CrawlCheck

Guides · 2026-09-30 · By · 2 views

Staging site leak: how to find a staging URL in your live pages and close it

How a managed host's staging hostname gets into production page source, what a leaked staging copy costs, how to find it from outside and how to close it without breaking the workflow.

A staging leak is a production page whose source references the site's staging hostname on a managed host (cloudwaysapps, wpengine, kinsta, pantheonsite), usually because a search-and-replace missed a serialised option, page-builder JSON or a theme file. Search the raw page source, not the rendered page, for those domains. If one is found, the staging host is a discoverable, often indexable and often out-of-date copy of the site. Fix the reference with a serialisation-aware replace, then put the staging host behind authentication or X-Robots-Tag: noindex.

Check your own site free

About a minute. First scan free with your email, nothing installed.

Managed WordPress hosts give every site a staging copy on a hostname of their own: something.cloudwaysapps.com, something.wpengine.com, something.kinsta.cloud, something.pantheonsite.io. The copy is meant to be private. It stops being private the moment the production page's source references it, because a crawler that reads the production page now has the staging hostname, and a staging hostname that answers 200 is a second copy of the site with no canonical pointing home. This guide is how the reference gets into the page, how to find it from outside, what a leaked staging host does to the production site's standing, and how to close it without breaking the workflow that created it.

How the hostname ends up in production #

Almost always by a search-and-replace that missed. A site is built or restored on the staging host, then pushed to production with a database replace of the staging URL for the live one. Anything the replace did not reach keeps the old host: a serialised option a plugin stored, an image URL inside a page builder's JSON, a hard-coded asset path in a theme file, a CSS url() in a saved stylesheet. The page renders correctly because the staging host serves the asset, so nobody notices. You find it by reading the source rather than the render: search the homepage source for the four staging domains. A leak like this shows up on 0% of scanned sites.

What the leak costs #

Three things, in order of how often they happen. First, the staging copy is discoverable. Any crawler that follows the reference reaches a host that usually answers 200 with a full copy of the site, and unless that host sends noindex or requires authentication, it is a duplicate of every page with a different hostname. Second, the production page depends on it: an image or script loaded from the staging host breaks the day the host is deleted or the staging copy is refreshed. Third, and the one that reaches the business, a staging copy is often out of date, and an engine that reads a price, a phone number or an address from the staging host has read last month's.

Where the host hidesWhy the replace missed itHow to find it
serialised plugin optionstring length is stored beside the value; a naive replace corrupts it and is skippedsearch the options table for the host
page-builder JSON in post metaescaped slashes: https:\/\/search post meta with escaped and unescaped forms
theme file or child themenot in the database at allgrep the theme directory
saved CSS url()custom CSS stored as one blocksearch the customizer/CSS option
image srcseteach width is a separate URLread the rendered source, not the editor

Finding it from outside #

Fetch the production homepage and search the response body, not the browser's rendered DOM, for the four staging domains. Then fetch a few deep pages the same way; the homepage is the usual place, but a page-builder template used on inner pages can carry the reference where the homepage does not. If a host is found, fetch it directly and read three things: the status, whether an X-Robots-Tag: noindex or a meta robots tag is present, and whether it demands authentication. A staging host that answers 200 with no noindex and no login is the case to close first.

Closing it #

Fix the reference and fix the host, in that order. For the reference, run a serialisation-aware search-and-replace across the database for the staging hostname, then grep the theme directory, then re-read the rendered source to confirm nothing remains. For the host, most managed platforms offer a password or basic-auth toggle on staging environments; turn it on. If the platform does not, send X-Robots-Tag: noindex, nofollow from the staging host's server configuration, which a crawler honours without reading the page, and disallow everything in the staging host's own robots.txt as a second layer, remembering that robots.txt alone does not remove a URL that is already linked.

Reading the finding #

A CrawlCheck report names the staging host it found in the homepage source. That is all the homepage proves: it does not show whether the staging copy is indexable or how many references the site carries, and one reference is enough to make the host discoverable. Treat it as a pointer. The reference is the symptom, the host’s exposure is the risk, and the two are fixed separately. A site that has moved hosts recently and kept the old platform’s staging copy alive is the common case, and the old platform’s host is the one to check first, since nobody is logging in to it any more.

A staging leak is a foreign host serving your site. The nearer relative is your own host serving it twice: www and the bare domain both answering 200 with the same page and neither redirecting to the other. That shows up on 5.5% of scanned sites and is covered in full in the canonical URL guide, with the case where it was found on a site whose canonical tag was correct in its own post. The two findings share a fix philosophy: one host answers, every other host redirects to it.

Preventing the next one #

Make the search-and-replace part of the push, not a step someone remembers, and make it serialisation-aware. Keep staging behind authentication permanently rather than toggling it. And after every push, run the outside check above: it takes a minute, and a scan on crawlcheck.io runs the homepage half of it on every report. The share of scans carrying each finding is on the dataset page.

A staging or campaign subdomain often points at a hosting service through a CNAME. When the service is cancelled and the DNS record stays, the name points at nothing you control, and on some providers someone else can claim that target and serve pages on your subdomain. Check every CNAME in the zone against an active resource in your own account; an error page alone does not prove the target can be claimed. (Microsoft guidance)

Share of scanned sites with each finding in this articlewww and bare domain both serve the page5.5%Staging host linked from live pages0%Share of scanned sites with eachfinding in this articlewww and bare domain both serve the page5.5%Staging host linked from live pages0%
Read live from the same counters the dataset page uses, at the moment this page was served. Bars are scaled to the largest value shown, not to 100%.

Share this finding:LinkedInXEmailFollow Emmanuel on LinkedIn

Questions this post answers

What is a staging leak?

A production page whose source references the site's staging hostname on a managed host such as cloudwaysapps.com, wpengine.com, kinsta.cloud or pantheonsite.io. A crawler that reads the page learns the staging host, which is usually a full, indexable copy of the site.

How does the staging hostname get into a live page?

A search-and-replace that missed during the push to production: a serialised plugin option, page-builder JSON with escaped slashes, a theme file, saved custom CSS or an image srcset. The page renders because the staging host still serves the asset.

Is robots.txt on the staging host enough?

No. robots.txt stops crawling, not indexing of a URL that is already linked. Put the staging host behind authentication, or send X-Robots-Tag: noindex from its server configuration, and use robots.txt only as a second layer.

What does a CrawlCheck scan look for?

A URL on a managed host's staging domain in the homepage source the server sent. It reads what was delivered, not the rendered page, so a reference that renders fine is still found. Whether the staging copy itself is indexable is a separate check you run by fetching it.

What is the difference between a staging leak and a duplicate host?

A staging leak is a foreign host serving your site. A duplicate host is your own www and bare hostnames both answering 200 with the same page and neither redirecting. Both have the same fix philosophy: one host answers, the rest redirect.

Why is an out-of-date staging copy a business problem?

An engine that reads a price, phone number or address from the staging host has read an old version, and a duplicate with a different hostname competes with the real page for the same query.

Related findings

The main product

Found this on your own site? We fix it for $100.

Scan free to see where you stand. The fix is one site, every finding implemented and re-measured, with a sealed before and after.

Free browser extension

Check this on any site you open, without leaving the page.

CrawlCheck for Chrome grades every site you visit A–F and shows, from your own browser, what search engines and AI assistants are actually served. Free, no account.

Write for the library

Measured something yourself? Submit an article.

Send your finding, the data and the method. A person reads every submission and replies by email; accepted articles are published under your name with a link to your study. Free to submit. Articles are accepted on merit; no one can pay to have one published.

Submit an article

How it works · a reply within about a week

Spotted a mistake or a bug on this page? Report an error. Confirmed errors are fixed and logged on the corrections page.

Comments

Comments are read before they appear. Nothing is published automatically, and no account is needed. An AI agent with a published signing key can post without the queue: it signs the request and the signature is checked against its own domain (how). Those appear at once, marked verified.

Writing about this? Facts, live figures and marks — every number on that page is dated and traceable to a scan.

All guides · The dataset · How the dataset works