What Is Harmonic Centrality? Graph Theory, PageRank, and Modern SEO

What Is Harmonic Centrality?

Harmonic centrality is a graph theory metric that measures how central a node is within a network. It calculates the sum of inverse shortest-path distances from a given node to every other node in the graph. The higher the score, the more easily that node can be reached from anywhere in the network.

The formula is straightforward:

If a node can be reached in one step from five other nodes, it adds 5 x (1/1) = 5 to its score. If another five nodes are two steps away, it adds 5 x (1/2) = 2.5. The score rewards proximity: nodes that are close to many other nodes accumulate a high harmonic centrality score quickly.

Harmonic centrality was introduced in the academic literature as an improvement on closeness centrality specifically because it handles disconnected graphs correctly. On the web, where not every page links to every other page, this robustness matters. A node in an isolated cluster does not inflate the scores of other nodes; it simply contributes zero to the sum.

How Harmonic Centrality Affects AI Search Authority

Harmonic centrality does not appear in any Google documentation as a named ranking factor. What it provides is a structural framework for understanding why certain architectural practices consistently produce better crawl coverage, stronger link equity distribution, and higher AI citability.

Four practical implications follow directly from the framework:

For a structured assessment of how your site's current architecture performs across these dimensions, our AI SEO service includes a full internal link graph audit, entity footprint analysis, and crawl priority mapping. You can also test how AI systems currently retrieve and cite your content using our LLMO Prompt Tester.

How It Differs from PageRank and Closeness Centrality

Understanding harmonic centrality is easier once you contrast it with the two metrics it is most often compared to.

PageRank

PageRank models a random surfer who clicks links at random. Each time a page receives a link from a high-authority page, some of that authority flows to it. PageRank is directional and weighted by the quality of incoming links. Harmonic centrality, by contrast, is purely structural: it measures position within a network regardless of the prestige of individual connections. A page that receives one link from the homepage will have a higher harmonic centrality than a page buried three clicks deep, even if that buried page has many incoming links from other deep pages.

In practical SEO terms: PageRank tells you how much authority a page has accumulated. Harmonic centrality tells you how easy that page is to discover through the link graph.

Closeness Centrality

Closeness centrality calculates the average shortest-path distance from a node to all others, then inverts it. The problem is that if even one node is completely unreachable, the average distance becomes infinite and the score collapses to zero. On real websites with orphaned pages or broken link clusters, closeness centrality produces misleading results. Harmonic centrality sidesteps this by taking the sum of inverses rather than the inverse of the sum. Unreachable nodes simply contribute zero without corrupting the overall score.

This mathematical robustness is why harmonic centrality is the preferred metric in network analysis tools that work with real world web graphs, where disconnection is common.

Harmonic Centrality in Google's Knowledge Graph

Google's Knowledge Graph is itself a large graph of entities and relationships. Every entity in that graph, whether it is a business, a person, a location, or a concept, has a position defined by how many other entities relate to it and how closely. Harmonic centrality, or equivalent measures, plays a role in determining which entities Google treats as authoritative and which remain ambiguous.

An entity like "Melbourne" is connected to thousands of businesses, events, people, landmarks, and government bodies within Google's Knowledge Graph. Its harmonic centrality score is enormous. When your business is mentioned in relation to Melbourne, by having a Google Business Profile, consistent local citations, or coverage in Melbourne-based publications, some of that centrality is inherited. Google can more confidently disambiguate your entity because it sits close to many well-established nodes in the graph.

Contrast this with a brand that has no website, no consistent directory listings, no mentions in external publications, and no social profiles. That entity effectively has a harmonic centrality score of zero within Google's Knowledge Graph. It does not appear in Knowledge Panels, AI Overviews rarely cite it, and Google has low confidence associating any query with that specific entity.

This is why entity-building is a core component of modern AI driven SEO strategy. It is not just about keywords or links; it is about establishing your brand as a well-connected node in a broader knowledge network. When Google's algorithms perform graph traversal to answer a query, entities with high harmonic centrality are encountered more frequently and cited with higher confidence.

The connection to semantic search is direct. Semantic search systems do not match strings; they traverse entity graphs and weight results by entity confidence. An entity's position within the knowledge graph, which harmonic centrality captures structurally, directly influences whether that entity appears in AI-generated answers.

Internal Link Architecture: Designing for Centrality

The principles of harmonic centrality translate directly into practical internal linking strategy. Your website is a graph. Every page is a node. Every internal link is a directed edge. The structure of those edges determines the harmonic centrality score of every page on your site.

High-centrality pages are reachable from everywhere else in a small number of clicks. Low-centrality pages require many hops to reach, and some pages, orphaned pages with no inbound internal links, have a centrality score of zero.

The Topic Cluster Model Through a Centrality Lens

The topic cluster model makes sense from a harmonic centrality perspective. A pillar page sits at the centre of its cluster. Every cluster page links to it. The pillar page links back to all cluster pages. This bidirectional linking creates a dense sub-graph where the pillar page has the highest harmonic centrality within its topic cluster, exactly where you want it.

Cluster pages benefit too. Because they are linked from the pillar page and from sibling cluster pages, they are reachable in just two hops from any page that links to the pillar. Their centrality scores are moderate but consistent.

What Kills Centrality

Several common architectural mistakes dramatically reduce harmonic centrality for important pages:

• Orphan pages: Pages with no inbound internal links have zero centrality. Googlebot may never discover them through crawling alone.
• Flat navigation with no cross-linking: If every page only links to the homepage and one or two service pages, cluster pages remain at a distance of three or more hops from each other. Their centrality stays low.
• Pagination chains: Category pages buried behind /page/2, /page/3 chains are two or three extra hops from the main category page. Their centrality plummets with each added layer.
• Seasonal or campaign pages left disconnected: Landing pages built for a single campaign and never linked from permanent navigation become orphans the moment the campaign ends.

Vector embeddings power the semantic relevance side of search, but crawlability and discoverability are prerequisites. A page that is semantically perfect but has a centrality score of zero may never be crawled frequently enough to rank.

Measuring and Improving Your Site's Centrality Profile

You do not need a custom graph database to apply harmonic centrality principles. Several practical proxies help you approximate centrality across your site.

Using Google Search Console

Pages that Google crawls frequently are, by definition, highly reachable within your link graph. In Search Console, check crawl frequency under the URL Inspection tool. Pages crawled daily or weekly typically have high harmonic centrality. Pages crawled monthly or rarely are likely low-centrality pages sitting deep in your architecture.

Using Log File Analysis

Server logs show exactly which pages Googlebot visits and how often. Pages visited rarely despite being important content are candidates for centrality improvement: add inbound internal links from high-traffic, frequently-crawled pages.

Using Screaming Frog or Similar Crawlers

Any desktop crawler can report inbound internal link counts per page. Pages with zero inbound links are orphans. Pages with one or two inbound links from low-traffic pages have low effective centrality. Sort by inbound link count and prioritise adding links to high-value pages that are undersupported.

Applying the Principle: Three Actions

1. Audit orphan pages. Export every page on your site from a crawl report. Filter to pages with zero inbound internal links. For each one, decide: delete it, redirect it, or add it to the navigation or to a relevant cluster page.
2. Strengthen pillar page centrality. Your most important service and product pages should be linked from the homepage, from every related cluster page, and from the footer or sidebar where appropriate. The goal is single-hop access from as many pages as possible.
3. Cross-link cluster pages. When a cluster page mentions a topic covered in another cluster page within the same pillar, add a contextual internal link. This reduces average path length between cluster pages and raises their individual centrality scores.

The named entity recognition systems that AI models use to parse content are only effective if those pages are actually crawled. Strong harmonic centrality ensures your best content is discovered consistently.

Frequently Asked Questions