Foglift
AEO Readiness Across 311 Websites: The Median Site Scores 46/100
Foglift scanned 1,386 pages across 344 distinct domains between 2026-03-14 and 2026-05-23. On the 311 domains where we computed a full AEO score, the median is 46 out of 100. The median SEO score on the same population is 86. The 40-point gap is the gap between being ready for Google and being ready for an AI engine, and most sites are sitting inside it.
Last updated Data window 2026-03-14 to 2026-05-23 (1,386 scans across 344 domains; 311 with a full AEO score)
Methodology
Foglift's public scanner audits a page against eight AEO dimensions (Structured Data Richness, Heading Clarity, FAQ Quality, Entity Identity, Content Depth, Citation Formatting, Topical Authority, AI Crawler Access) plus parallel SEO and GEO subscores. Every public and authenticated scan since 2026-03-14 is stored in the public.scans table. For this artifact we deduplicate to one row per registered domain by taking each domain's most recent scan, then filter to rows that produced a numeric AEO score. That leaves 311 distinct domains in the sample. Per-domain results are not enumerated. The aggregation SQL is published verbatim alongside this report.
The readiness gap is 40 points wide
Two numbers tell the entire story. The median SEO score across the 311 scanned domains is 86 out of 100. The median AEO score on the same domains is 46 out of 100. The same population, the same scans, the same Saturday morning. A 40-point gap between how ready the typical site is for a traditional search engine and how ready it is for an answer engine.
The SEO discipline has had 20 years to bake itself into the defaults of every CMS, every static site generator, every theme marketplace. Meta descriptions, canonical tags, sitemap files, heading hierarchies are all rituals a publisher gets for free at this point. AEO does not yet get any of that for free. A site that ships an excellent SEO baseline can still leave most of the AI-extraction surface area on the table. The data below is the size of that gap, broken into the pieces a publisher can actually act on.
Headline stats (n = 311 distinct domains)
46/100
Median AEO score
86/100
Median SEO score
85/100
Median GEO score
43.8/100
Mean AEO score
67/100
P90 AEO score
85/100
Max AEO score observed
The score distribution: zero sites above 85
Of the 311 domains with a numeric AEO score, exactly two cleared 80. The 90th percentile sits at 67. The 95th sits at 73. The 99th sits at 77. The single highest observed score is 85. There is no site in the sample that we would call AEO- complete by the obvious definition of clearing 90.
The shape of the distribution is the interesting part. The fattest single band is 50 to 59 (82 domains, 26.4% of the sample), and the modal site is closer to a solid B-minus than to either tail. But the bottom of the distribution is heavy: 67 domains (21.5%) score below 30, and 18 score in single digits. These are not abandoned sites; many of them ship perfectly functional homepages with clean copy. They are simply running on the SEO playbook unchanged, which currently is not enough.
| AEO band | Domains | % of sample | At or above |
|---|---|---|---|
| 80–100 | 2 | 0.6% | 2 (0.6%) |
| 70–79 | 22 | 7.1% | 24 (7.7%) |
| 60–69 | 28 | 9.0% | 52 (16.7%) |
| 50–59 | 82 | 26.4% | 134 (43.1%) |
| 40–49 | 73 | 23.5% | 207 (66.6%) |
| 30–39 | 37 | 11.9% | 244 (78.5%) |
| 20–29 | 25 | 8.0% | 269 (86.5%) |
| 10–19 | 24 | 7.7% | 293 (94.2%) |
| 0–9 | 18 | 5.8% | 311 (100.0%) |
The SEO-ready, AEO-broken population: 44.5% of the strong-SEO cohort
Restrict the sample to sites whose SEO score is 70 or higher. That is 236 of the 311 domains, roughly three quarters of the population. Now ask how many of those SEO-strong sites also score below 50 on AEO. 105 of 236, or 44.5%. Almost half of the sites that pass a traditional SEO audit are still failing the AI-extraction layer that runs on top of it.
The correlation between SEO and AEO is 0.71, which is real but loose. The correlation between GEO and AEO is 0.88, which is much tighter. That is the right number to anchor on. If your stack treats AEO as a function of SEO, you are extrapolating off a moderate correlation. If your stack treats AEO as a function of GEO (the answer-quality layer: entity identity, structured data, citations, content depth), you are extrapolating off a much stronger signal. The 44.5% gap is the cost of choosing the first model over the second.
The most common missing signals
The Foglift scanner emits a per-issue diagnostic on every scan. We collapsed the 311 latest-per-domain scans into a flat issue list and ranked by how many distinct domains carried each issue. The top of the list is below. Two cosmetic items dominate (Open Graph and Twitter Card tags), but the structurally meaningful gap is structured data: 29.6% of scanned domains ship no JSON-LD at all. That single missing signal is the largest contributor to the AEO median sitting where it sits.
| Missing signal | Layer | Sites missing | % of sample |
|---|---|---|---|
| Missing Open Graph tags | SEO | 125 | 40.2% |
| Few internal links | SEO | 119 | 38.3% |
| Missing Twitter Card tags | SEO | 106 | 34.1% |
| Title tag too long | SEO | 106 | 34.1% |
| No structured data (JSON-LD) | SEO/GEO | 92 | 29.6% |
| Missing H1 heading | SEO | 76 | 24.4% |
| Missing meta description | SEO | 71 | 22.8% |
| Meta description too long | SEO | 68 | 21.9% |
| No sitemap.xml found | SEO | 67 | 21.5% |
| Missing canonical URL | SEO | 59 | 19.0% |
| No H2 subheadings | SEO | 58 | 18.6% |
| No FAQ section | GEO | 55 | 17.7% |
| Multiple H1 headings found | SEO | 42 | 13.5% |
| No favicon detected | SEO | 41 | 13.2% |
| No entity identity markup | GEO | 25 | 8.0% |
Top decile vs. bottom quartile: a 4.4x issue-load gap
The cleanest way to see what separates the readiness leaders from the laggards is per-site issue count. We bucketed the 311 domains into three bands: a top decile (AEO ≥ 67, n = 33), a middle (AEO 30–66, n = 211), and a bottom quartile (AEO < 30, n = 67). For each domain we counted GEO and SEO issues flagged by the scanner, then averaged inside the band.
| Band | Domains | Avg. GEO+SEO issues per site |
|---|---|---|
| top_decile (AEO ≥ 67) | 33 | 2 |
| middle (AEO 30–66) | 211 | 3.2 |
| bottom_quartile (AEO < 30) | 67 | 8.8 |
A typical top-decile site is carrying 2.0 flagged issues. A typical bottom-quartile site is carrying 8.8. The gap is 4.4x. There is no magic dimension that the leaders are running ahead on; they are just carrying fewer of every issue at once. That is consistent with how the AEO score itself composes: it is the sum of many small signals, and the way to push it up is to close many small gaps rather than win on any one.
What this means for a publisher
Three implications if you operate a site you want AI engines to cite.
One. Treat AEO as a separate discipline from SEO. The 0.71 correlation between the two is loose enough that an SEO investment no longer guarantees an AEO outcome. Build a parallel measurement cadence (scan, score, fix, re-scan) on the AEO surface specifically.
Two. Ship Schema.org JSON-LD. If your site is in the 29.6% of the sample with zero structured data, that single fix is worth more median points than any other deliverable in the engineering backlog. Start with Organization, layer in FAQPage, then the most-fitting page-type schema (Product, Article, HowTo).
Three. Aim for the top decile, not the median. The leaders carry 2.0 issues per site and the middle carries 3.2. The difference is roughly one more closed gap. The leaders are not ahead because they did one big thing right; they are ahead because they closed a few more small things. Plan in that shape.
Other reports in this series
This is Day 6 of an ongoing research cadence. The first five reports all draw from the Q2 2026 AI Search Citation Benchmark, which measures the engine side of the same problem: which domains ChatGPT, Claude, Gemini, Google AI Overview, and Perplexity actually cite when a buyer asks a question.
- Top 100 most-cited domains in AI search (Q2 2026). The canonical AI authority list across five engines.
- ChatGPT vs. Google AI Overview: the same prompt, two different webs. 4.1% Jaccard overlap across 75 prompts.
- Buyer intent reshapes AI citations. Discovery, shortlist, and variation cite different domain sets.
- When AI engines cite the reviewer vs. the brand. The 70-point vendor-vs-aggregator gap.
- AI Search Citation Benchmark, Q2 2026. The underlying dataset for the first five reports.
Frequently Asked Questions
Why is the median AEO score so much lower than the median SEO score?
Traditional SEO grades a page on signals that have been industry-standard for 20 years: meta description, heading structure, canonical, sitemap, internal linking. Most CMS templates ship those by default. AEO grades a page on signals that AI engines extract: rich JSON-LD structured data (Organization, Product, FAQPage, HowTo), explicit entity identity, FAQ blocks, citation formatting, content depth tied to authoritative references. Those are not yet template defaults. The 40-point median gap is the cost of that lag.
Is the sample biased toward sites that already care about AI search?
Yes, and the bias runs in the direction of overstating readiness, not understating it. Most of the 311 domains in this sample arrived by deliberately running a Foglift scan on their own site, which selects for operators who already know AEO matters. Random-sample baselines on the open web are very likely to be lower than 46. Treat this number as an upper bound on the typical site, not a lower bound.
If 44.5% of SEO-ready sites still score below 50 on AEO, what's the single highest-leverage fix?
Schema.org JSON-LD. About 30% of the scanned population ships no structured data at all, and Foglift's per-issue weighting puts structured data among the single largest contributors to the AEO score. A page that already has clean SEO fundamentals but no JSON-LD will typically gain 10–20 AEO points from adding Organization plus FAQPage plus the most-fitting page-type schema (Product, Article, HowTo). That alone moves a median site from the 40s into the 60s.
Why did 33 of the 344 scanned domains not get an AEO score?
Three reasons, in descending frequency: the scan returned a non-200 HTTP status (404, 403, 5xx), the rendered HTML did not contain enough content to compute a meaningful score, or the scan predated a server-side schema field added to the scans table in late March 2026. Those rows are excluded from this artifact. The 311 number is the AEO-scored sample; 344 is the population of scanned domains and 1,386 is the count of individual scan events.
How does this connect to the Q2 2026 Citation Benchmark?
Two complementary lenses on the same problem. The Citation Benchmark measures the engine side: which domains AI engines actually cite when a buyer asks a question. This AEO Readiness study measures the publisher side: how technically prepared the scanned domains are to be cited. Combine them and the picture is: most sites are not yet ready, and even the ones that are face a fragmented citation graph where winning on ChatGPT does not mean winning on Google AI Overview. Both lenses point at the same conclusion. The work to be done is on the publisher side first.