Do many crawled addresses need a redirect?
It checks how much of the crawled URL set travelled through a redirect.
What Check Engine measures
The check selects crawled pages with a nonempty saved redirect chain. Their count must exceed the larger of 5 pages or a 0.25 share of all crawled pages before the finding appears. The comparison is strict, so equality does not trigger it. This check uses redirect volume as the measured signal. It does not add network timings to calculate a total delay. For example, exactly 5 redirected pages cannot exceed a threshold that is at least 5 pages.
The full crawl's request history supplies the counts, including pages outside the successful content set. It is distinct from the check for a particular loop or a chain containing several hops.
How to read the result
The finding is low severity. A redirected address can be a correct legacy route, and a large share can reflect how the crawl was seeded rather than proof that every internal link is wrong.
Why it matters
Many redirects can indicate that the site still advertises old addresses. The useful follow-up is to identify the sources of those requests, rather than removing valid redirects that protect visitors arriving through older links.
How to fix it
Update internal links and sitemaps to the final intended addresses where you control them. Keep redirects that support old external links. Check for loops and unnecessary intermediate moves. Measure response times separately when you need to quantify delay.
What we found in field audits
Problems in Speed & Core Web Vitals were found on 13 of the 14 sites we audited. These are area totals. Read the field-audit method.
Method note
backend/seo_engine/checks/performance.py, redirect_cost, revision 0bb8e8d3.
Common questions
Can exactly five redirected pages trigger the redirect-volume finding?
No. The count must exceed a threshold of at least 5 pages. Larger crawls can raise the limit through the share rule.
Does the redirect-cost check add up network delay?
No. It counts the saved redirect chain entries rather than adding network timings. Measure response time separately to quantify delay.
How serious is a high volume of redirected crawl requests?
It has low severity. A high count can reflect the crawl's starting addresses as well as outdated links.