How much link equity does a 301 redirect actually pass?

Ask how much link equity a 301 redirect passes and every answer sounds precise: 90%, 99%, “almost all,” “no loss at all.” None of them is a measurement. Nobody outside Google can observe how PageRank moves through Google’s index, which means every figure in this piece, including the ones attributed to Google itself, is a public statement of position rather than a number anyone verified from the outside.

What Google has actually said, and never said#

Google’s own position has moved twice on the record, and neither move came with a percentage. In a February 2013 video, Matt Cutts said “the amount of PageRank that dissipates through a 301 is currently identical to the amount of PageRank that dissipates through a link” - a statement that a redirect loses no more than an ordinary hyperlink does, without saying how much a link loses either. In July 2016, Gary Illyes went further in a tweet: “30x redirects don’t lose PageRank anymore,” dropping even the comparison to a regular link. John Mueller has since added the caveats that matter in practice: redirecting to a topically unrelated page risks Google treating it as a soft 404, and a chain of several redirects behaves worse than one clean hop, even though no single hop is said to have a fixed cost.

Put together, this is thirteen years of consistent Google messaging in one direction - “close to full value, and increasingly no measurable loss” - and zero years of Google naming a number. That gap is not an oversight. PageRank’s internal calculation is one of the things Google has never published in disclosed detail for any purpose, redirects included, so a percentage was never on the table to begin with.

The industry’s own estimates: a number and a shrug#

The one specific figure that circulates widely, “301 redirects pass 90-99% of link equity,” comes from Moz’s own redirection guide. Moz states the range as guidance, not as a finding from a disclosed crawl or study, and does not publish the methodology behind it. The figure also predates Illyes’ 2016 statement by several years, which is part of why it keeps getting cited alongside claims that contradict it: the 90-99% range and “redirects don’t lose PageRank anymore” cannot both be describing the same current reality, and only one of the two sources says how it arrived at its position (neither, in fact - Moz gives a range with no visible study, and Google gives a tweet with no visible study either).

Semrush, by contrast, declines to give a number at all. Its own 301 redirect guide says search engines “typically transfer almost all of the old page’s ranking power to the new page,” choosing the same kind of unquantified language Google itself uses rather than repeating Moz’s range or proposing a figure of its own. That is itself informative: a tool vendor with its own crawl data and a direct commercial reason to publish a precise number chose not to.

Why we care#

Every number in this piece describes a public position, not a measurement. No source cited here, Google included, has published a disclosed methodology for isolating how much authority a 301 redirect passes versus every other signal that changes when a URL moves. Treat “90-99%,” “almost all,” and “no loss anymore” as three ways of saying the same thing with different confidence, not as three competing facts to average.

The actionable guidance sits underneath the disagreement, not inside it. Every source agrees, without needing a number, that a 301 to a genuinely equivalent page passes most or all of a page’s authority, and that the real risk is a redirect to an unrelated destination, which Google has said on the record it may not credit at all. That distinction is worth more than any percentage in this piece.

A single redirect’s estimated pass-through rate is a separate question from what a real redirect chain does to traffic. This report covers the per-hop estimate; a documented, measured outcome from one migration belongs in our case studies rather than in a comparison of publicly stated positions like this one, and the two should never be quoted as if they answer the same question.