The Reasonable Surfer Model — Google's patent describing how link equity is distributed — weights links based on the probability that a typical user would actually click them, which means the same <a> tag on a page passes very different amounts of link equity depending on its position, context, anchor text, and visual prominence
The previous articles on this site covered internal linking strategy, external link auditing, nofollow/sponsored/ugc attributes, link rot, and the footer vs content link equity difference. This article addresses link equity mechanics — specifically what determines how much equity flows through a link, the relationship between anchor text and relevance signals, and the specific internal linking architectures that maximise equity distribution.
PageRank fundamentals: the original model
PageRank (the algorithm behind Google's original ranking breakthrough) models web pages as a graph where links are votes. In its simplest form:
PR(A) = (1 − d) + d × Σ(PR(Ti) / C(Ti))
Where:
- d is the damping factor (typically ~0.85)
- Ti is any page linking to page A
- C(Ti) is the number of links from page Ti
- The sum is over all pages Ti that link to page A
What this means practically:
- A page's PageRank depends on how many links it receives and how much PageRank the linking pages have
- A page that links to 100 other pages passes less equity to each than a page linking to 10 pages
- Every additional outgoing link dilutes the equity passed per link
The damping factor (0.85): represents a "random surfer" model — with probability 0.85, the surfer follows a link; with probability 0.15, they navigate to a random page. This prevents rank accumulation on isolated clusters of mutually-linking pages.
The Reasonable Surfer Model: probability-weighted link equity
Google's 2004 patent "Information Retrieval Based on Historical Data" (updated since) introduced the Reasonable Surfer Model, which replaced the simple equally-distributed model with probability-weighted distribution:
Core idea: not all links on a page are equally likely to be clicked. Google estimates the click probability for each link based on:
Position on page:
- Header navigation links: high click probability for branded queries
- First paragraph in-content links: high click probability
- Body content links (mid-article): moderate click probability
- Footer links: low click probability (most users don't scroll to footers)
- Sidebar links: moderate, varies by relevance to content
Anchor text relevance: links with descriptive anchor text matching the linked page's topic have higher estimated click probability than generic anchors ("click here," "read more").
Link context: a link in a paragraph about "mortgage amortisation" pointing to an amortisation calculator has higher contextual relevance than the same link in an unrelated section.
Visual prominence: links in large text, high-contrast colours, or within highly-viewed page areas have higher estimated click probability.
Anchor text as a relevance signal
Anchor text — the clickable text in a hyperlink — is one of Google's strongest on-page relevance signals for the linked page:
The original insight: the anchor text you use when linking to another page tells Google something about what that other page is about, from the perspective of an outside party. A page receiving hundreds of links with anchor text "cheap flights" signals to Google that this page is about cheap flights, regardless of the page's own content.
The Penguin algorithm (2012) and anchor text spam: before Penguin, SEOs manipulated anchor text by building links with exact-match anchor text ("best mortgage calculator," "buy cheap flights"). Penguin penalised over-optimised anchor text profiles — pages with unnaturally high percentages of exact-match anchors.
Current anchor text best practices:
- Branded anchors ("Company Name"): safest, most natural
- Descriptive partial-match anchors ("amortisation schedule guide"): natural, topically relevant
- Generic anchors ("click here," "read more"): low relevance signal but natural
- Exact-match keyword anchors ("mortgage calculator"): useful in moderation; problematic if dominant
For internal links specifically: exact-match anchor text for internal links is safer than for external link building — Google understands that a site linking to its own pages with topically precise anchor text is likely editorial rather than manipulative.
Internal linking architecture: silos vs hubs
Two major internal linking architectures:
Silo structure: topic clusters where pages only link internally to pages in the same topic cluster. A mortgage section links only within the mortgage section; no links cross into the property investment section.
Hub structure: high-authority hub pages link out broadly to relevant content across the site; detailed content pages link back up to hub pages. No strict topical isolation.
The current consensus: strict silos are generally no longer recommended. Cross-topical links between related content are natural and appropriate. The more important consideration is that links flow toward your most important pages — hub pages, cornerstone content, pages you want to rank most.
PageRank sculpting: deliberately withholding links (using nofollow on internal links) to concentrate PageRank on specific pages was once a known SEO tactic. Google has discouraged this — nofollow on internal links is now recommended only for genuinely editorial reasons (login pages, privacy policy), not for PageRank manipulation.
The internal link audit: what to look for
A link extractor run on your own site reveals:
Orphaned pages: pages with no internal links pointing to them. Google has to discover these via sitemaps or external links — they receive no internal link equity. Orphaned pages are often thin content or outdated pages that should either be linked to (if they have value) or removed/consolidated.
Link depth: how many clicks from the homepage to reach important content pages. Pages 4+ clicks from the homepage typically receive lower crawl priority and less link equity. Important pages should ideally be within 3 clicks from the homepage.
Over-linked pages: pages receiving an unusually high number of internal links (the global header navigation creates internal links to every page it links to, on every page of the site). Footer links similarly. These global links pass equity but at lower probability than contextual links.
Under-linked pages: high-value pages (in terms of content quality and target keywords) that have few internal links pointing to them. Adding internal links from relevant content improves both crawlability and equity distribution.
How to use the Link Extractor on sadiqbd.com
- For internal link audit: extract links from key pages and verify that high-priority pages receive contextual in-content links (not just footer or navigation links), and that anchor text is descriptive rather than generic
- For orphan page discovery: extract links from your sitemap pages and find URLs that appear in the sitemap but receive no internal links — these are orphaned pages that need either links added or consolidation
- For competitor link analysis: extract links from competitor pages to understand their internal linking structure — which pages they consider important enough to link to from their highest-traffic content
Frequently Asked Questions
How many internal links from one page is too many? There's no definitive maximum, but Google has suggested keeping it "reasonable" — typically cited as under 100 links per page. The more relevant consideration is dilution: each link on a page receives a share of that page's equity. A page with 200 internal links passes less equity per link than a page with 20 links. More importantly, very dense internal linking can reduce crawl efficiency — Googlebot may not follow all links on a page with hundreds of links. Practically: focus on whether each link serves the user's navigation needs and passes meaningful equity. If a page is heavily linked only because of global navigation (header + footer + sidebar) but has few contextual body links, the contextual body links carry more weight despite the page's high total link count.
Is the Link Extractor free? Yes — completely free, no sign-up required.
Try the Link Extractor free at sadiqbd.com — extract and analyse every link on any web page instantly.
