Three in ten redirects
go nowhere new

The question the rot report could not answer

When we measured how much of the Australian web is broken, dead links were only part of what we saw. A steady 8.9% of every link we followed never answered directly — it answered with a redirect. That report had no room to ask the obvious follow-up: a redirect to where, and why?

A redirect is not a fault. Moving a page and redirecting the old address is exactly right, and most of the redirects here are doing precisely that. But a meaningful minority take you nowhere new at all: they correct the scheme, add or drop a www, or fix a trailing slash — work the link itself could have done.

What the redirects were for

Permanent or temporary

A 301 tells search engines the move is permanent and the destination should inherit the original's standing. A 302 says the opposite — keep indexing the old address, this is temporary. Sites that use 302 for a permanent move quietly hold their own ranking signals at the old URL.

What a housekeeping redirect actually costs

Individually, almost nothing — a few hundred milliseconds. At the scale of a whole site it adds up in four ways:

• A round trip per link. Every visitor who follows one waits for a request that exists only to be told to make another request. On a slow mobile connection that is the difference between fast and merely acceptable.
• Crawl budget. Search engines spend a fetch on the redirect and another on the destination. On a large site that is a meaningful share of what a crawler was willing to spend on you.
• Chains. Housekeeping redirects compose. An http:// link to the wrong hostname variant can take two hops before anything is served, and chains lose more at every step.
• They hide real breakage. A link checker that follows redirects silently reports all of these as working, so nobody ever finds out the links are wrong.

The fix is not to remove the redirects — they are the safety net, and they should stay. It is to stop relying on them from your own pages: link to the canonical address directly, so the safety net only catches links you do not control.

Method

Where the redirects came from

The same scan behind our public sector link rot census — every .gov.au, .edu.au and .org.au domain in Tranco list K9QPW of 1 September 2026, up to 15 links per homepage. One scan, two reports, so the servers involved were asked once.

How a redirect is classified

We compare the link's own URL with the Location the server answered with, looking only at scheme, host and path. Query strings and fragments are ignored — they vary for tracking reasons that say nothing about the redirect. Hostnames are compared with and without a leading www., so a www change is recognised as such rather than counted as a move to another site.

What "avoidable" means

A scheme upgrade, a www change, both together, or a trailing slash. In each case the destination is the same page on the same site, reachable by writing the link differently. It does not mean the redirect should be switched off.

What we did not follow

We record the first hop only. Our checker deliberately does not follow redirects, so a chain is counted once, by its first response. Real chain depth is therefore undercounted here.

Known limits

One request per link, at one moment, from one country. Servers that vary their response by client or region will be misread. Every redirect seen in this scan was classified.

Data

The full aggregate is published as au-redirect-report.json under a CC BY 4.0 licence.

← All reports