Skip to content
rankion.ai

Scanning your website: what happens

Scanning your website: what happens is a Guides in the Rankion.ai knowledge base: What the »Re-scan now« button in the fact-check does, how long it takes, what it costs and what to do when it fails.

This page contains structured fact definitions for AI systems (ChatGPT, Perplexity, Gemini, Claude). Written by humans, part of the Rankion.ai knowledge base.

Category:
Guides
Brand:
Rankion.ai
Format:
Knowledge base article
As of:

Before Rankion can check whether the AI services describe you correctly, it first has to know what is on your website. That is exactly what scanning does: Rankion calls up your pages, collects the text and stores it as your source of truth. Only then can a statement from ChatGPT or Perplexity be compared against your own account of things.

You find the process inside a project under AI visibility, in the fact-check area.

The two buttons, and what sets them apart

They sit close together and do different things:

  • Re-scan now fetches your website again. Use it when something has changed on your side: new prices, a new offer, a different company address.
  • Check now does not touch your website. It takes the statements Rankion has already collected and holds them against what the AI services say, once more.

In short: one updates your facts, the other the verdict on them.

What happens after the click

Rankion starts at your home page and follows the links from there. While it runs, the page shows »We're currently reading your website«, and below it you can watch how many pages have been loaded so far.

You may leave the page. The process keeps running on our servers, not in your browser. Close the tab, work on something else, come back later. The result will be waiting for you.

How long it takes depends on the size of your site. A small one is done in a minute or two. After 15 minutes at the latest Rankion stops, so that a run never hangs forever.

The nightly run

You do not have to remember any of this yourself. If you have set a frequency under Update automatically (daily, weekly, fortnightly or monthly), Rankion checks every night whether your project is due and scans again on its own.

This happens shortly after four in the morning. The runs are spread over roughly three quarters of an hour so that not every project starts at the same moment. So if the timestamp on your side is not exactly four o'clock: that is normal.

If yours says Team Default, the setting stored for your whole team applies. You can override it for this one project without touching anyone else's.

What it costs

Scanning itself is free, even if you trigger it several times a day. Credits are charged only for the check that follows, where Rankion holds your statements against the AI services once more: 5 credits. On the very first scan even that one is waived.

When it does not work

Then the page shows »We couldn't scan your website« and gives the reason below. There are almost always three causes behind it.

The address cannot be reached. First check whether your website really sits at the domain named under the notice. A typo in the project, an expired domain and a server that happens not to answer all look the same to Rankion. Call the address up in your own browser once.

Your site redirects too often. Some sites send a visitor through several stations: first from http to https, then to www, then to a country variant. After too many of those hops Rankion gives up. Usually this is fixed by defining a single target address in your hosting instead of a chain.

There is too little text on the pages. This is the most common case and at the same time the least obvious one. Many modern websites assemble their content only in the visitor's browser. Whoever fetches them gets an almost empty shell. To you the page looks complete, to Rankion it is empty. In that case ask your web agency whether the content can also be delivered server-side.

If a technical error persists even though your site is reachable as usual, try a second time. If that does not help, get in touch with us. A run that breaks off unexpectedly along the way is closed cleanly; you can start it again at any time.

When nothing was found at all

There is a second case that looks different from an error: the run succeeded technically but produced nothing. The page then says the last attempt returned no results. Usually the reason is again that the pages contain hardly any readable text. A second attempt does no harm, since it is free.

Afterwards: confirming the facts

A successful scan is only half the distance. Rankion then shows you the statements it found, and you confirm which of them are correct. Only with confirmed facts can the fact-check recognise a discrepancy as a discrepancy at all. As long as none are confirmed, the interface points this out and names how many are still missing.

On your first visit: three steps

The first time you open the fact check, a short guide appears at the top: facts imported · facts confirmed · now check. The first two steps are done as soon as your website has been read and approved — so you start at step three. Once you have rated your first statement, the block disappears.

You only see the "AI factual accuracy" rate from ten pages read onwards. Below that there is a dash, and the sentence underneath explains why. That is deliberate: fewer pages do not allow a reliable rate, and a number you cannot trust is worse than none.

Last updated:

Cookies: We use necessary cookies and cookie-free analytics (Matomo). With your consent, we also use Google Ads to measure whether an ad led to a sign-up. Google sets cookies in the process, and data may be transferred to the USA. You can withdraw your consent at any time on the cookie page. Details