SITE SCAN · PAGE SCAN · FREE · NO SIGNUP

Find the pages worth keeping

Enter your domain. Credify reads your pages and compares them with each other, then tells you which ones are near-copies, which are too short to stand on their own, and which carry text that appears nowhere else on your site. Every number names the pages behind it, so you can open them and check.

Site scanwhole site or one page · free

Reads your pages and compares them with each other, which is the one thing a single-page check cannot do: how much they repeat each other, how many are too short, and which titles and descriptions are reused or filled in from a template. Free scans read 25 pages, spread across the sitemap, in about 20–40 seconds. Pro’s deep scan reads up to 250.

25
pages read, free
Taken from across your whole sitemap. Up to 250 on Pro
8
words at a time
Two pages are compared on the runs of eight words they share
0
access to your accounts
It reads your public pages. Nothing else, ever

Why a normal audit misses this

A town page with 280 words and a tidy title passes every check on its own. Three hundred of them with only the town swapped is a different thing, and nothing about any single page shows it. The problem is not inside a page. It is between them, and it only appears when the pages are read side by side.

So that is what this does. It reads your pages together and tells you how much of each one appears on the others. What you get back is not a verdict: it is a list of your own URLs with a number against each, and you can open any two of them and see the shared sentences for yourself.

What it looks at

Seven things, each about the site as a whole rather than one page. The numbers underneath are the points where the scan starts saying something.

How much your pages repeat each other

On average, how much of one page’s wording turns up on another. Pages written separately sit near 2%.

Raised at 8%, 18% and 35%

Pages that are near-copies

How many of your pages have a twin somewhere on the site, and every pair that is 60% or more the same, with both addresses.

Raised when 20%, 35% or 60% of pages have a twin

Pages too short to stand alone

How many are under 300 words. Chinese, Japanese and Thai are counted by characters, so they are not called short by mistake.

Raised at 15%, 30% and 50% of pages

The same title on more than one page

Titles or descriptions used on several pages at once, with the text itself and the pages sharing it.

Serious once 30% of pages share one

Titles that differ by a word

Two titles or descriptions that are the same except for one word: what a template with a blank in it leaves behind.

Pairs sharing 75% of their words

Descriptions that repeat the title

A description that says the title again and finishes with a stock line like “Try it free now.” Written descriptions do not look like that.

Raised at 10%, 25% and 50% of pages

The keep list

Every page ranked by how much of its writing appears nowhere else on your site. The top carry something of their own. The bottom cost the least to merge away.

Guidance, adds nothing to the score

Also mentioned, adding nothing to the score: which language folder was compared if your site has several, pages whose text only appears once JavaScript has run, overview and hub pages that summarise the rest of the site on purpose, and any address in your sitemap that could not be opened.

When it is worth running

Traffic fell across the whole site at once
Many pages lost visits together, yet every page passes a normal audit. Whatever is shared between your pages is the place left to look.
Your pages come out of a template
Town and service pages, product and category combinations, one page per month or per year, comparison pages. See how much of each one is the template and which pairs are near-copies.
Your site is in more than one language
Pages are compared inside one language folder, so translations cannot hide repetition behind each other. Enter a URL inside the folder you want, such as yoursite.com/de/.
You have to decide what to merge
The keep list ranks your pages by how much of the writing is theirs alone. Legal and contact pages are marked, because nothing repeating a page does not make it valuable.
You just published a batch of pages
Run it once they are live and find out whether the new set reads as a set of copies, while changing them is still cheap.
Someone else has to see the evidence
Pro turns a scan into a PDF or a link you can send: the pairs, the short pages and the keep list, with every address, for them to check.

What happens when you press scan

  1. It finds your list of pages. Your sitemap, at /sitemap.xml or wherever your robots.txt says it lives. If that file points at other sitemap files, it follows them. You can also paste the address of the sitemap yourself.
  2. It picks pages from across the whole list. 25 pages on a free scan, up to 250 on Pro, taken two at a time from evenly spaced points between the first URL and the last. Sitemaps tend to list similar pages next to each other, so taking neighbours in pairs is what catches two near-copies. If your site has language folders, it stays inside one of them.
  3. It compares every page with every other. The menu, header, footer and sidebars come off each page first, so a shared layout is never counted as repeated writing. What is left is compared run of eight words by run of eight words.
  4. It tells you what it found, and where. Each finding names the pages it came from and what to do about it. Without an account you see the top three; a free account shows all of them; Pro adds a PDF and a link you can send to someone.

Looking at one page instead

Once the site scan has pointed at a page, the Page Scan tab checks that one URL against 33 things.

Who wrote it, and how you know4 checks
  • A named author on the page
  • Links out to at least two sources
  • A date, and how old it is
  • An About, Contact and Privacy page on the site
Can a search engine read it14 checks
  • Served over HTTPS
  • Not accidentally set to noindex
  • Not blocked in robots.txt
  • One main heading, not none and not five
  • Sub-headings in a sensible order
  • Works on a phone
  • A preview image and title for sharing
  • Description length
  • Title length
  • Says which URL is the original
  • Structured data for search results
  • Alt text on images
  • Links to your other pages
  • How long the server takes to answer
How the writing reads6 checks
  • Length, and whether it is too short
  • How hard it is to read
  • The same keyword forced in over and over
  • Blocks of text repeated from other pages
  • A description that repeats the title and ends in a stock line
  • A page that is mostly links
  • Wording that reads as AI-written (a note, never scored)
Can an AI assistant quote it4 checks
  • Questions as headings, and answers under them
  • Lists and tables that can be lifted out
  • A file telling AI crawlers what they may use
  • Concrete numbers worth quoting
How fast it loads for a visitor4 checks
  • Overall speed score out of 100
  • How long until the main thing appears
  • Whether the layout jumps while loading
  • How long the page is frozen to taps
Money and health pages1 check
  • Pages about health, money or law are held to a stricter standard: no named author becomes a serious finding rather than a small one

Which of your pages are the same page written again?

Nobody writes the same run of eight words twice by accident. Two people writing about the same subject, separately, share almost none of them. So when two of your pages share most of their eight-word runs, they were not written twice: they were filled in twice from the same template. That is the whole measurement, and it is why the result is something you can check by opening two tabs.

A free scan reads 25 of your pages, taken from across the whole sitemap rather than off the top, and compares each one with all the others. You get back: how much your pages repeat each other on average, every pair that is 60% or more the same with both addresses, how many pages are under 300 words, which titles and descriptions are used more than once, and a list of every page ranked by how much of its writing appears nowhere else. When one page needs a closer look, the E-E-A-T Checker goes through it on its own, and the history of the Panda update explains where judging a whole site at once came from.

Check every number yourself

A stranger telling you something about your own site deserves suspicion, so nothing here asks to be believed. Every finding names the pages it came from. Open the two addresses, read them side by side, and the shared sentences are either there or they are not. The two site owners who have acted on one of these scans both re-derived the numbers from their own pages first, and one of them got a slightly different answer for the same reason two people counting a crowd do. It changed nothing about which pages were copies.

Descriptions that were assembled, not written

The shape is easy to recognise once you have seen it. The description under a search result opens by saying the page title again, word for word, then finishes with a line like Try it free now. Get started today. One page like that is a wasted line of text. Two hundred pages like that is a pattern. The check is a plain string comparison: it does not guess at anyone's intent, and no model is involved.

Why it matters more than it looks: comparing the body text of a whole site is expensive. Comparing titles and descriptions is not — two short strings per address, checkable in one pass. So a description shape repeated across every page is the cheapest evidence there is that the pages were produced in bulk, and it says nothing about whether the writing underneath is original. That is the trap. You can write every word of every article yourself and still be judged on a metadata job that took a script four seconds.

It is almost never deliberate. It comes from filling in every description in one go, from a spreadsheet that glues two columns together, and from a site template that fills the same pattern for every page in the database: {{page_title}} | {{site_name}}. {{tagline}} Try it free now.. Each one produces a perfectly valid description that no person ever read. This tells you which of yours were written and which were assembled.

Before: flagged
Title
Best Running Shoes for Flat Feet | ShoeGuide
Meta description
Best Running Shoes for Flat Feet | ShoeGuide. Find your perfect pair instantly. Shop now.
The description says the title again word for word, drags the " | ShoeGuide" part of it along, and ends on a stock line. Three things are raised: the description looks assembled (serious, 14 points), the brand name has leaked into it (minor, 4 points), and if the two strings match outright, the description simply repeats the title (8 points).
After: clean
Title
Best Running Shoes for Flat Feet | ShoeGuide
Meta description
Nine stability shoes tested over 300 miles by a runner with fallen arches: which held their shape, which collapsed, and which to skip.
Same title, untouched. The description now says something the title does not: how the shoes were tested, by whom, and what the reader gets out of it. Nothing raised, and a reason to click.

Why this check exists: it is what happened to this site

In June 2026 this site lost most of its search traffic. The obvious suspect was the writing: the articles must be too alike. So we measured it. Across the whole site, pages shared 2% of their eight-word runs with each other, which is what separately written pages look like. The articles were fine.

The descriptions were not. Twenty-six pages carried one that opened with its own title and ended with the same stock line, because they had all come out of a single pass instead of being written one at a time. We rewrote all twenty-six by hand, then built the check so it would catch the shape on other sites. It is the check we wish we had been running.

We are not claiming that fixed it. The rewrite is recent, search moves slowly, and the honest answer is that we do not know. What we can say is what it cost to find out. If your descriptions were filled in as a batch, this is the fastest thing on the list: a writing job, not an engineering one, and it does not touch a line of your actual content.

What this cannot tell you

It cannot tell you why your traffic changed. It has no view of Google, it never asks for access to your Search Console or your analytics, and anything that sounded like a verdict from Google would be invented. Plenty of tools will give you one anyway. This is a measurement of your own pages against each other, and that is all it is.

It also cannot tell you a page is good. A page can be the only one of its kind on your site and still say what a hundred other sites already say — the comparison stops at your own domain. And unless your site is small, it reads a sample rather than every page, so it describes the shape of your site rather than auditing all of it. Where a number is too small a sample to mean anything, the scan says so instead of printing a percentage.

Frequently Asked Questions

Related tools

Check one page on its ownWhat Google changed, and whenThe helpful content update, explainedWhen Google started judging whole sitesWhat Pro adds

Scan your site now.

Free for 25 pages, no signup. See which of your pages share their wording, which are thin, and which are worth keeping.

Want the history behind all this? See the Google algorithm updates timeline, every core and spam update from 2011 to 2026, or the Helpful Content Update survival guide if a sitewide decline is what brought you here.