How well are today's websites prepared for search engines, AI-powered discovery, and modern user expectations?
To investigate, we analyzed 200 websites using SEO GEO Reports, evaluating five key areas of website quality: Search Engine Optimization (SEO), Generative Engine Optimization (GEO), accessibility, security, and performance.
The results reveal significant room for improvement.
Across the 200 websites, our analysis identified 7,795 issues spanning 172 distinct issue types. GEO accounted for the largest share of findings, while traditional SEO generated the highest number of critical issues.
Security configuration problems were particularly widespread, and accessibility findings showed a relatively high proportion of critical issues.
What do these results tell us about the current state of websites? Which problems occur most frequently? And where should website owners focus their optimization efforts?
Let's examine the data.
1. Website Quality in Numbers: 200 Sites, 7,795 Issues
Our research evaluated five fundamental dimensions of website quality:
- SEO (Search Engine Optimization): Technical and content-related factors affecting traditional search engine visibility.
- GEO (Generative Engine Optimization): Content structure, clarity, and machine readability that may support discovery and interpretation by generative AI systems.
- Accessibility (WCAG): Website usability for people with disabilities, including those using assistive technologies.
- Security: HTTP security headers and other protective mechanisms.
- Performance: Loading speed, Core Web Vitals, and technical efficiency.
Key Findings
MetricResult
Websites analyzed
200
Total issues and recommendations
7,795
Unique issue types
172
Average findings per website
39
Critical findings
807
Warning-level findings
3,403
Informational findings
3,585
More than 10% of all detected findings were classified as critical by the auditing system.
However, not all findings have the same practical consequences.
A missing H1 heading, poor Largest Contentful Paint performance, and the absence of an optional llms.txt file represent very different optimization concerns.
For this reason, our analysis examines not only the number of issues but also their severity, frequency, and technical significance.
2. GEO Has the Most Findings — Accounting for 28% of All Issues
One of the most notable findings concerns Generative Engine Optimization.
GEO generated 2,184 findings, representing approximately 28% of the entire dataset.
This makes GEO the category with the highest number of reported optimization opportunities.
Distribution of Issues by Category
CategoryIssuesShare of Total
GEO
2,184
28.0%
SEO
1,812
23.2%
Security
1,612
20.7%
Performance
1,368
17.5%
Accessibility
819
10.5%
Total
7,795
100%
GEO also received the lowest average quality score in the dataset: 36.6 out of 100.
Why Did GEO Score So Poorly?
Many websites are still structured primarily around traditional web publishing and search engine optimization.
While their content may be accessible to users and search engines, it is not always organized in a way that makes information easy for automated systems to interpret.
Our audit frequently identified issues and recommendations involving:
- Missing or insufficient structured data.
- Limited information about authors and content sources.
- A lack of concrete figures, statistics, and measurable facts.
- Missing definitions or explanations of specialist terminology.
- Insufficiently structured or contextualized content.
One subcategory stood out in particular: AI Readability.
Across the sample, AI Readability accounted for 1,117 findings affecting 199 out of 200 websites.
This does not mean that 199 websites are invisible to ChatGPT, Gemini, or other AI services.
Rather, almost every website received at least one recommendation related to improving the clarity, organization, or machine readability of its content.
Key takeaway: GEO represents a substantial optimization opportunity in the analyzed sample, especially when it comes to presenting clear, structured, and trustworthy information.
3. The 10 Most Common Website Issues
Among the 172 unique issue types detected, some appeared far more frequently than others.
The following table shows the ten most common findings, ranked by the percentage of analyzed websites affected.
Top 10 Most Frequently Detected Issues
RankIssueWebsites Affected
1
Missing Permissions-Policy header
88.0%
2
Missing Referrer-Policy header
76.5%
3
Missing llms.txt file
75.5%
4
Missing security.txt file
74.5%
5
Missing Content-Security-Policy header
74.0%
6
Missing HSTS header
69.0%
7
Missing Speakable schema
63.5%
8
Missing X-Content-Type-Options header
62.0%
9
Missing clickjacking protection
61.5%
10
Missing FAQ schema
55.5%
Perhaps the most surprising result is the dominance of security-related findings.
Seven of the ten most frequently detected issues involved security configuration.
Security Headers Are Frequently Overlooked
The most common issue in the entire dataset was the absence of the Permissions-Policy HTTP header.
It was detected on 176 out of 200 websites (88%).
Permissions-Policy allows website administrators to control access to certain browser capabilities, such as cameras, microphones, and other APIs.
Other frequently missing security headers included:
- Referrer-Policy
- Content-Security-Policy
- Strict-Transport-Security (HSTS)
- X-Content-Type-Options
A missing security header does not necessarily indicate an exploitable vulnerability.
Some headers provide additional protection, while others require configuration appropriate to the website's architecture and functionality.
Nevertheless, the findings demonstrate how frequently HTTP security configuration is overlooked during website development and maintenance.
Does Every Website Need llms.txt?
The absence of an llms.txt file was detected on 151 websites (75.5%).
The proposed llms.txt convention provides a way to publish structured information about website resources for language-model-oriented workflows.
However, it is important to distinguish emerging practices from established technical requirements.
There is currently no general requirement for websites to publish llms.txt to be indexed, discovered, or referenced by generative AI systems.
Similarly, Speakable and FAQ structured data may be relevant for particular content types and use cases, but they are not universally necessary.
The absence of these elements should therefore be interpreted as an optimization recommendation rather than automatic evidence of poor AI visibility.
For most websites, content quality, accessibility, factual clarity, crawlability, and logical information architecture remain more fundamental priorities.
4. Traditional SEO Has the Highest Number of Critical Issues
Although GEO accounted for the largest number of overall findings, traditional SEO generated the greatest number of critical findings.
Our analysis identified 319 critical SEO issues.
Even more importantly, critical SEO findings appeared on 185 out of 200 websites (92.5%).
Three groups of problems deserve particular attention.
Poor Core Web Vitals
Largest Contentful Paint (LCP) problems were detected on 80 websites.
LCP measures how quickly the largest meaningful content element becomes visible within the viewport.
Poor results can be associated with oversized images, render-blocking resources, slow server responses, or inefficient page rendering.
In our dataset, LCP findings appeared under both SEO and Performance.
These represented the same 80 affected websites rather than 160 independent cases.
This distinction matters when interpreting automated audit results because individual technical problems may affect multiple quality categories.
Missing H1 Headings
A missing H1 heading was detected on 53 websites.
The H1 element helps communicate the primary topic of a page and contributes to a logical document structure.
Although a missing H1 does not automatically lead to lower search rankings, it may indicate weaknesses in semantic markup and content organization.
Missing Structured Data
Critical findings associated with missing JSON-LD were reported on 63 websites.
Structured data can help search engines understand information about organizations, products, articles, and other entities.
However, schema markup should accurately represent the page's actual content.
Not every website needs every Schema.org type, and adding markup does not guarantee enhanced search appearances or AI citations.
Key takeaway: Traditional SEO fundamentals remain essential. Improving AI readiness should not come at the expense of technical SEO, content structure, or crawlability.
5. Accessibility: Fewer Findings, but a Higher Proportion of Critical Problems
Accessibility generated 819 findings, making it the category with the lowest total number of reported issues.
However, 193 accessibility findings were classified as critical.
That represents approximately 23.6% of all accessibility findings — the highest critical-issue proportion among the five categories.
Most Common Critical Accessibility Findings
IssueWebsites Affected
Duplicate HTML IDs
70
Form fields without labels
62
Images missing alternative text
41
These problems can have meaningful consequences for website usability.
Duplicate IDs may interfere with assistive technologies or programmatic relationships between elements.
Missing form labels can make it difficult for screen reader users to understand what information is required.
Images without appropriate alternative text may prevent important visual information from being communicated to users who cannot see them.
Accessibility is not only about meeting technical requirements. It is also about making websites usable for a broader audience.
Furthermore, semantic HTML and logically structured content can support automated content interpretation.
It is important to note that automated accessibility testing cannot establish full WCAG conformance. Some accessibility requirements must be evaluated manually.
6. Performance Remains a Significant Weakness
The average Performance score across all 200 websites was 38.8 out of 100, making it the second-lowest category score after GEO.
Frequently detected performance-related problems included:
- Limited use of modern image formats.
- Missing resource-loading optimizations.
- Poor Core Web Vitals results.
- Inefficient loading of critical page resources.
For example, findings related to insufficient use of modern image formats appeared on 106 websites (53%).
Performance optimization can improve user experience and reduce unnecessary resource consumption.
Depending on the underlying cause, improvements may include:
- Compressing and resizing images.
- Using modern image formats where appropriate.
- Optimizing caching.
- Reducing unnecessary JavaScript.
- Improving server response times.
- Prioritizing critical rendering resources.
An Important Data Quality Observation
Our research also identified an anomaly affecting some quality scores.
A total of 72 websites received a score of zero in both Accessibility and Performance, without corresponding issue records in either category.
This may indicate incomplete scans or unavailable measurements rather than genuinely poor website performance.
When zero scores are excluded, the average scores change significantly:
CategoryAverage Across All SitesAverage Excluding Zero Scores
Accessibility
41.7
65.2
Performance
38.8
60.6
These differences highlight the importance of distinguishing genuinely poor results from missing or incomplete measurements.
For transparent reporting, both interpretations should be considered until the scan-completion status is verified.
7. Which Website Quality Category Needs the Most Improvement?
The answer depends on how website quality is measured.
If we look at the total number of findings, GEO ranks first.
If we focus on critical findings, traditional SEO presents the largest number of flagged problems.
Accessibility stands out because a comparatively high proportion of its findings are critical.
Overall Category Comparison
CategoryAverage Score (0–100)Critical Findings
GEO
36.6
138
Performance
38.8
148
Accessibility
41.7
193
SEO
48.2
319
Security
66.9
9
Three conclusions emerge.
GEO represents the largest pool of identified optimization opportunities. It has both the highest number of reported findings and the lowest average category score.
SEO accounts for the highest number of critical findings. Technical SEO and fundamental content structure remain major areas of concern.
Accessibility problems deserve particular attention. Even though the category produced fewer total findings, its critical-issue proportion was higher than that of any other category.
Category scores are based on different groups of checks and should not be treated as directly interchangeable measures of business risk.
The most effective optimization strategy should consider severity, user impact, technical dependencies, and the relevance of each recommendation to the website.
8. What Should Website Owners Prioritize?
The results of this study suggest that modern website optimization requires a broader approach than simply adjusting keywords, titles, and meta descriptions.
A practical improvement plan should address the following priorities.
Step 1: Fix High-Impact Technical Problems
Start with issues that genuinely affect website functionality, crawlability, security, and accessibility.
Prioritize confirmed technical defects rather than treating every automated warning as equally important.
Step 2: Improve Website Performance
Evaluate Core Web Vitals, optimize images, reduce unnecessary scripts, and investigate slow server responses.
Performance improvements can benefit both users and the technical quality of the website.
Step 3: Make Content Easier to Understand
Use clear headings, logical document structures, precise answers, meaningful definitions, and factual information.
Where appropriate, add structured data that accurately reflects the content.
Step 4: Strengthen Content Trustworthiness
Provide accurate organizational information, identify authors where relevant, cite reliable sources, and maintain up-to-date content.
These practices can improve the usefulness and credibility of information for both human readers and automated systems.
Step 5: Review Accessibility and Security
Check form labels, alternative text, semantic HTML, and relevant HTTP security headers.
Address accessibility barriers and security weaknesses according to their actual impact.
Step 6: Run Regular Website Audits
Website quality is not static.
CMS updates, design changes, new content, integrations, and infrastructure modifications can introduce new issues.
Periodic audits make it easier to identify regressions and maintain technical standards over time.
9. SEO and GEO Should Work Together
One of the broader lessons from our findings is that SEO and GEO should not be treated as competing disciplines.
Traditional SEO helps ensure that websites are technically accessible, well structured, and discoverable through search engines.
GEO focuses on additional considerations associated with presenting information clearly and making it easier for generative AI systems to interpret.
These goals often overlap.
Clear headings, useful factual content, consistent entity information, accessible HTML, and well-organized documents can support both traditional search and machine interpretation.
However, improving a GEO audit score does not guarantee inclusion in AI-generated answers.
Similarly, technically correct SEO does not automatically result in visibility across generative AI products.
The most sustainable strategy is to build websites that are accessible, reliable, understandable, and genuinely useful.
10. Final Conclusions: Thousands of Issues, Different Levels of Importance
Our analysis of 200 websites using SEO GEO Reports identified 7,795 findings across SEO, GEO, Accessibility, Security, and Performance.
The study revealed several clear patterns:
- GEO accounted for the largest share of reported findings, at 28%.
- SEO had the highest number of critical findings, with 319.
- Security configuration problems dominated the list of most frequently detected individual issues.
- Accessibility had the highest proportion of critical findings within its category.
- Performance and GEO received the lowest average quality scores in the full dataset, although incomplete measurements may affect some category averages.
Together, these findings indicate that many websites still have considerable room for improvement across several technical and content-related dimensions.
At the same time, they demonstrate why automated audits must be interpreted carefully.
An optional optimization recommendation is not equivalent to a serious accessibility defect, a security vulnerability, or a problem preventing search engine crawling.
The value of a website audit is not simply how many issues it detects, but how effectively it helps website owners identify what matters most.
How Does Your Website Compare?
Is your website prepared for traditional search engines, AI-powered discovery, and modern technical standards?
With SEO GEO Reports, you can analyze your website across five key areas:
SEO, GEO, Accessibility, Security, and Performance.
Identify potential weaknesses, understand optimization opportunities, and take a more structured approach to improving website quality.
Analyze Your Website with SEO GEO Reports
Research Methodology
This study is based on audit results for 200 sample websites generated using SEO GEO Reports.
The analyzed dataset contained 7,795 issue records covering 172 distinct issue types across five categories: SEO, GEO, Accessibility, Security, and Performance.
Issue frequency was calculated as the proportion of websites affected by each issue type. Average category scores were calculated at the website level.
The study is exploratory and should not be interpreted as a statistically representative sample of all websites on the internet.
Issue severity classifications reflect the audit system's assessment and do not necessarily indicate verified business impact or exploitable security vulnerabilities.
Automated checks cannot replace comprehensive technical, security, or accessibility assessments.
In addition, GEO audit results should not be interpreted as direct measurements of visibility, citations, or referral traffic from generative AI platforms.
Some category averages may be affected by incomplete scans, as described in the Performance section.