Discussion
  • Technical SEO

On a domain with ten years of layered redirects, how many hops have you seen Google tolerate before signals weakened, and did flattening chains pay off?

Published 1 reply

Discussion starter: a technical SEO question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.

Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.

1 reply

  • admin
    Admin
    Original poster

    Google documents a hard limit, not a decay curve. Its crawler documentation says: “By default, Google’s crawlers follow up to 10 redirect hops. However, specific products’ crawlers may have different limits.” A chain that runs past the limit, loops or contains a malformed URL fails, and Search Console reports it as a “Redirect error”. Below that limit, Google’s current documentation gives no hop count at which signals start to weaken, and no controlled public test exists that I can point to that isolates one.

    So on a domain with ten years of layered redirects, the useful question is less “how many hops is too many” and more “which chains are broken, weakened or slow, and which of those carry links or traffic”. Flattening pays off where it repairs those. Where chains already reach a relevant page in two or three permanent hops, the gains are crawl efficiency and speed, and I would not promise a ranking lift from them.

    What Google has documented, and where the per-hop myth came from

    • The hop limit. Ten by default, with product-specific exceptions, according to the HTTP status code documentation. The same page says Google uses a 301 as a strong signal and a 302 as a weak signal that the target should be canonical, and treats 308 like 301 and 307 like 302.
    • Crawl cost. The crawl budget guide says: “Avoid long redirect chains, which have a negative effect on crawling.” It does not define “long”.
    • Failure states. The Page indexing report documentation lists four kinds of “Redirect error”: a chain that was too long, a loop, a redirect URL that eventually exceeded the maximum URL length, and a bad or empty URL in the chain. “Page with redirect” is not an error. It means the URL redirects and will not be indexed itself, while the target “might or might not be indexed”.
    • PageRank. In 2016, Google’s Gary Illyes said 30x redirects no longer lose PageRank. The same article recounts Matt Cutts’s 2013 position that the PageRank dissipating through a 301 was identical to the amount dissipating through a link. That earlier position is the likely root of the idea that every hop leaks a slice of equity.

    Two gaps are worth stating plainly. The 2016 statement was about redirects, not chains: Google has not quantified what happens across five or eight hops. And “a negative effect on crawling” is about crawl efficiency, not ranking. A precise per-hop loss figure is a model someone built, not a measurement Google published.

    How ten years of migrations stack hops

    Long chains accumulate rather than get designed. Each migration adds rules that point old URLs at the URLs of the day, and nobody rewrites the older rules when those URLs move again. Normalisation rules (HTTP to HTTPS, bare domain to www, trailing slash, lowercase) can run in different layers, such as the CDN, the load balancer, the web server and the CMS, and each layer that fires on its own adds a hop.

    Hypothetical 2016 backlink: http://example.com/Products/Widget.aspx
    ↓
    CDN rule: HTTP to HTTPS (301)
    ↓
    Web server rule: bare domain to www (301)
    ↓
    2018 migration rule: /Products/Widget.aspx to /products/widget (301)
    ↓
    2021 replatform: /products/widget to /shop/widgets/blue-widget (302, never switched to 301)
    ↓
    2024 CMS setting: add a trailing slash (301)
    ↓
    Final: https://www.example.com/shop/widgets/blue-widget/ (200)

    That is five hops, inside Google’s limit, and nothing in the documentation says it fails. The risks sit elsewhere. The 302 leg weakens the signal about which URL should be canonical. Every Googlebot request for the old URL costs five fetches before it reaches content. And if someone deletes or reuses /products/widget, the chain breaks for every link upstream of that point.

    Where old chains lose value

    Chain condition What Google gets How to spot it Priority
    Ends in a 404 or 410 No destination, so links into the chain point at nothing Google will index Final status code in a chain crawl High if the first URL has links or traffic
    Loops, runs past the hop limit or hits a malformed URL A “Redirect error” Page indexing report; chain crawl High
    Ends at the homepage or a loosely related page Possibly treated as a soft 404 rather than a move Many unrelated source URLs sharing one broad final URL High for URLs with links
    Contains a 302 or 307 leg A weaker signal about which URL should be canonical Status code of each hop Medium, and cheap to fix
    Final URL is noindexed, canonicalised elsewhere or redirected again later Signals consolidate onto a page Google will not index Indexability check on the final URL Medium
    Contains a meta refresh or JavaScript hop left by an old CMS A hop that relies on Google parsing or rendering the page Chain crawl with JavaScript rendering switched on Medium
    Three to five clean 301s ending at a relevant 200 page Works, at a crawl and latency cost Log requests to intermediate URLs Low to medium; rises with scale

    Did flattening pay off? What the evidence supports

    Public evidence here is thin, and before-and-after comparisons from migrations cannot isolate hop count because too much else changes at the same time. What the documentation and the mechanism support is narrower:

    • Repairing broken chains is recovery, and it can show in rankings. If a 2016 link feeds a chain that now ends in a 404, fixing it restores a path that was lost entirely. That is the case where destination pages can reasonably be expected to gain, and the gain is measurable.
    • Shortening healthy chains mostly changes crawling and speed. Each hop is another request for Googlebot and another round trip for a user’s browser. On a site where Googlebot still requests hundreds of thousands of legacy URLs, cutting hops frees real crawl activity. On a 2,000-page brochure site, it is housekeeping.
    • Swapping 302 legs for 301s is cheap and removes an ambiguity about which URL should be canonical.
    • Internal links that point at redirects are the ones Googlebot meets every day, unlike a 2016 backlink. Updating templates, navigation, canonicals, hreflang and XML sitemaps to final URLs is where I would start.

    An audit plan for a decade of redirects

    1. Build the URL inventory from every era. Pull old XML sitemaps, backlink exports of linked URLs, analytics landing pages as far back as you have them, server or CDN logs of Googlebot requests that returned a 3xx, and URL lists from past migrations.
    2. Crawl the inventory in list mode with a tool that records every hop, for example Screaming Frog’s redirect chains report. Keep each hop’s status code, the final status, and whether the final URL is indexable (200, self-canonical, no noindex).
    3. Classify by health and by value. Take health from the table above and value from referring domains, recent Googlebot requests and traffic. The broken, high-value group is the project. The healthy, low-value group can wait.
    4. Collapse rules to one hop. Map every legacy URL, including intermediate URLs from earlier eras, straight to its final destination, combining protocol, host and path changes in a single rule. Keep the map in one layer, the one that runs first, so other layers do not stack hops on top of it.
    5. Point everything you control at final URLs: internal links, canonicals, hreflang and XML sitemaps.
    6. Re-crawl and confirm one hop, 301 or 308, ending on an indexable 200 page. Keep redirects for any URL that still has links or traffic; they cost little to maintain.

    Watch for this:

    Rule order is where flattening projects break things. A new direct rule placed below a broad pattern rule, such as a regex that rewrites a whole legacy folder, can be shadowed and never fire. Test the full URL inventory after deployment, not a sample of the rules you just wrote.

    How to tell whether it worked

    Choose metrics that match what flattening can change, and give Googlebot time to meet the new rules. Legacy URLs can go weeks or months between Googlebot requests, so a fixed calendar window is a poor proxy. Watch the logs until most of the high-value legacy URLs have been requested again, and measure from there.

    Metric Where What success looks like
    Googlebot requests answered with a 3xx Logs; the Crawl stats report’s breakdown by response Falls as internal links are updated
    Hops per legacy request Logs, or a re-crawl of the inventory One
    “Redirect error” URLs Page indexing report Zero for priority URLs
    Destination pages whose inbound chains were broken, compared with destinations whose chains were already healthy Performance report, filtered to each group The repaired group gains relative to the healthy group
    Response time for legacy journeys Synthetic checks from a few locations Lower, by at least one round trip per removed hop (more where a hop changes host)

    The group comparison is what turns this into evidence. If only the destinations with repaired chains gain, repair was the driver, not hop count. If both groups move together, something else changed. Add a Search Console custom annotation on the release date, and note any core update or large release in the same window, because either can move both groups.

    Where I would land

    If this were my domain, I would not pick a hop threshold and chase it. I would fix every chain that ends badly, loops, passes through a 302 or lands on an irrelevant page, starting with URLs that have external links. I would collapse the rest to one hop because a single-layer map is easier to maintain, and update internal links first because those are the redirects Googlebot meets every day. Then I would judge the project on crawl activity and on the pages whose inbound chains were broken, not on sitewide rankings.

    Need help with this on your own site?

    Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.

    See SEO services

Add a reply

Replies are open to approved members. Log in or apply to join.