becoming more pronounced, not less. The gap between the prepared and the unprepared is widening.

The good news, unchanged from June: nothing in the gap is architecturally difficult. The actions that move a site from Foundation to Good — publish a complete llms.txt, add the right schema types, expose AI discovery files — do not require a redesign, a backend migration, or a content overhaul. They are publishing decisions. They require knowing what to publish, not building anything new. The July data shows that the teams that make these decisions are pulling ahead of those that do not.

The July 2026 quick-win checklist

Based on where audited sites lose the most points in July 2026 — and the month-over-month trends — these are the highest-leverage actions, ordered by expected point recovery per hour of work:

1. Publish a full llms.txt (not a stub). llms.txt adoption slipped to 54.2% in July, and only 26.9% have a full, structured file. If you are one of the 72 new domains that entered the cohort without one, this is the single fastest way to separate yourself from the pack. Include at minimum: a brief site description, sections for your most important pages, and URLs to your docs, blog, and product pages. Use the free llms.txt generator to build a starter in under 5 minutes.
2. Add Organization and WebSite JSON-LD. Only 52.2% of sites have Organization schema and 58.9% have WebSite schema — despite WebSite schema seeing the largest adoption gain this month (+6.5pp). Include name, url, logo, sameAs, and description at minimum. Sites with schema score 28.2 points higher than those without.
3. Publish a /.well-known/ai.txt. Only 17.5% of sites do this — up from 16.0% in June, but still remarkably rare. It is a single file with a handful of lines that places a site in the top 17.5% of AI discovery readiness immediately. AI Discovery remains at 10.0% efficiency — the worst-performing category by a factor of 3.7×.
4. Add FAQ schema where FAQ content already exists. FAQ schema grew from 13.2% to 18.1% in July, and FAQ content on page rose to match at 18.1% — but 81.9% of sites still have no FAQ schema at all. Every page with a Q&A section that lacks FAQ JSON-LD is a missed citation opportunity. This is the schema type with the most headroom.
5. Audit your robots.txt for accidental AI bot blocks. The average site allows 23.2 AI bots — down marginally from 23.3 in June. Confirm that GPTBot, ClaudeBot, PerplexityBot, and GoogleOther are either explicitly allowed or not listed (default allow).
6. Re-audit after each change. The GEO rubric rewards compound improvements: each signal reinforces the others. The July data shows the correlation gaps are widening — the sites that publish llms.txt + Organization schema + AI discovery files are pulling further ahead of those that do not. A site with all three is not just the sum of three parts — it is an entity that AI models can triangulate across multiple signals.

Methodology

This report covers 360 unique domains audited by GeoReady between 1 July and 31 July 2026. When a domain was audited multiple times in the period, only the most recent audit is included — this prevents repeat users from inflating the cohort size and ensures each domain appears once at its best-known state.

• Engine version: GEO Optimizer 4.12.x, the open-source 100-point, 8-category rubric. Source available on GitHub.
• Audited sample: Sites that chose to audit with GeoReady — not a random sample of the web. The cohort grew 25% from June (288 → 360 domains), and the new entrants skew toward teams already interested in AI visibility, so these figures likely overestimate average readiness across all websites.
• Anonymized: Domain names are stored as salted HMAC-SHA256 hashes in the benchmark dataset. No individual domain is identifiable in the public figures.
• Monitoring audits excluded: Only user-initiated audits (web, API, CLI) are included. Scheduled monitoring re-checks are excluded to avoid inflating the cohort with the same domains audited repeatedly.
• Consistent rubric: Every domain is scored with the same engine and the same weights. Category maxima: Robots 18, LLMs 18, Schema 16, Meta 14, Content 12, Brand & Entity 10, Signals 6, AI Discovery 6.
• Month-over-month comparison: June 2026 figures are drawn from the same benchmark dataset using the same methodology, covering 288 domains audited between 10 June and 30 June 2026. The two cohorts are independent — domains audited in both months appear once in each report at their most recent state for that period.

The underlying benchmark dataset is the benchmark_audit_events table in the GeoReady production database. The public API endpoint (GET /api/public/benchmark) returns aggregated figures only — no per-domain data is ever exposed.

Frequently asked questions