Link juice and link equity: what actually passes through a link

Google's documentation makes two statements about PageRank on two pages carrying the same last-updated date. The ranking-systems guide says PageRank "continues to be part of our core ranking systems". The SEO starter guide files PageRank under "Things we believe you shouldn't focus on". Both say they were last updated on 10 December 2025. I read both on 11 September 2026.

Link juice and link equity are informal names for whatever a link passes when it passes something. The mechanism behind the names is PageRank. The arithmetic almost everyone teaches for it, value divided by the number of links on the page and a fixed share lost on the way, comes from a paper published in 1998. Both are true at once.

What follows is what Google has said, when, what it has never said, and which columns to delete from your prospecting sheet. If that last part is what you came for, jump to what to change on your prospecting sheet. The rest comes down to this: nobody outside Google can see what a link is worth, anyone can see whether it is still there, and a link you paid for in March can be gone, unfollowed or redirected by August with nothing to announce it.

Link juice, link equity, link authority: same thing

Three names, one idea, no authoritative definition anywhere. Google does not define "link juice" or "link equity" in its documentation. The tool vendors use both terms freely in their own writing, but none of them sells either as a metric you can read off a screen. They are shop talk, and have been since the early 2000s.

Glossary pages like to draw a distinction: juice is the casual term, equity the professional one. That distinction is invented. It describes which word a writer prefers in front of a client, not two different quantities.

What the names point at is real. Links carry ranking value, and Google has never stopped saying so. What no name carries is a unit, a scale, or any way for someone outside Google to read the number.

Does PageRank still exist in 2026?

Yes. Google's guide to Google Search ranking systems says so in the present tense: "Among these is PageRank, one of our core ranking systems used when Google first launched... How PageRank works has evolved a lot since then, and it continues to be part of our core ranking systems."

That sentence gets quoted constantly. The one that almost never gets quoted beside it sits in the same documentation set. Google's SEO starter guide has a section headed "Things we believe you shouldn't focus on", with PageRank as an entry: "While PageRank uses links and is one of the fundamental algorithms at Google, there's much more to Google Search than just links."

Google's position, stated twice on the same day: PageRank runs, and you should not build your week around it. Those are not in conflict. And note "has evolved a lot since then", because every claim about how link value gets distributed rests on a model Google itself calls much changed.

The number died in 2016. The system didn't.

This is where "PageRank is dead" comes from, and both halves get repeated as one fact.

From 2000, Google's browser toolbar showed a public PageRank score from 0 to 10. By the time it was switched off it had gone years without a refresh. On 8 March 2016, Barry Schwartz reported in Search Engine Land that Google had confirmed the removal, and that "the company still uses PageRank data internally within the ranking algorithm, but the external PageRank values shown in the Toolbar are going away completely".

What ended in 2016 was visibility, not the system. Ten years on, nobody outside Google has seen a PageRank value for any page. Every number you can see today, DR, DA, UR, PA, Authority Score, is a vendor's model of the outcome built from that vendor's own crawl. None is PageRank, and none of their makers says it is.

Where "link juice divides and drains" actually comes from

One paper. Sergey Brin and Larry Page, The Anatomy of a Large-Scale Hypertextual Web Search Engine, published in 1998 at Stanford. Section 2.1.1 holds every piece of the arithmetic the industry still recites.

The paper says PageRank "extends this idea by not counting links from all pages equally, and by normalizing by the number of links on a page". It names the divisor: "C(A) is defined as the number of links going out of page A." And it sets the constant: "The parameter d is a damping factor... We usually set d to 0.85." That is the whole origin of every "your link is diluted" argument ever made.

One thing to know before leaning on it. The section explaining the damping factor describes d as "the probability at each page the 'random surfer' will get bored and request another random page", while the formula uses d as the share that carries forward. Read literally, those sentences disagree about what 0.85 is the probability of. The paper gets quoted as a specification. It is not one.

Then the circular part. Google's ranking-systems page sends you to this exact paper, and in the next sentence says the mechanics have "evolved a lot since then". Cited and disowned in one breath. That is the strongest available statement about how link value moves in 2026, and it is not a statement about how link value moves.

What the 1998 model predicts, if it still held

Worked plainly, because everyone gets shown this formula and almost nobody gets shown its date. Take a page with ten outbound links. Under the 1998 formula each link receives a tenth of what the page has to give, and the damping factor means a fixed share never arrives. Add ninety more links and each carries a hundredth. That is the source of "don't buy a link on a page with a hundred outbound links", and of "cap your outbound links at 100". It is a model describing a prototype, and nobody at Google has restated it as current mechanics since 2009.

2009: the year link sculpting stopped working

Sculpting meant marking some of a page's links so the value they would have taken went to the others instead. Standard practice at the time. On 15 June 2009 Matt Cutts, then head of Google's webspam team, wrote on his own blog that it had already stopped working. "Originally, the five links without nofollow would have flowed two points of PageRank each... More than a year ago, Google changed how the PageRank flows so that the five links without nofollow would flow one point of PageRank each." Marking a link no longer hands its share to the neighbors. The share evaporates.

Everyone quotes that. Almost nobody quotes the clause that matters most: more than a year ago. The tactic had been dead for over twelve months before anyone was told, and we only know because Cutts chose to say so.

The same post carries the line that undercuts every arithmetic argument built on the paper. Before he starts, Cutts writes: "Even when I joined the company in 2000, Google was doing more sophisticated link computation than you would observe from the classic PageRank papers." He will write the rest, he says, "in the framework of 'classic PageRank' but bear in mind that it's not a perfect analogy".

So the last first-party account of the arithmetic opens by saying Google's real link computation had been ahead of the paper since at least 2000. Even the "15 percent" comes from that post, where Cutts calls the decay factor something "you could think of" as 10 to 15 percent of a page's value disappearing. An intuition, inside a disclaimer, about a model he has just called imperfect.

2020: the last thing Google said about the phrase itself

On 29 July 2020, Google's John Mueller was asked how to direct link juice from a home page to the rest of a site. His answer, reported the next day by Search Engine Roundtable and two days later by Search Engine Journal: "I'd forget everything you read about 'link juice.' It's very likely all obsolete, wrong, and/or misleading. Instead, build a website that works well for your users."

That is not documentation. It is a reported tweet and should be weighed as one. But it is the most recent thing anyone at Google has said about the term, and it says the advice built on the term is obsolete.

Six years on, the arithmetic still sits at the top of the results for "link juice". Open the first three glossary pages defining it, count how many print the division as current mechanics, then look at what any of them cites for it. Not because anyone is lying, but because the 1998 model is teachable in ninety seconds and yields confident, checkable advice. Nothing published since yields any advice at all.

Is PageRank the same as Domain Authority?

No, and the distinction is the practical core of this. PageRank is Google's internal link-analysis system, with no public value since 2016. Domain Authority is Moz's score, built from Moz's own crawl to predict ranking rather than report what Google holds. Ahrefs' DR is the same kind of thing from a different crawl, which is why the two disagree so often; see how Moz's and Ahrefs' scores are built and which Ahrefs numbers are calculated and which are modeled.

Nobody can show you a PageRank score. No API, no report, no export. A seller offering you one, or an "equity score" derived from it, is selling a model with a confident label. Page-level scores are the more informative family of guesses, per why the page-level score is the better diagnostic, but guesses is what they are.

What to change on your prospecting sheet

Most people typing "link juice" into Google are sizing up somebody else's page, not tidying their own. So: what to delete, and what survives.

Delete any column called link juice, equity score, or power. Whatever sits in it is a vendor score under a different name, or a number somebody invented.

Delete outbound-link count as a scoring input. This is the one nobody expects. That column is C(A) from the 1998 paper wearing a suit. It entered the trade as a divisor in a formula, and no first-party source since 2009 has confirmed link value is still divided that way. Keep the number for reading the page, as in pulling the outbound list on your own pages. Stop scoring on it as a dilution calculation.

What survives is what you can observe:

• Is the page indexed?
• Is the link followed, or does it carry a rel value saying otherwise?
• Is it a real page a human could reach in two clicks from the site's navigation?
• Does the site have traffic visible anywhere but the seller's deck?

None of those measures value. All are conditions the question of value depends on.

Keep DR or DA, but demote them to a coarse filter for throwing out the bottom of a list. The gap between DR 68 and DR 74 is not something you can act on. Counting links versus counting sites is a steadier read.

One column to add: a re-check date, set when you buy the link. Monthly for links you paid for, quarterly for the rest. The value of a link is unknowable. Its existence is not, and it changes. A link bought in March can be gone, unfollowed or redirected by August, and nothing will tell you unless you look.

What this means for your own internal links

Some of the old advice survives, because it never depended on the arithmetic. A page nothing links to receives nothing, under any model of link analysis including one nobody has published. Fewer, more relevant links generally beat a pile of them, which is a claim about relevance rather than division. And depth matters: a page two clicks from your home page is treated differently from one buried six levels down.

What does not survive: outbound caps, sculpting with nofollow, "spread the juice evenly", and link-count arithmetic in any form. Cutts's 2009 post kills the second directly, and his advice there was to "let PageRank flow freely within your site".

The figures people quote for internal linking, three to ten links per thousand words and the variants, are somebody's heuristic in a lab coat. Google has published no number. When one is presented as guidance, ask whose it is.

You can't check what a page passes. Check these instead.

You cannot measure link equity. Nobody outside Google can, and every tool reporting a figure is reporting its own model. What you can check are the three conditions that decide whether the question applies to your link at all.

Is the link still on the page? Is it still followed, or has a rel value appeared that was not there when you paid? Does it resolve to your page directly, or through a chain ending somewhere you never agreed to. You can check the rel attribute on any list of links without an account, pull every link leaving a page the same way. For the third, see what a redirect chain really costs you.

Be clear about what those checks are. Observations about markup and HTTP responses. They tell you the link exists in the form you bought, and nothing about what it is worth. Any tool claiming otherwise, ours included, would be overselling itself.

Questions people ask

Is link juice the same as link equity?

Yes. Both are informal names for the ranking value a link passes. Google defines neither, and no tool vendor sells either as a metric. "Equity" as the formal term and "juice" as the casual one is a preference about register.

Does PageRank still exist?

Yes. Google's ranking-systems documentation, updated 10 December 2025, says PageRank "continues to be part of our core ranking systems". The public toolbar score was switched off in March 2016, so no PageRank value has been visible outside Google for ten years.

Does a nofollow link pass link juice?

Generally no. Google's documentation says links marked nofollow, sponsored or ugc "will generally not be followed", so do not buy one expecting ranking value. Full picture, including the 2019 hint change, in what nofollow, sponsored and ugc actually do.

Do 301 redirects lose link juice?

Google has said two different things, three years apart, and never published a number for either. In February 2013 Search Engine Land reported Matt Cutts saying "the amount of PageRank that dissipates through a 301 is currently identical to the amount of PageRank that dissipates through a link", which assumes a link dissipates something. In July 2016 the same publication reported Gary Illyes's tweet: "30x redirects don't lose PageRank anymore." The percentages that circulate, 90 percent through a 301 or 99, come from neither.

How many outbound links is too many per page?

Google has never published a number, and its current Search Essentials carries no link-count guidance. "Cap it at 100" has two ancestors: an old crawl-size limit dropped in 2013, and the 1998 formula's habit of dividing by link count. History in pulling the outbound list on your own pages.

Is there a PageRank checker that works?

No. Anything sold as one returns a vendor's score with PageRank written on the label, or a number with nothing behind it. The last real value was the toolbar score, switched off in March 2016.

Can you measure how much link equity a page passes?

No. Every number claiming to is a third-party model of a system only Google can see. Ask instead whether the link still exists, is still followed and still resolves to the page you paid for, because those are observable and they change.

What a link is worth, and what it isn't

People want link juice to be a quantity because a quantity can be managed. It fits in a column. You can sort by it, report it, defend a budget with it. "Some links matter and nobody outside Google can see how much" is a true sentence no process can be built around, which is why the 1998 arithmetic has outlived every correction to it.

So stop scoring on it. The formula is from 1998, and the last first-party account of it — Matt Cutts, 15 June 2009 — arrived with his own warning attached: he was writing "in the framework of 'classic PageRank'", and it was "not a perfect analogy". Nobody has updated it since. Deleting that column costs you nothing, because it was never measuring anything.

What you can build a process around is narrower and duller. Not how much a link is worth, but whether it is still there, still followed, still pointing where you paid for it to point. That is not a measurement of value. It is the precondition for value, and the only part anyone outside Google can check — which also makes it the only part that can go wrong without you knowing. A link bought in March can be gone by August, and the sheet will not tell you. Keeping that column honest means opening every donor page again, one at a time, on a schedule you actually keep. That, and not a number, is what we built LinkGuard to do.