This page contains no new data. It differences the frozen figures in the August 2026 and September 2026 editions — which is why no number here can ever disagree with the pages it came from.
Reading a monthly difference. Each month is a separate sample, and its size changes from month to month. A change in sample size can move a median and a distribution by itself, so a monthly difference should not be mistaken for a change in scores: it does not show that sites improved or got worse.
The change, tool by tool
From August 2026 to September 2026, 13 of the 25 tools had a lower median, 10 the same and 2 a higher one. The largest fall was on the Entity Extractor (79 → 58) and the largest rise on the Mobile-Friendly Checker (84 → 92). In the chart below each tool’s grey dot is its August 2026 median and its coloured dot the September 2026 one.
- August 2026
- September 2026 — median up
- September 2026 — median down
| Tool | Median then | Median now | Change | Scored 80+ | Issues/run |
|---|---|---|---|---|---|
| Entity Extractor | 79 | 58 | −21 pts | 19% −5 pp | 1.9 +0.3 |
| Entity Coverage Analyzer | 67 | 52 | −15 pts | 27% −5 pp | 2.5 −0.1 |
| Mobile-Friendly Checker | 84 | 92 | +8 pts | 65% +3 pp | 2.0 no change |
| AI Visibility Checker | 67 | 59 | −8 pts | 24% −11 pp | 6.5 +0.6 |
| On-Page SEO Audit | 82 | 74 | −8 pts | 31% −28 pp | 11.3 +3.0 |
| GEO Checker (Generative Engine Optimization) | 60 | 53 | −7 pts | 17% −5 pp | 7.5 +0.3 |
| Technical Site Audit (Crawler) | 67 | 60 | −7 pts | 26% −9 pp | 4.6 no change |
| GEO & AEO Readiness Checker | 64 | 58 | −6 pts | 13% −8 pp | 16.1 +1.3 |
| NLP Content Analyzer | 71 | 65 | −6 pts | 10% −5 pp | 2.9 +0.3 |
| Author Page Audit | 54 | 49 | −5 pts | 8% −8 pp | 3.8 +0.3 |
| Topical Authority Map | 42 | 47 | +5 pts | 8% +1 pp | 3.1 no change |
| All-in-One SEO Audit | 76 | 73 | −3 pts | 28% −12 pp | 15.1 +1.3 |
| Meta Tag Analyzer | 83 | 80 | −3 pts | 51% −10 pp | 3.3 +0.5 |
| LLM Readability Score | 75 | 72 | −3 pts | 39% −6 pp | 3.2 +0.3 |
| AI Content Score | 64 | 63 | −1 pts | 10% −1 pp | 6.3 +0.2 |
| Page Speed & Size Test | 90 | 90 | no change | 75% −6 pp | 1.2 +0.1 |
| E-E-A-T Analyzer | 38 | 38 | no change | 1% no change | 6.8 +0.2 |
| AI Citation Checker | 46 | 46 | no change | 7% −1 pp | 7.4 +0.1 |
| Content Freshness Score | 13 | 13 | no change | 1% no change | 4.4 −0.1 |
| Thin Content Checker | 36 | 36 | no change | 21% −7 pp | 3.0 +0.1 |
| Local SEO Audit | 22 | 22 | no change | 1% +1 pp | 6.2 +0.2 |
| AEO Checker (Answer Engine Optimization) | 40 | 40 | no change | 1% no change | 6.0 −0.2 |
| AI FAQ Optimizer | 30 | 30 | no change | 1% no change | 0.0 no change |
| Anchor Text Optimizer | 79 | 79 | no change | 45% +11 pp | 1.2 +0.1 |
| FAQ Gap Finder | 0 | 0 | no change | 2% no change | 2.8 no change |
“pp” means percentage points: a share moving from 35% to 40% has risen 5 pp. Green marks a figure moving in the favourable direction — so a falling share of unmet checks shows green too. The colour states direction only; it does not mean sites improved.
How the sample changed
This is the first thing to look at before reading anything else. Scans went 34,236 → 86,983, unique sites 5,586 → 22,549. The September 2026 edition measured about 4 times as many sites. The two months are not the same sites, so a move on this page need not mean sites changed — it can equally mean the cross-section being measured did.
Check by check: which rates fell and which rose
The percentages below are the share of scans in which a SeoMods check was not met; a falling rate means the check went unmet in a smaller share of that month’s sample. For the full audit tool, biggest movers first.
| Check | Not met then | Not met now | Change |
|---|---|---|---|
| llms.txt present | 71% | 81% | +10 pp |
| Healthy text-to-HTML ratio | 85% | 78% | −7 pp |
| Structured data for AI | 53% | 60% | +7 pp |
| Structured data (Schema.org) | 53% | 60% | +7 pp |
| All images have ALT text | 55% | 60% | +5 pp |
| Meta description length (70–160) | 50% | 55% | +5 pp |
| Title length optimal (30–60) | 53% | 55% | +2 pp |
| No fixed desktop widths | 64% | 63% | no change |
| Question-style headings | 60% | 59% | no change |
| Author & date (E-E-A-T) | 91% | 91% | no change |
AI visibility checks
| Check | Not met then | Not met now | Change |
|---|---|---|---|
| llms.txt present | 70% | 82% | +12 pp |
| Social / sameAs profiles linked | 41% | 51% | +10 pp |
| JSON-LD structured data | 48% | 57% | +9 pp |
| Organization / WebSite schema | 53% | 61% | +8 pp |
| About page linked | 52% | 57% | +5 pp |
| Author / byline | 60% | 57% | −3 pp |
| In-depth content (≥800 words) | 37% | 40% | +3 pp |
| Content in raw HTML (server-rendered) | 33% | 30% | −3 pp |
Every other tool, check by check
The two tables above open up the full audit and AI visibility only. Every tool has the same breakdown on its own September 2026 page: the figure on each card is its September 2026 median, and the line under it the August 2026 median and the gap.
On AI visibility the median went 67 → 59. That difference, too, is between two separate samples; the AI visibility score is calculated against SeoMods’ own criteria.
How to read this comparison
- Cross-section, not panel. Each month measures whatever sites were scanned that month. This does not follow the same sites over time.
- Small moves are noise. Differences under half a point are marked “no change” and should not be read as a trend.
- Sample size comes first. If the scan count changed by an order of magnitude, a shift in the median is far more likely to be composition than behaviour.
- One methodology. Both editions were measured with the same checks and the same scoring; the method is on the hub.
What these scores do not measure
- It is not a Google ranking. A high score does not guarantee a better position, and a low one does not mean Google has penalised the page.
- It is not an organic traffic estimate. No click, impression or visit data goes into it.
- It is not Google’s E-E-A-T score. Google publishes no such score; E-E-A-T is a concept from its search quality rater guidelines.
- It is not a Google PageSpeed Insights score. Nothing here is run in a browser, and Core Web Vitals (LCP, INP, CLS) are not measured.
- It does not represent the whole of the real user experience. What is measured is one URL’s HTML and server response at the moment of the scan.
- It is not a direct measurement of the signals every AI system uses. Which system uses which signal, and how, is not public; the score measures the checks SeoMods chose.
About this benchmark. This benchmark is calculated from data on the sites SeoMods has scanned. The results do not necessarily represent all websites, or a random selection of them. The scores express SeoMods’ own measurement and scoring criteria; they are not ranking, quality, E-E-A-T or AI visibility scores issued by Google. Sample, anonymization and licence: the full methodology.
Frequently asked questions
Did SEO scores go up or down between August 2026 and September 2026?
The median audit score moved from 76 to 73. Two things govern how that should be read: the sample size changed (34,236 → 86,983 scans), and the sites measured in the two months are not the same sites. So this does not mean individual sites improved — it is the difference between two cross-sections of the web.
Why is the difference not the same as sites improving?
Because each month’s sample is whatever domains were scanned that month, and those sets do not overlap exactly. A fall in the share of scans not meeting a SeoMods check can come from existing sites starting to meet it, or from a different — and differently sized — mix of sites being scanned. Read the comparison as two separate cross-sections, not as a panel study following the same sites over time.