What Are Redirects? Complete Guide

How redirects work

A redirect is an instruction that tells a client to request a different location. In a typical server-side redirect, the client requests URL A, receives a 3xx HTTP response containing a Location header, and then requests URL B. The final URL must return its own response, ideally 200 when it contains an accessible page.

Redirects operate at the URL and protocol level. They are not the same as canonical tags. A redirect moves the user and crawler, while a canonical tag identifies a preferred indexable URL without moving the visitor. A redirect is also different from forwarding a domain only at the registrar level, which may use masking or other behavior that search engines cannot interpret as a clean site move.

Common applications include HTTP to HTTPS enforcement, www or non-www normalization, domain migrations, changed slugs, content consolidation, temporary maintenance, campaign routing and replacement of deleted pages. Because every redirect creates another request, the destination should normally be the final intended URL rather than an intermediate address.

301 vs 302 vs 303 vs 307 vs 308

The correct status depends on permanence and request semantics. The distinctions below are defined by RFC 9110 and summarized in practical form by MDN.

Decision rule: choose 301 or 308 when the old URL should ultimately be replaced in search results. Choose 302 or 307 when the source URL should remain the long-term address. Use 303 when a completed non-GET request should lead to a separately retrievable resource.

For ordinary page migrations using GET requests, 301 is the most familiar choice. For APIs, checkout flows, uploads or forms, method preservation deserves explicit testing. MDN warns that redirects are especially consequential for non-GET requests because an unintended method change can alter or break the transaction.

How redirects affect SEO

Google describes permanent server-side redirects as a strong canonicalization signal. A permanent redirect tells Google that the destination should replace the source as the preferred URL. Temporary redirects generally signal that the source should remain the canonical URL, although search systems interpret signals collectively rather than mechanically.

Redirects do not guarantee equivalent rankings. A migration can change page content, internal links, rendering, crawl access, structured data, canonicals and user intent at the same time. The strongest transfer case is a direct mapping between substantially equivalent pages. Redirecting an old product page to an unrelated category or every retired URL to the homepage weakens the relationship and may be treated as a soft 404.

Google recommends server-side redirects where possible. JavaScript and meta refresh redirects can be interpreted, but they depend on rendering or delay and are weaker operational choices when server access is available. Redirects should not be used as cloaking, doorway routing or a method of showing crawlers a destination different from the one users receive.

The same technical consistency helps answer systems. Google states that its AI search features require no special redirect markup. A final page still needs to be crawlable, indexable and eligible for Search. Bing, Copilot and ChatGPT-linked discovery similarly benefit from stable final URLs, accessible evidence and consistent canonical references, but no redirect can guarantee citation or answer inclusion.

A safe URL migration sequence

1. Inventory the old site. Export indexable URLs, sitemaps, analytics landing pages, backlinks, canonicals, hreflang references and URLs seen in server logs.
2. Create a destination map. Assign every valuable old URL to the closest equivalent new URL. Mark URLs with no genuine replacement for 404 or 410 instead of forcing an irrelevant redirect.
3. Resolve normalization rules. Decide the final protocol, hostname, path case, trailing-slash behavior and parameter policy before writing rules.
4. Implement direct server-side redirects. Avoid old URL to interim URL to final URL mappings, even when previous migration rules already exist.
5. Test before release. Validate status, Location, final status, chain depth, content equivalence, encoded characters, query strings, subpaths, ports and non-GET requests where applicable.
6. Update owned references. Change internal links, canonicals, hreflang, structured-data URLs, XML sitemaps, advertising destinations and important profile links to the final URLs.
7. Launch and monitor. Submit the new sitemap, inspect representative URLs and compare crawler logs, index coverage, traffic, revenue and conversion behavior.

Google advises retaining site-move redirects for generally at least one year. Keeping them longer can help users and links that still reference old addresses. Do not remove them merely because the new pages have begun appearing in search.

Redirect diagnostic and decision framework

Test with an HTTP client or crawler that exposes every response, not only the page displayed by a browser. Browser-only checks can hide intermediate hops through cached redirects, service workers, cookies or automatic protocol upgrades. Compare anonymous and authenticated behavior where personalization is involved.

Chains, loops and soft 404s

A redirect chain occurs when one redirected URL points to another redirected URL. A loop occurs when the sequence returns to an earlier URL and never reaches a final resource. Chains add latency and create more points of failure. Google also notes that long chains can negatively affect crawling.

A 2025 Web Science study followed up to 10 hops across 11 million redirecting URIs. About half terminated successfully and half ended in errors. Only 0.06 percent exceeded 10 hops, but the research also identified 62,000 custom 404 URIs, nearly half of which behaved as soft 404s. The dataset is broader than SEO migrations alone, so it should be read as evidence of widespread redirect fragility, not as a ranking study.

Use one hop as the operating target: source to final destination. This does not mean every two-hop chain produces a measurable ranking loss. The practical case for cleanup is cumulative: fewer requests, lower latency, clearer canonical signals, more efficient crawling and easier incident diagnosis.

A soft 404 returns a success or redirect-like response while the destination effectively says that no relevant content exists. Homepage fallbacks and thin search-result destinations are frequent causes. If no closely corresponding page exists, an accurate 404 or 410 can be better than a misleading redirect.

Advanced auditing, KPIs and crawl prioritization

Prioritize redirects using business and discovery signals rather than fixing an arbitrary URL list. Start with URLs receiving organic landings, conversions, external links, recurring bot requests or internal links. Then address widespread templates and rules that generate defects at scale.

• Coverage: percentage of mapped legacy URLs producing the intended status and destination.
• Chain depth: average and maximum hops before the final response.
• Final health: share ending in 200, 4xx, 5xx or another redirect.
• Signal consistency: final URLs used in internal links, canonicals, hreflang and sitemaps.
• Crawler behavior: requests to old URLs, final URLs and error endpoints in server logs.
• Search recovery: indexed destination count, clicks, impressions and query coverage compared with the pre-migration baseline.
• Business continuity: transactions, leads, revenue and conversion rate by migrated landing-page group.

Segment results by page type, directory, country, device and migration wave. A healthy site-wide average can conceal a broken language directory or high-value product group. Log-file analysis is particularly useful because it shows which legacy URLs crawlers continue requesting and whether crawl activity reaches the intended final pages.

HTTP Archive offers longitudinal crawl datasets that can support broader analysis of root-document and redirect behavior. Such datasets are useful for trends, but they do not replace first-party logs because crawler refusals and misleading status responses can distort observed outcomes.

Redirects in content consolidation and site architecture

Redirects can support content consolidation when several pages satisfy substantially the same intent. Select the strongest destination, merge unique useful material, redirect superseded URLs and update internal links. Preserve intent boundaries: an informational guide, product page and local service page should not be merged simply because they share a keyword.

For hub-and-spoke architecture, redirect retired spokes only to a hub that genuinely answers the same need. Otherwise, keep a useful replacement or return 404 or 410. Update contextual links so search engines and answer systems discover the consolidated resource directly rather than through historical URLs.

Redirect cleanup also supports content decay remediation. When refreshing a declining page, retain its URL if the intent remains stable. Redirect only when consolidation or a permanent URL change is necessary. Constantly changing addresses creates avoidable dependencies and weakens the durability of external citations.

After a migration, reclaim high-value links by asking owners of important backlinks to update their destination. Use link-intersect analysis, unlinked brand mentions and digital PR to earn links directly to durable final resources. Original datasets, statistics pages, comparison assets and expert contributions create stronger natural link demand than redirecting expired promotional pages into generic commercial content.

Implementation choices and buying criteria

Redirects can be configured in the origin web server, application framework, content management system or edge network. Origin rules offer control but may require deployment access. CDN or edge rules can process large lists before a request reaches the application. CMS plugins are convenient for smaller editorial sites but can introduce database overhead or conflicts with server rules.

Choose a solution based on rule volume, bulk import and export, regex safety, query-string handling, path-suffix preservation, testing, version control, rollback, access permissions and audit logs. Cloudflare, for example, documents options for status code, query-string preservation, subpath matching and path-suffix preservation. Features vary by platform and plan.

For a few unambiguous changes, a simple reviewed configuration may be enough. For a multi-domain or enterprise migration, require a source-to-destination mapping, staging validation, automated regression tests, monitoring and an immediate rollback path. Ownership should be explicit across SEO, engineering, infrastructure, analytics and product teams.

Be cautious with wildcard and regular-expression rules. A compact rule can redirect thousands of URLs incorrectly, expose open-redirect vulnerabilities or override specific mappings. Security review is essential when a destination can be influenced by user input. Never use deceptive redirects that send users and crawlers to materially different content.

What is proven, what is consensus, and what is uncertain

Proven by standards and official documentation: HTTP status codes have distinct semantics; 307 and 308 preserve the request method; redirects create additional requests; permanent redirects are a Google canonicalization signal; and irrelevant destinations can be interpreted as soft 404s.

Strong practitioner consensus: direct one-hop mappings, updated internal references, retained migration redirects and automated testing reduce risk. Auditing logs as well as crawls usually reveals issues that manual browser checks miss.

Still uncertain or context dependent: there is no universal ranking-loss percentage for each extra hop, nor a fixed time in which every migrated URL will be replaced in search. Recovery depends on site size, crawl frequency, mapping quality, content equivalence, internal linking and other changes made during the migration. Community reports of gains after chain cleanup are useful observations, but they do not establish that redirect cleanup alone caused the gains.