What to do with the result
Five things, in an order. The order is by how certain each one is, because a list ordered by payoff would be a forecast, and a forecast here would be invented.
Why this order and not another
No AI engine publishes how it decides which businesses to name. There is no documentation, no weighting, no confirmed factor list. So the honest way to rank a to-do list is by how sure anyone can be that the item matters — not by what it might return, which nobody can know.
The first item is mechanically certain. The rest get less certain as you go down, and the last one is explicitly a checkbox. Do the first one even if you do nothing else on this page.
1. Make sure an assistant can open your pages
Sent here by: cited on nothing, or named without ever being cited.
An assistant that cannot fetch your page cannot quote your page. Nothing further down this list is even in
the room until that is true. Start with what your robots.txt says to the AI crawlers —
the crawler check reads the file and lists what is allowed and what is
blocked.
Blocks arrive by accident far more often than by decision: a plugin toggle switched on by a previous developer, a security setting at your CDN that challenges anything not obviously a human, a staging site copied to live with its “hide from search engines” box still ticked. None of those announce themselves.
Blocking a crawler does prevent your page being fetched. Unblocking one does not cause anything to be quoted. That asymmetry is the whole reason this is first rather than last: it is the only item on the page where the mechanism is certain.
Who does what. Reading the file takes thirty seconds and costs nothing. Changing it, or undoing a CDN rule, is a job for whoever maintains the site — send them the exact line and the file it is in.
2. Make sure your words arrive with the page
Sent here by: named but never cited, on questions your site does answer.
Some sites deliver a nearly empty page and then build the visible text in the visitor’s browser afterwards. Whether a given assistant waits for that to finish is not published by any of them. Rather than research an unanswerable question, remove it: if the text is already in the page as it is delivered, what the fetcher does or does not wait for stops mattering. The render check fetches your page and shows you what is actually in it.
Who does what. Running the check is yours. Getting text into the delivered page is a development job, and usually not a small one — which is why it is worth knowing whether you have the problem before anyone quotes for it.
3. Make it unambiguous who you are
Sent here by: named as something you are not, or not recognised as one business across the answers.
An assistant answering “who does boiler repairs in Sheffield” has to decide whether the business on your website, the one in the directory listing and the one on the review site are the same company or three different ones. What any engine does after that guess is not published. What you can do is remove the guess.
- One name, spelled one way, everywhere.
- One address and one phone number, identical on your site, your Google Business Profile, your social profiles and any trade directory you are in.
- One page that states what you do, where you do it and since when, in sentences.
- A machine-readable copy of those same facts. The Organization schema generator writes the block; someone with access to the site pastes it in.
Who does what. The name, address and profile clean-up is entirely yours and needs no technical help at all. The block is generated free and then takes about five minutes for whoever built the site.
4. Write the page that answers the question you were missing from
Sent here by: missing, with the same competitor named instead.
This is the item that is real work, and it is the one worth doing. Your result names the exact questions you were absent from and who was offered in your place. Pick the question a customer would genuinely type, and put a page on your site that answers that question in your own words — what it is, where you are, what it costs, who it is not for.
If a competitor is being named on that question and you are not, the plainest available explanation is that they have a page answering it and you do not. The FAQ schema generator turns your question-and-answer pairs into a block a machine can read without inferring anything from your layout.
5. llms.txt, described as what it is
A file at the root of your site listing what you would like an assistant to read. No major AI vendor has confirmed reading it. It ships here as a checkbox and it is described as a checkbox — the generator writes one in a minute, it costs nothing, and it is last on this list for a reason. Do not expect it to move anything on its own, and be suspicious of anyone who tells you it will.
What none of this promises
Nobody controls what an assistant answers. Everything above is either the removal of a specific, identifiable blocker or the creation of a page that did not exist — both real, both yours to keep, neither a lever on what a model says. Any company in this category telling you it can make an AI name your brand is selling you something it does not own.
After a change, re-run the check on a different day and read the difference as one sample against another. Why that qualification matters is the previous page.
If this page did not answer it
Write to the one inbox. There is no chat widget, no ticket queue and no phone number — a message arrives with the person who wrote the thing you are asking about, and gets answered by hand. Send the domain you checked, what came back, and what you expected instead. That difference is the whole message.
Any question that arrives twice becomes a page here, so writing in is also how this section gets longer.
The surfaces that cannot be measured, the numbers refused, and what is not built.