Redirect logs show what happens after a URL redirect rule is published: which retired URLs receive requests, whether crawlers and visitors pass through unnecessary hops, which destinations fail, and where valuable traffic may be leaking. This guide explains how to turn server, CDN, and analytics data into a repeatable redirect health report, with practical filters, estimation formulas, investigation examples, and a prioritization method you can revisit after migrations or on a regular schedule.
Overview
A redirect log analysis combines request logs with analytics and URL mapping data. The goal is not simply to count redirects. It is to identify redirects that deserve action, such as:
- A redirect chain where an old URL points to a second redirect before reaching the final page.
- A retired URL that continues receiving substantial requests from backlinks, bookmarks, email campaigns, or internal links.
- A redirect that ends at a missing, blocked, irrelevant, or unintended destination.
- A crawler repeatedly requesting obsolete URL patterns that could be consolidated or removed from the site.
- A campaign link whose query string is lost, duplicated, or routed to the wrong landing page.
- A rule that creates a loop, a growing number of hops, or inconsistent behavior between HTTP and HTTPS or between hostnames.
Logs are especially useful because they record requests that may not appear in standard analytics. Bots, prefetchers, API clients, privacy-conscious visitors, and failed page loads can all be absent or grouped differently in analytics platforms. At the same time, a log entry does not prove that a human saw a page. Treat the user agent, referrer, cookies, and analytics session as signals rather than perfect identities.
For status-code context, review 301, 302, and 307 redirects. A permanent migration redirect and a temporary campaign redirect may both appear in access logs, but they should be evaluated against different expectations.
How to estimate redirect cost and priority
You can estimate the operational value of a redirect fix with a simple scoring model. The model is not a search-engine formula; it is a way to make decisions consistently when a site has thousands of redirect requests.
For each source URL, record:
- Request volume: the number of requests during a defined period.
- Human-likelihood share: the proportion that appears to come from visitors rather than known crawlers, monitoring tools, or internal services.
- Business value: a rating based on conversions, revenue, lead quality, campaign importance, or strategic content.
- Redirect severity: a rating for chains, loops, errors, wrong destinations, or query-string loss.
- Fix effort: the estimated effort to update the source link, redirect rule, destination, or campaign asset.
A practical priority estimate is:
Priority = request volume × human-likelihood share × business value × redirect severity ÷ fix effort
Use small, consistent scales—for example, ratings from 1 to 5—and document what each rating means. A high-volume old URL with a relevant final destination may need less urgent work than a lower-volume campaign URL sending visitors to an unrelated page. The point is to compare like with like, not to create a universal benchmark.
For crawl-efficiency reviews, use a separate view. Count requests from identified crawlers by source URL, status code, and number of hops. Repeated crawler requests to URLs that should no longer exist may indicate outdated internal links, an incomplete migration, an overly broad redirect rule, or external links that continue to circulate. Avoid assuming every automated request is a search crawler; verify user-agent strings against other request characteristics and, where appropriate, your platform’s bot-verification process.
Inputs and assumptions
Before filtering data, define the scope. A useful minimum is one complete reporting period, such as a calendar month, plus a comparison period before or after a known change. Export raw logs where possible so you can reprocess them when your classifications improve.
Log fields to preserve
| Field | Why it matters |
|---|---|
| Timestamp | Shows trends, campaign timing, and bursts after a migration. |
| Request method | Separates ordinary GET requests from other traffic that may require different redirect handling. |
| Host and requested URL | Reveals HTTP-to-HTTPS, www-to-non-www, old-domain, path, and query-string patterns. |
| Status code | Identifies redirects, successful destinations, client errors, and server errors. |
| Location header | Shows where a redirect sends the requester, when captured by the server or CDN. |
| Referer | Helps distinguish internal links, external backlinks, search referrals, and direct or hidden sources. |
| User agent | Provides a starting point for classifying browsers, crawlers, previews, monitoring tools, and suspicious automation. |
| Response time and bytes | Helps identify slow redirect handling or unusually large responses. |
| Request ID or cache status | Connects CDN events with origin events and helps prevent double-counting. |
Normalize URLs before grouping them. Decide whether case, trailing slashes, default ports, encoded characters, and query parameters should be treated as equivalent. Preserve campaign parameters separately: removing all query strings may hide a tracking problem. For guidance on retaining attribution during redirects, see how to handle query strings without losing tracking data.
Also establish exclusions. Internal health checks, uptime monitors, static assets, administrative paths, and known preview services can overwhelm a report without representing page traffic. Keep the excluded records in a separate dataset rather than deleting them permanently.
Worked examples
Finding a redirect chain
Group requests by source URL and follow each recorded destination. A simplified query pattern might look like this:
SELECT source_url,
COUNT(*) AS requests,
COUNT(DISTINCT redirect_target) AS targets,
AVG(hop_count) AS average_hops
FROM redirect_requests
WHERE occurred_at >= :period_start
AND occurred_at < :period_end
GROUP BY source_url
HAVING MAX(hop_count) > 1
ORDER BY requests DESC;
This is illustrative SQL, not a platform-specific schema. Confirm the chain with a redirect checker or command-line request that does not automatically follow every redirect. If /old goes to /intermediate, which then goes to /current, update the rule so the source resolves directly to the final relevant URL when the change is permanent and safe to consolidate. Validate that query strings, protocol, hostname, and trailing-slash behavior remain correct.
Finding broken destinations and traffic loss
Join redirect records to final response records, or test a sampled set of high-volume targets. Flag redirects where the final response is a client or server error, where the content is clearly unrelated, or where the destination requires an unexpected login. Then compare the source URL’s referrers with the destination’s landing-page activity. A source with external referrers but no corresponding destination sessions deserves review, especially if the source was used in a campaign or has valuable backlinks.
Do not automatically redirect every broken URL to the homepage. A relevant replacement is usually more useful; when no equivalent exists, consider whether a clear not-found response is more accurate than a misleading redirect. Use the redirect audit checklist to combine log findings with crawl and link data.
Estimating the effect of a fix
Suppose one retired URL receives 4,000 requests in a month. After excluding automated checks and other clearly non-human traffic, you estimate that 35% may represent visitors. If the URL currently uses two redirects and ends at a relevant page, the immediate value may be reduced latency and simpler crawling rather than recovered traffic. If it instead ends at a missing page, the same volume indicates a more urgent destination problem.
Record the baseline: requests, status codes, hop count, final response, referrer categories, and campaign parameters. After the fix, compare the same measures. A successful change should normally reduce unnecessary hops or errors without causing a rise in lost tracking parameters, wrong-page landings, or new redirect loops.
When to recalculate
Run a focused redirect log analysis after a domain change, HTTPS implementation, platform migration, URL restructuring, campaign launch, or major redirect-rule update. A broader monthly review is useful for sites that publish frequently or maintain many campaign links. Smaller sites can use a quarterly schedule, then increase the frequency during a migration.
Each recurring report should include:
- Total redirect requests and the share of requests covered by the reporting period.
- Top source URLs by requests, human-likelihood estimate, and business value.
- Number of chains, loops, error-ending redirects, and wrong-destination cases.
- Top referrer domains and campaign parameters associated with retired URLs.
- Crawler requests to obsolete or repeatedly redirected URLs.
- Changes in average hops, redirect errors, and destination response time since the previous report.
- Owners, due dates, and verification status for each corrective action.
Recalculate the estimates when traffic mix, campaign tagging, hosting, CDN configuration, bot-filtering rules, or redirect mappings change. Keep the assumptions beside the results so a future comparison does not mistake a classification change for a traffic trend. For migration planning, pair the report with a URL redirect mapping template. For security-sensitive rules, review potential open redirect vulnerabilities before deployment.
Finish each review with a short action list: fix the highest-value wrong destinations first, collapse confirmed chains, repair internal links that keep generating obsolete requests, preserve required campaign parameters, and retest the affected URLs from representative devices and referrers. A redirect log becomes valuable when it changes the next rule, link, or destination—not merely when it produces another report.