Original research, November 2026
AI Search Readiness of US Small Businesses, 2026
We checked the public websites of 3,000 US small businesses across 25 industries for the basics AI search tools rely on. Few block AI crawlers. Many give them little to read.
In short
Of 2,111 reachable sites, only 4.3% block any of eight major AI crawlers in robots.txt, so closing the door is rare. The bigger gap is structure: 32.1% publish no JSON-LD at all, 45.5% have no LocalBusiness-type or Organization markup, and 5.6% have FAQPage markup. Electricians, real estate agents and pain management clinics trail most on schema. This is a snapshot of what sites publish on one day, not proof of who AI tools cite.
Rather have it done for you? See our AI search optimization service, from $1,500.
Key findings
Percentages use the 2,111 sites whose homepage returned HTTP 200 (n shown for every group). Method and limits are below.
- Only 4.3% of reachable sites block any of eight AI crawlers in robots.txt. 91 of 2111 reachable sites. 29 (1.4%) block all eight. Among the 1735 sites that have a robots.txt file, the share is 5.2%.
- Most deliberate blocks target training bots, not ChatGPT's user agent. 69 sites name at least one AI crawler. 58 of them (84.1%) do not block ChatGPT-User. Another 22 sites block everything through a * rule (n=2111 reachable).
- 32.1% of sites publish no JSON-LD structured data at all. 678 of 2111 reachable homepages. Schema is the plainest way to tell a machine who you are, where you are and what you sell.
- 45.5% have no LocalBusiness-type or Organization markup; only 5.6% have FAQPage. 960 of 2111 sites have neither. FAQPage appears on 118 sites. LocalBusiness-type markup is on 35.2% (743).
- 30.5% serve an llms.txt file, but it looks like a platform default. 643 of 2111 sites. Wix: 103 of 110. Shopify: 16 of 19. Squarespace: 0 of 100. Outside Wix and Shopify the rate is 26.4% (524 of 1982). We did not test whether any AI system reads these files.
- Electricians lag on schema: 49.4% have none, against 21.8% of HVAC sites. Electricians n=81, HVAC n=87, dentists 22.0% (n=82). Real estate agents (43.5%, n=69) and pain management clinics (42.5%, n=80) are also high.
- Real estate agents and doctors block AI crawlers most; pest control sites not at all. Real estate agents 11.6% (n=69), doctors and clinics 9.8% (n=82), pest control 0.0% (n=74). Gaps this size are beyond the margin of error only at the extremes.
- Site platform tracks schema use: 3.6% of Wix sites have none, against 14.6% on WordPress and 53.7% where we found no common platform fingerprint. Wix n=110, WordPress n=889, no fingerprint n=885. Association only; platform choice and business size are tangled together.
Overall results
| Signal | Sites | Share of reachable sites |
|---|---|---|
| HTTPS | 2085 | 98.8% |
| robots.txt present | 1735 | 82.2% |
| llms.txt present | 643 | 30.5% |
| JSON-LD present | 1433 | 67.9% |
| No JSON-LD | 678 | 32.1% |
| LocalBusiness-type schema | 743 | 35.2% |
| Organization schema | 748 | 35.4% |
| FAQPage schema | 118 | 5.6% |
| Meta description | 1626 | 77.0% |
| Viewport meta | 1935 | 91.7% |
| H1 present | 1644 | 77.9% |
| Title 30 to 65 characters | 1141 | 54.1% |
| Blocks 1+ of 8 AI crawlers | 91 | 4.3% |
Basic hygiene is mostly in place: 98.8% use HTTPS and 91.7% declare a mobile viewport. 77.9% have an H1 and 77.0% have a meta description. Structured data and machine-readable summaries are where the gaps are.
Free · no obligation
Get a free SEO and AI visibility check of your site
We look at how Google and AI tools like ChatGPT see your site and send you the top fixes, within one working day.
Request received
Thanks. Your free check is in the queue.
We will send the top fixes on WhatsApp or email within one working day.
Results by industry
Industries with at least 60 reachable sites, sorted by share with no JSON-LD. Each row is one stratum of the sample, so n is small and each figure carries a margin of about 5 to 10 points.
| Niche | Sites (n) | No JSON-LD | LocalBusiness-type | FAQPage | llms.txt | Blocks 1+ AI bot |
|---|---|---|---|---|---|---|
| Electricians | 81 | 49.4% | 27.2% | 3.7% | 33.3% | 4.9% |
| Real estate agents & investors | 69 | 43.5% | 23.2% | 4.3% | 27.5% | 11.6% |
| Pain management clinics | 80 | 42.5% | 23.8% | 10.0% | 22.5% | 3.8% |
| Remodeling & custom homes | 84 | 40.5% | 27.4% | 2.4% | 32.1% | 1.2% |
| Property management | 98 | 38.8% | 21.4% | 1.0% | 30.6% | 1.0% |
| Solar | 91 | 38.5% | 27.5% | 5.5% | 27.5% | 6.6% |
| Pest control | 74 | 36.5% | 32.4% | 9.5% | 35.1% | 0.0% |
| Pool contractors | 87 | 35.6% | 25.3% | 5.7% | 24.1% | 6.9% |
| Mortgage brokers & loan officers | 88 | 34.1% | 19.3% | 5.7% | 31.8% | 4.5% |
| Senior & home health care | 85 | 34.1% | 29.4% | 2.4% | 35.3% | 1.2% |
| Physio / rehab / acupuncture | 91 | 33.0% | 26.4% | 1.1% | 25.3% | 4.4% |
| Plumbing | 91 | 33.0% | 28.6% | 8.8% | 39.6% | 2.2% |
| Doctors & clinics | 82 | 32.9% | 22.0% | 2.4% | 15.9% | 9.8% |
| Financial advisors & wealth | 89 | 32.6% | 30.3% | 4.5% | 15.7% | 3.4% |
| Water-damage restoration | 82 | 31.7% | 46.3% | 4.9% | 32.9% | 6.1% |
| Insurance agencies & agents | 94 | 29.8% | 42.6% | 3.2% | 19.1% | 2.1% |
| Garage doors | 90 | 27.8% | 43.3% | 5.6% | 37.8% | 4.4% |
| Chiropractors | 87 | 26.4% | 46.0% | 8.0% | 23.0% | 2.3% |
| Roofing | 74 | 25.7% | 44.6% | 9.5% | 40.5% | 4.1% |
| Plastic surgery / med-spa / aesthetics | 80 | 25.0% | 43.8% | 8.8% | 32.5% | 6.2% |
| Personal injury / property-damage lawyers | 87 | 24.1% | 43.7% | 5.7% | 31.0% | 5.7% |
| Law firms & lawyers | 90 | 22.2% | 43.3% | 6.7% | 28.9% | 4.4% |
| Car & boat dealers | 68 | 22.1% | 75.0% | 4.4% | 61.8% | 7.4% |
| Dentists & orthodontists | 82 | 22.0% | 54.9% | 11.0% | 23.2% | 4.9% |
| HVAC | 87 | 21.8% | 41.4% | 6.9% | 42.5% | 1.1% |
Results by state
States with at least 40 reachable sites (15 of 50), sorted by share with no JSON-LD. State comes from the listing address, and the lead file skews toward California, Texas and Florida. Differences of under about 10 points are within the margin of error.
| State | Sites (n) | No JSON-LD | LocalBusiness-type | FAQPage | llms.txt | Blocks 1+ AI bot |
|---|---|---|---|---|---|---|
| Florida | 164 | 37.8% | 36.0% | 4.3% | 30.5% | 4.3% |
| Massachusetts | 61 | 36.1% | 32.8% | 9.8% | 24.6% | 1.6% |
| Texas | 328 | 35.4% | 34.5% | 6.1% | 31.4% | 6.4% |
| Arizona | 88 | 35.2% | 36.4% | 11.4% | 33.0% | 6.8% |
| California | 388 | 34.0% | 33.8% | 5.2% | 31.4% | 3.9% |
| Ohio | 110 | 32.7% | 39.1% | 5.5% | 31.8% | 0.0% |
| Michigan | 44 | 31.8% | 36.4% | 9.1% | 36.4% | 0.0% |
| Illinois | 101 | 31.7% | 39.6% | 4.0% | 31.7% | 5.0% |
| New York | 112 | 31.2% | 29.5% | 1.8% | 26.8% | 5.4% |
| Colorado | 53 | 30.2% | 34.0% | 11.3% | 24.5% | 0.0% |
| Washington | 44 | 29.5% | 27.3% | 4.5% | 20.5% | 13.6% |
| Indiana | 51 | 29.4% | 29.4% | 2.0% | 21.6% | 0.0% |
| North Carolina | 56 | 23.2% | 41.1% | 12.5% | 33.9% | 1.8% |
| Pennsylvania | 58 | 22.4% | 32.8% | 10.3% | 25.9% | 10.3% |
| Tennessee | 57 | 21.1% | 36.8% | 1.8% | 31.6% | 5.3% |
What the AI crawler blocks look like
69 sites name at least one AI crawler in robots.txt. 58 of them do not block ChatGPT-User. 22 more block every crawler with a * rule, which usually means a site closed to search engines too. We did not check whether any crawler obeys these rules.
| Crawler | Sites blocking it | Share |
|---|---|---|
| GPTBot | 88 | 4.2% |
| ChatGPT-User | 31 | 1.5% |
| OAI-SearchBot | 31 | 1.5% |
| ClaudeBot | 83 | 3.9% |
| anthropic-ai | 82 | 3.9% |
| PerplexityBot | 30 | 1.4% |
| Google-Extended | 80 | 3.8% |
| CCBot | 85 | 4.0% |
Method
- Source: Liftup's file of US business listings gathered from public directories. 96,157 rows had a website. Chains, directories and listing or social hosts were removed.
- Sample: 25 industries, 120 distinct domains drawn at random from each, 3,000 in all. Overall figures therefore weight each industry equally.
- Crawl: 8 November 2026, one pass. For each domain we fetched /robots.txt, /llms.txt and the homepage with GET requests. Public pages only, no forms or logins, 10 second timeout, at most 25 requests at once, redirects followed, User-Agent including LiftupResearchBot (+https://liftup.tech/research/).
- Reachable: homepage returned HTTP 200. 2,111 of 3,000 (70.4%) did. 291 returned 403 and 256 had no usable DNS. A 403 may be bot protection, so unreachable does not mean offline.
- robots.txt: parsed under RFC 9309. The group naming the crawler wins, otherwise the * group. Longest matching rule wins, Allow beats Disallow on a tie. Blocked means Disallow covers the root path. Sites with no robots.txt count as not blocking.
- llms.txt: HTTP 200, non-HTML body, more than 20 bytes.
- Schema: JSON-LD in the delivered HTML only. Markup injected by JavaScript after page load is not seen.
- Margin of error: about +/-1.9 points overall at 50%, about +/-5 to 10 points per industry, wider per state.
- Limits: homepage only, one point in time, no test of whether AI tools actually cite any site. No causal claims.
What owners can do with this
- Check your own robots.txtOpen yourdomain.com/robots.txt and look for Disallow rules against GPTBot, ClaudeBot, PerplexityBot or a blanket * rule. Our AI visibility checker reads it for you.
- Add structured data that matches your businessUse LocalBusiness or the closest subtype, with name, address, phone, hours and services. Build it with the schema markup generator and test it before publishing.
- Answer real questions on the pageIf customers ask the same five questions, put them on a page in plain text and mark them up as FAQ only where the page genuinely shows them.
- Consider an llms.txt fileIt is a short Markdown summary of your site and key pages. Nobody has shown it changes what AI tools say about you, but it costs little. Use the llms.txt generator.
- Decide on purpose, not by defaultBlocking training crawlers while allowing search and user agents is a legitimate choice. Make it deliberately. If you want help, see AI search optimization.
Cite this study
Liftup Technologies. "AI Search Readiness of US Small Businesses, 2026." Published 8 November 2026. https://liftup.tech/research/ai-search-readiness-2026/
Please quote figures with their sample size and the date. Underlying counts per niche and state are on this page. For the raw per-site table or a custom cut for your region or industry, email rohit@liftup.tech.
Sources: RFC 9309: Robots Exclusion Protocol, llms.txt proposal, Schema.org LocalBusiness.
Updated 8 November 2026
Questions, answered
How many sites did you check?
Does blocking AI crawlers hurt my visibility?
Does having an llms.txt file help me get cited?
Why are some sites missing from the results?
Is this representative of all US small businesses?
Can I use the data?
Keep exploring
AI visibility checker
A short audit of the things you control: crawler access, page structure, entity signals and measurement. You get a score out of 20 and a fix list in…
Free toolllms.txt generator
Add your site name, a short summary and your most useful pages with a note on each. The generator returns a formatted llms.txt file ready to upload…
Free toolSchema markup generator
Pick a type, fill in the fields that match your page, and copy valid JSON-LD. The generator leaves out empty fields so you never publish a blank…
Curious how your own site compares?
Send us your domain and we will run the same checks and tell you what to fix first, with a fixed quote after a short call.