Google’s original PageRank formula, redrawn as buckets, pipes and valves. Read the guide, then pour some authority of your own and watch where it goes.
This infographic turns Google’s original PageRank formula into buckets, pipes and valves, so you can see exactly where that value goes, where it pools and where it leaks. Below, we walk through every part of it. And whenever you’d rather stop reading and start pouring, the playground lets you build the same diagram with your own pages.
1The Infographic
2The Big Idea: Buckets, Pipes and Valves
The whole graphic runs on one metaphor. Learn these three pieces and you can read any part of it at a glance.
3Walking Through the Diagram
Twelve buckets, one rule. Here’s how the value moves from the top of the diagram to the bottom.
Everything starts with Page A, a strong page holding 80 units of value it earned from links outside the diagram. Page A links to two pages, so it splits its value two ways.
But not all 80 flows. PageRank uses a damping factor of 0.85, so only 85% of a page’s value travels through its links. The other 15% goes to random jumps spread across the whole web.
So Page A sends 80 × 0.85 = 68 through its pipes, 34 to each page. The remaining 12 never enters the pipes at all.
The bottom row (G, H, I, J and K) are dangling pages. They collect value but don’t pass any further.
4What the Pipes Teach Us
The four numbered markers on the diagram each point to a lesson about how link equity behaves.
5What This Means for Your Site
You’ll never see Google’s real numbers, but the patterns in the diagram hold up. Here’s how they translate into everyday SEO decisions.
6The PageRank Formula, Piece by Piece
The buckets are a simplified picture of the real thing. Here’s the full update rule the infographic is based on.
It looks intimidating, but it’s three simple ideas added together. Click through each part:
The surfer doesn’t always click links. With d = 0.85, they jump to a random page 15% of the time. vu is page u’s slice of those jumps. In the infographic, this is the 15% that never enters the pipes.
This is the part the buckets show. For every page w that links to u, take its PageRank, multiply by 0.85 and divide by its number of outbound links, L(w). Add up all those streams and you get the water flowing into page u. Every pipe in the diagram is one of these terms.
Dangling pages, like the bottom row of buckets, have no outbound links, so the surfer gets stuck there. The formula takes the value sitting on those dead ends and hands it back out like a random jump, so none of it disappears.
Each pass, a page’s score = its slice of random jumps + value flowing in through links + its slice of the dead-end pages.
Repeat that for every page until the numbers stop changing. That loop is called power iteration, and it’s exactly what the playground runs every time you add a page or flip a valve.
PageRank can be thought of as a model of user behavior. We assume there is a “random surfer” who is given a web page at random and keeps clicking on links, never hitting “back” but eventually gets bored and starts on another random page.
Sergey Brin and Lawrence Page, The Anatomy of a Large-Scale Hypertextual Web Search Engine (1998)
7Before PageRank, Search Was Easy to Fool
In the ’90s, search engines like AltaVista, Lycos and Excite ranked pages mostly on what the page said about itself: keyword counts, title tags and meta keywords. The result was easy to game, and pages stuffed with repeated words and hidden text filled the top results.
PageRank changed the question. Think of it like academic research, which is where Larry Page got the idea. Anyone can call their own paper groundbreaking. What matters is how often other researchers cite it, and a citation in a top journal outweighs dozens from obscure ones. PageRank treated links the same way, so a page earned its rank through the respect of others, not its own claims.
rel="nofollow" arrivesrel="sponsored" and rel="ugc" and treats all three attributes as hints rather than strict rules.8How Much Juice Do Big Brands Have?
Google stopped showing PageRank in 2016, so the closest public stand-in is Open PageRank . These scores come from its September 2026 release.
Reading about link equity only gets you so far. Pour authority into a page, connect the pipes and watch every bucket recalculate in real time.
?Frequently Asked Questions
Link juice is SEO shorthand for the ranking value one page passes to another through a link. The more value a page has and the fewer links it splits that value across, the more each link passes. The formal term is link equity.
Not the page itself. In the formula, a page’s value comes only from the links pointing to it. Its outbound links decide how that value is split among the pages it links to.
No. Since 2009, a nofollow link still takes its share of the split and that share is lost, so nofollowing internal links wastes value instead of steering it somewhere else.
A page with dozens of links cuts its value into tiny slices. Many directories also have little authority to begin with, and Google tends to ignore links from low-quality directories.
Google has said PageRank is still one of its signals. Today’s version is far more complex than the 1998 formula and works alongside many other ranking systems.
No. The bucket values are made up for teaching, and the brand scores come from Open PageRank, not Google. The math follows the original PageRank formula, simplified so you can follow it by hand.
Yes. Open the interactive link juice playground, start from the blank canvas and add your own pages and links. You can rename every bucket and share the finished graph with a link.
Copy and paste the code below into your page. Please keep the credit link intact.
Twelve buckets, one formula and as many pipes as you want to draw. See where your link juice really goes.
Sources
- Brin, S. and Page, L. (1998). The Anatomy of a Large-Scale Hypertextual Web Search Engine.
- Cutts, M. (2009). PageRank sculpting.
- Google Search Central (2019). Evolving “nofollow”: new ways to identify the nature of links.
- Open PageRank. Methodology and Top 10 Million Domains (September 2026 release).
