Myth vs reality
www vs non-www backlinks: which myths survive a check
Everyone has a theory about which host collects the links. Most theories come from reading a third-party tool, not from how Google processes your pages.
The common thread
Pick one host, redirect the other, and stop comparing tools
Link tools list hosts separately, so one host can look richer. That says something about the tool's view, not about Google's. If both hosts resolve, set a 301 from one to the other and a matching canonical, then measure the survivor.
- A 301 and a self-referencing canonical agree with each other
- Different tools disagree; a monthly index like ours is no exception
Seven claims, checked
Myth: the host that tools credit more should be canonical
Reality: a tool shows the links it found, per host. Choose the host you can serve consistently, redirect the other, and consolidate. Differences between tools reflect each crawler's reach.
Myth: a canonical tag moves link credit by itself
Reality: it is a hint, not a command. A 301 is the stronger signal when two URLs are really the same page.
Myth: robots.txt on both hosts is an error
Reality: robots.txt is read per host, so both hosts can answer. The safer pattern is to redirect one host to the other, so one file exists.
Myth: an hreflang alternate may differ from its canonical
Reality: alternates should point at canonical URLs and confirm each other. An alternate that redirects or canonicals elsewhere weakens the cluster.
Myth: a sitemap alone is enough for hreflang
Reality: sitemap hreflang is valid. What matters is picking one method per site and keeping it complete, not mixing methods halfway.
Myth: more links from one site equals more value
Reality: our refdomains call counts each site once because ten links from one domain rarely behave like ten votes.
Myth: every unusual crawler visit is a link problem
Reality: bot traffic in logs is mostly noise. Check user agents and rates in your server logs before you read it as a penalty or a plan.
How to test your own site in ten minutes
Request all four address variants, meaning http and https with and without www, and each one should land on a single final URL after one 301 and nothing more. Log it. One row each. Then compare what Ahrefs, Semrush and our index each report for the two hosts, and expect three different answers, because each crawler has its own reach.
Next, open Search Console and compare the Domain property with any URL-prefix properties, because a Domain property covers every host and protocol and has been available since 2019. Many people prefer it for that reason. Quick check.
Then call overview for the bare domain and note the figures beside your results. It is free. Nothing here asks you to take our word for it, because you can run all of it on a site you own, at no cost, in about ten minutes.
Planning a move? The migration runbook gives the order of operations, and the Sheets join shows how to merge host-level link counts into one view, so you can compare the old and the new host without guessing.
Questions we keep getting
- Will switching from www to non-www lose my links?
- Not if you 301 every old URL to its exact new twin and keep the redirects. Expect some wobble while Google processes the change. The links still point at the old host, and the redirect carries their signal across.
- Can I check backlinks for both hosts at once?
- Query the bare domain with
backlinks, then look at how the first response labels its rows. Splitting them by host in Sheets takes one formula, so it does not need a second paid call. - Does a self-referencing canonical hurt?
- No. It is a common and sensible pattern because it names the preferred URL for the page itself. Problems start when a canonical points to a page that redirects, or to a different page altogether.