AI Visibility Benchmark: What SRP Found Across 321 Small Business Websites

SRP's analysis of 321 small and midsize business websites found four recurring barriers to AI readiness. Missing JSON-LD structured data affected 78 percent and 71 percent lacked a direct answer after a question. In addition, 65 percent had a poorly organized heading hierarchy, and 51 percent lacked identifiable authorship. These findings measure website readiness, not guaranteed citations or recommendations.

By Sebastian Pincetti, Partner, Digital Marketing Specialist, and AI Developer at SRP Communication & Brand Design. Published October 1, 2026. Last updated October 10, 2026.

Your next step: get started with SRP's free online AI Visibility Audit to identify automated issues on your website. Then contact SRP if you need help reviewing or implementing the recommendations.

Why Does an AI Visibility Benchmark Matter?

An AI visibility benchmark matters because it replaces assumptions with observable evidence. For example, a company may rank in traditional search and still make its identity, services, or key answers difficult for machines to interpret. A benchmark reveals how often those barriers occur across a defined group of websites and gives business leaders a practical basis for deciding what to fix first.

This article is built for business owners, CEOs, managers, and marketing decision makers at companies with approximately 5 to 250 employees.

What Did SRP Measure?

SRP measured technical readiness for AI-assisted discovery. The automated audit examined whether a website could be accessed, whether it presented consistent business information, and whether important content was organized clearly enough to extract. The benchmark did not measure whether ChatGPT, Gemini, Perplexity, or another platform recommended the business in live answers.

• Crawler access. Whether robots.txt rules, server responses, or security controls prevented relevant automated crawlers from accessing public content.
• Entity clarity. Whether visible company information and structured data clearly identified the organization, location, services, and official website.
• Content structure. Whether headings, paragraphs, navigation, and service descriptions made important answers easy to locate and interpret.
• Supporting signals. Whether the site included elements such as author attribution, publication dates, source links, contact details, and consistent business information.

How Was the Benchmark Created?

SRP created the benchmark by running its automated AI Visibility Audit on 321 unique business websites from June through [end month] 2026. The process had four steps.

1. Step 1, collect the scans: SRP gathered audits from businesses in manufacturing, IT services, legal services, clinics, medical practices, bakeries, restaurants, professional services, marketing, digital marketing, general contracting, and construction.
2. Step 2, clean the sample: each domain was counted once. Repeat tests, staging sites, and incomplete scans were excluded.
3. Step 3, anonymize the data: SRP removed company names, URLs, contact information, and other identifying details before aggregating the findings.
4. Step 4, calculate the results: each percentage is the share of eligible domains that failed a check.

Because one website could fail more than one check, the category percentages may total more than 100 percent.

What Were the Most Common AI Visibility Barriers?

The most common AI visibility barriers were missing JSON-LD structured data, missing direct answers, disorganized heading hierarchies, and missing author attribution. The table presents the findings from most common to least common.

What Does Missing JSON-LD Structured Data Mean?

Missing JSON-LD structured data means the website does not provide a standardized machine-readable layer that describes the organization, page, article, service, or another relevant entity. Missing JSON-LD affected 78 percent of audited websites, according to SRP (321 audits, June to [end month] 2026).

Structured data gives Google explicit clues about the meaning of a page, according to Google Search Central (2026). However, valid markup does not guarantee a particular search appearance. So the same caution applies to AI visibility. Structured data can improve clarity, but it cannot force an AI platform to cite a business.

What Does a Missing Direct Answer Look Like?

A missing direct answer appears when a page asks a useful question but delays the response, starts with background information, or never provides a concise answer. SRP found this problem on 71 percent of the audited websites. A stronger section begins by repeating the main idea in the question and answering it in the first sentence, followed by explanation and evidence.

What Does a Disorganized Heading Hierarchy Look Like?

A disorganized heading hierarchy appears when a page has no clear H1, uses several H1 headings without a deliberate reason, or skips levels. It also appears when headings are chosen for visual size rather than content structure. SRP found heading-organization problems on 65 percent of the audited websites. A logical hierarchy helps readers scan the page and helps machines identify the relationship between topics and supporting details.

What Does Missing Author Attribution Mean?

Missing author attribution means the page does not identify who wrote or reviewed the content. SRP found this issue on 51 percent of the audited websites. A visible author name, a relevant biography, and accurate Article and Person structured data can help readers and machines connect the content with accountable expertise.

However, strong answer structure does not mean turning every page into a list of robotic questions. The best pages combine direct summaries, useful explanations, examples, evidence, and natural transitions. Human readability remains the priority.

How Do SRP's Findings Compare With Other Benchmarks?

SRP's findings are directionally consistent with other public benchmarks, particularly around answerability, headings, and authority signals. Unlike broad crawls of thousands or millions of sites, SRP's benchmark focuses only on small and midsize business websites and tests four readiness checks together. So the percentages should not be compared as if the studies used the same test. Each benchmark has a different sample, scoring model, and definition of failure.

• Answerability. Only 17.9 percent of 3,001 domains survived the complete funnel through the answerable stage, according to Nyman Media (2026). Its Content Answerability category had a mean score of 39.6 and a median of 40. Those results support SRP's finding that 71 percent of sites lacked a direct answer after a question, although the two methodologies are not identical.
• Self-contained sections. Nyman Media reported that retrievable, self-contained sections passed on 93.3 percent of top-quartile sites but only 31.7 percent of the rest. That gap reinforces the practical value of question-led sections with concise answers.
• Heading structure. Skipped heading levels appeared on 41.8 percent of one million home pages, according to WebAIM (2026). In the same study, 7.5 percent had no headings and 18.1 percent had more than one H1. SRP's 65 percent is higher because SRP used a broader heading-organization check and a smaller, self-selected business sample.
• Authorship and freshness. Nyman Media found Content Freshness and Authority to be its weakest scored category, with a mean of 34.7 and a median of 25. That category includes bylines, publication dates, modification dates, and author signals, making it directionally consistent with SRP's 51 percent missing-author result.
• Structured data. Other AI-readiness frameworks evaluate JSON-LD, Organization schema, Article schema, and related markup, but the public benchmarks reviewed did not provide a directly comparable missing-JSON-LD rate. SRP's 78 percent should therefore be presented as an original finding from this specific sample.

What Else Should Future SRP Benchmarks Measure?

Future SRP benchmark editions should retain the four current checks and add a small set of complementary signals. These additions would make the report more complete without confusing technical readiness with observed AI citations.

• Crawler access. Check robots.txt, meta robots, X-Robots-Tag, firewall responses, and access for relevant search crawlers.
• Fetch and rendering. Record HTTP status, HTTPS, redirect chains, response time, and whether important content is available without client-side JavaScript.
• Discovery and architecture. Check XML sitemaps, canonical URLs, internal links, and whether the audited page appears in the site structure.
• Entity consistency. Compare the company name, services, location, and profile links across visible copy, metadata, JSON-LD, and major external profiles.
• Publication evidence. Measure publication date, last-updated date, sources, author biography, and review attribution where appropriate.
• Observed AI visibility. Track a separate set of branded, non-branded, and comparison prompts across relevant platforms. Report this separately from the automated website-readiness score.

What Should a Business Fix First?

A business should first confirm that important pages are accessible, then repair the four problems identified in the benchmark. Access remains a prerequisite even though it was not one of the four findings reported in this edition.

1. Confirm intended crawler access. Review robots.txt, HTTP status codes, firewall rules, and bot-management settings. Allow only the crawlers that support the company's publishing and privacy choices.
2. Add accurate JSON-LD. Implement page-appropriate structured data that reflects information visible to readers. Validate the markup and keep it consistent with the site.
3. Reorganize headings and answers. Use one clear page topic, a logical H1-to-H3 hierarchy, and a direct answer immediately after important question headings.
4. Identify accountable authors. Publish named authors, relevant biographies, dates, sources, and review attribution where appropriate.
5. Test outcomes separately. After technical improvements, run a documented set of branded, non-branded, and comparison prompts across relevant AI platforms. Technical readiness and observed mentions are different measurements.

What Does the Benchmark Not Prove?

The benchmark does not prove that a website will appear in an AI-generated answer. It measures observable website conditions that can support or obstruct discovery and interpretation. AI outputs also depend on the user's question, location, model, retrieval system, available sources, external reputation, and changes made by the platform provider.

For example, the benchmark also does not establish that structured data alone causes more AI citations. A valid causal study would require controlled before-and-after testing, comparable sites, repeated prompt measurements, and enough time for recrawling and source updates.

How Can You Compare Your Website With the Benchmark?

You can compare your website with the benchmark by running SRP's free online AI Visibility Audit. The audit checks issues that can be detected automatically and provides a practical plan for what to review next. It is self-service and requires no appointment.

However, treat the score as a diagnostic starting point. Review important findings manually, confirm that recommendations fit your business and security policies, and test live AI answers separately.

Frequently Asked Questions

What is an AI visibility benchmark?

An AI visibility benchmark is a comparison of measurable conditions that affect whether websites can be accessed, understood, and used as sources by AI-assisted discovery systems. A credible benchmark states its sample, dates, checks, exclusions, and limitations.

Does schema markup guarantee visibility in ChatGPT or Gemini?

Schema markup does not guarantee visibility in ChatGPT, Gemini, or any other AI platform. Accurate structured data can clarify business facts for machines, but citations and recommendations depend on many technical, content, authority, and platform-specific factors.

Can a website pass an automated audit and still receive no AI mentions?

A website can pass an automated audit and still receive no AI mentions. Automated checks measure readiness and detectable barriers. They do not measure every external source, brand signal, prompt variation, or retrieval decision involved in a live answer.

How often should AI visibility be reviewed?

AI visibility should be reviewed after material website changes and on a regular schedule. Quarterly technical checks are a practical starting point for many small and midsize businesses, while live prompt monitoring may be more frequent for competitive or fast-changing markets.

Methodology and Limitations

SRP's benchmark uses anonymized results from automated website scans. It evaluates deterministic technical and content signals available at the time of each scan. It does not inspect private analytics, verify every page manually, measure every AI crawler, or test every possible prompt. Results may also be affected by temporary server responses, security services, geolocation, and platform updates.

All websites were evaluated using the SRP AI Visibility Audit benchmark methodology, June through October 2026 edition. The benchmark data was exported on October 1, 2026. For the full audit method, read What Is an AI Visibility Audit and What Does It Actually Measure?

About the author

Primary Sources

• Google Search Central: Introduction to structured data markup in Google Search. Accessed October 13, 2026.
• Google Search Central: General structured data guidelines. Accessed October 13, 2026.
• OpenAI: Overview of OpenAI crawlers. Accessed October 13, 2026.
• Schema.org: Organization. Accessed October 13, 2026.
• Nyman Media: The State of AI Discoverability. Published August 9, 2026. n=3,001 domains. Accessed October 13, 2026.
• WebAIM: The WebAIM Million (2026). n=1,000,000 home pages. Accessed October 13, 2026.