Back to Main Site
Measured on 1,810 small business sites

Your site works fine for people. It may be turning AI assistants away.

139 of the 1,810 sites measured refused an AI assistant outright, while answering a browser perfectly.

Assistants do not browse your site the way a person does. They fetch it with their own named user agents, and some hosts and content delivery networks refuse those names by default. The owner is almost never told, and nothing in the site's own admin screens mentions it. The check below sends those exact names and shows you what came back.

No account, no email required. Takes about ten seconds.

What a run over 1,810 sites actually showed

139

refused an AI assistant while answering a browser normally.

444

already publish an llms.txt file. Almost none of those owners wrote one.

1,296

carried at least one thing an assistant would trip on.

The pattern is a vendor default, not a decision

Those two numbers point the same way. Almost nobody chose to block assistants, and almost nobody chose to publish a file for them either. Both were applied by a platform, a plugin or a content delivery network, and in both directions the site owner was the last to know. That is the useful finding: this is not a strategy anyone adopted, it is a setting that moved underneath them.

Which is why it is usually cheap to fix

Where a content delivery network is doing it, this is a switch and not a project. Where a managed host is doing it above your account, you may not be able to change it at all, and knowing that is worth something too. The check names which one you are dealing with, so the next step is either five minutes or nothing.

What this page is not claiming

Not that AI assistants are about to send you customers. Nobody can measure that honestly yet, and the claim is doing a lot of work in a lot of sales copy right now. What is measurable is narrower and still worth ten seconds: whether an assistant asked about your business can read your pages at all.

Find out in about ten seconds

The check requests your home page as a browser and again as the assistants' own published user agents, then looks for the three files they go after. It changes nothing on your site.

Run the free check

Questions

Because in almost every case you did not. Of the sites measured that refuse assistants, most sit behind a content delivery network or a hosting platform whose default was changed for them. Cloudflare added a per-site switch for it. WordPress.com applies its own. The owner is rarely told, and nothing in the site's own admin screens mentions it.

You would not, from a browser. That is the whole difficulty. These sites answer a person normally and answer a named assistant with a refusal, so every ordinary way you look at your own site shows it working. It takes a request that identifies itself the way an assistant does, which is what the free check sends.

Usually yes, and it is a setting rather than development work. It is a per-site policy, not something Cloudflare applies everywhere: in the sample it fronted about four sites in ten and only a handful of those were refusing anyone. If your check names Cloudflare, that is the good outcome, because you control it.

Then the honest answer is that you probably cannot, and you should know that rather than spend money trying. On some managed platforms the block is applied above your account and there is no switch in your dashboard. The check names which one you are dealing with so you are not guessing.

Nobody can honestly tell you that, and be careful of anyone who does. What is measurable is the other half: whether an assistant asked about your business can read your pages at all, and right now a lot of sites cannot be read. Whether that turns into lost work depends on how people end up searching, which is still moving. This is worth a few minutes and a setting change. It is not worth a rebuild on its own.

It overlaps and it is not the same. Search crawlers and assistant fetchers are different user agents with different rules, so a site can rank perfectly well and still be refused. Several of the sites measured had a clean sitemap, a sensible robots.txt and good search visibility, and still answered an assistant with a refusal.

If it is a setting, tell me what the check reported and I will tell you where the switch is, at no charge. If it turns out the site itself is the problem, which is a different finding and a bigger one, we would talk about scope first. Nothing here is a subscription and nothing starts without a written price.

Tell me what the check said

Run it first if you want something concrete to point at. Paste what it reported and I'll tell you whether it is a setting or a real piece of work.

Prefer to skip the form? Book directly on Calendly or call (440) 836-3059.