What Is Topical Relevance in SEO and How to Measure It
Topical relevance is how closely a page or a site relates to the topic behind a query, judged by the entities, attributes and context it shares with that topic rather than by the words it repeats. It differs from keyword relevance, which matches strings, and from topical authority, which adds coverage of the whole subject and the history of performing on it. Topical relevance works at two levels, the page and the site, and search engines estimate it from entities, meaning-based matching, internal links and the share of a site given to the topic. It rises with named entities, complete attributes, one contextual vector per page and links from related pages, and it falls when off-topic sections, a mixed source context or generic text dilute it. Contextual borders protect it, it can be measured with similarity scores, entity overlap and the query mix in Search Console, a new site builds it through publication order, and a six-step check tests it on any page. Holistic Radar’s own query mix is the worked example, and the questions at the end cover relevance without backlinks.
What topical relevance is
Topical relevance is the degree to which a document belongs to the topic a query is about. A page is topically relevant to “how to build a topical map” when it discusses the central entity, source context, attributes and rows that the topic contains, whether or not it repeats the exact query.
The unit is the topic, not the query string. A query is one way of asking about a topic, and the same topic generates many queries. A page that is relevant to the topic can rank for queries it never states word for word, and a page that repeats a query without covering its topic is relevant to the string and to little else.
Two pages show the difference. One repeats “topical map template” eight times and lists a few column names; the other never uses the phrase twice but defines the central entity, the source context, the core and outer sections, the raw and processed rows and the bridge table. The second is more relevant to the topic behind the query, and it is the one a meaning-based system matches to queries the first never anticipated.
Topical relevance is a prerequisite, not a ranking by itself. Among the pages that are relevant to a topic, ranking then compares their quality, their source and their history.
Topical relevance, keyword relevance and topical authority
Three terms describe how a page relates to what is searched, and each covers more than the one before it.
| Term | What it compares | What it misses |
|---|---|---|
| Keyword relevance | The query’s words with the page’s words | Synonyms, entities, and topics stated in other words |
| Topical relevance | The query’s topic with the page’s entities, attributes and context | Whether the site covers the whole subject, and for how long |
| Topical authority | The site’s coverage of the whole subject, with its history of performing on it | Nothing within the subject; it is the widest of the three |
Topical authority is a site’s demonstrated coverage of a subject, measured against the full set of questions the subject contains. It is built from many topically relevant pages over time, which is why topical relevance comes first in the order of work and topical authority last.
Page-level and site-level topical relevance
Topical relevance works at two levels, and the two can disagree. Page-level relevance is how well one page matches one topic. Site-level relevance is how much of a site is about the topic, and how the site’s other pages relate to the page in question.
A strong page on a site about something else is relevant on its own and isolated in its context. A weak page on a site that covers the subject thoroughly is supported by its neighbours, and often outranks the strong isolated page once the site’s coverage is recognized. The two levels compound: a relevant page inside a relevant site, linked from related pages, carries both signals at once.
The practical consequence is that a site cannot borrow topical relevance for one page. Publishing a single page on a new topic tests page-level relevance only; the site-level signal arrives as the related pages arrive.
How search engines estimate topical relevance
Search engines estimate topical relevance from four kinds of evidence, and each one is visible in how a page is written and linked.
- Entities and attributes. The entities a page names, and the attributes it states about them, show which topic the page belongs to.
- Meaning-based matching. Systems that represent queries and passages as vectors match a page to a query by meaning, so a page that answers the topic in other words still matches.
- Internal links and anchors. Links from pages on the same topic, with anchors that name the target’s attribute, place a page inside the topic’s cluster.
- Site share. The proportion of a site’s pages, links and queries that belong to the topic indicates whether the site is a source on it.
Relevance is also estimated below the page, at the level of passages. A long page can hold one passage that answers a narrow query better than any page written for it, and that passage can be retrieved on its own. Passage-level relevance rewards sections that state their own subject in their first sentence, because a passage that depends on the paragraph before it cannot be read alone.
None of the four reads keyword density. Repeating a query string adds nothing an engine that matches by meaning did not already have, and it can lower the quality of the passage that carries it.
What raises topical relevance
Topical relevance rises when a page states its topic in the forms a search engine extracts. Five changes do most of the work:
- Name the entities. State the central entity and the related entities the topic contains, defined in the sentences that introduce them.
- Complete the attributes. Cover the attributes searchers ask about for the topic, with exact values where values exist.
- Hold one contextual vector. Approach the topic from one angle per page, so the page does not split between a definition, a comparison and a product.
- Link from related pages. Add contextual links from pages on the same topic, with anchors that name the attribute the target owns.
- Lead with an extractive summary. Open with a paragraph that names the topic’s entities in the order the page covers them.
An entity gap list shows which entities the ranking pages share and a page lacks, which makes it the most direct check of page-level relevance.
What dilutes topical relevance
Topical relevance falls when a page or a site carries material that belongs to a different topic or a different source. Four patterns are common:
- Off-topic sections. A page about topical maps that adds a long section on link building tells the engine the page has two subjects.
- A mixed source context. A consultancy that publishes job listings and course reviews adds pages a different source would publish.
- Boilerplate. Repeated blocks of navigation, promotions and template text outweigh the main content on thin pages.
- Generic text. Sentences that would fit any topic, such as claims that SEO “has evolved”, carry no entity and no attribute.
Dilution is easier to prevent than to repair. A topical map with rejected rows keeps off-topic pages out before they are written, and the source context decides which rows are rejected.
Topical relevance and contextual borders
A contextual border is the point where a page stops covering its own attribute and a neighbouring page begins. Borders protect topical relevance in both directions: the page stays about one attribute, and the neighbour keeps the attribute it owns.
A page that crosses its border, explaining the neighbour’s attribute in full, becomes relevant to two topics and fully relevant to neither. It also competes with the neighbour for the same queries. The correct crossing is a bridge: one sentence that names the neighbouring attribute, states the relation, and links to the page that owns it. A topical map draws the borders before writing starts, so each writer knows where their page ends.
How to measure topical relevance
Topical relevance has no single score, but three measures approximate it at the page and site levels.
| Measure | Level | How it is taken |
|---|---|---|
| Similarity to ranking pages | Page | Cosine similarity between the page and the main content of the top results |
| Entity overlap | Page | The share of consensus entities the page names |
| On-topic query share | Site | The share of Search Console queries that map to a page in the topical map |
The first two are computed with the scripts in the Python for semantic SEO article and read as described in semantic analysis. The third needs only Search Console and the map: every query is assigned to the page that owns its attribute, and the queries left over are either noise or gaps.
Topical relevance for a new site
A new site has page-level relevance from its first page and site-level relevance only as its related pages arrive. The order of publication decides how quickly the second signal forms. Root attributes and quality nodes go first, published together, so the first pages a search engine reads already describe one subject from several angles. Rare attributes and outer articles follow at a steady rhythm that keeps adding related pages.
Holistic Radar launched 17 quality nodes on one day for this reason. A site that publishes one article a month across many topics spends a year being relevant to nothing in particular; a site that publishes one cluster at a time is relevant to that cluster within weeks.
A topical relevance check in six steps
A topical relevance check runs on one page and one topic, and takes six steps:
- State the topic and the attribute the page owns in the topical map.
- Read the first sentence: it should name the central entity or the attribute.
- List the entities the page names and compare them with the ranking consensus.
- Mark every section that belongs to a neighbouring attribute and replace it with a bridge.
- Count the contextual links into the page from pages on the same topic.
- Check the page’s Search Console queries for intents the page does not serve.
Steps 3 and 6 produce lists; steps 2, 4 and 5 produce edits. A page that passes all six is relevant at the page level, and its site-level relevance then depends on its neighbours.
Topical relevance on holisticradar.com
Holistic Radar measured its own on-topic query share on 21 November 2026. Of the 529 queries with impressions in the previous 90 days, 483 mapped to a page that owns their attribute in the topical map, a share of 91%. The 46 queries left over carried 153 impressions, and most were noise, such as “seo radio”, or concepts the site used without a page of its own.
The same audit showed what topical relevance does not do alone. 78% of the site’s impressions sat beyond position 50, on queries the site was relevant to. Google had placed the site inside the right topic; the history that turns relevance into position was 44 days old. Relevance was the condition, and authority was still being earned.
The leftover queries became new rows. Concepts such as source context, the central entity and lexical semantics were used across many pages with no page of their own, and each received an owner URL so its queries had a destination.
Topical relevance questions
Can a site be topically relevant without backlinks?
A site can be topically relevant without backlinks, because relevance is read from what the site publishes and how it links internally. Backlinks affect how a relevant site ranks against other relevant sites, not whether it belongs to the topic.
Does publishing more pages raise topical relevance?
Publishing more pages raises site-level topical relevance only when the pages belong to the topic and connect to each other. Pages on other topics lower the share of the site that is relevant.
Is topical relevance the same as search intent?
Topical relevance is not the same as search intent. Relevance asks whether a page is about the topic; intent asks whether the page does what the searcher wanted, such as define, compare or buy. A page needs both.
Topical Relevance in the semantic SEO topical map
Topical Relevance is the relevance attribute of the topical authority entity inside the Holistic Radar topical map, and it sits in the outer section under the Know predicate.
The outer section holds the page because it defines a measure without selling one, and its bridge passes session signals toward the Radar Scan, which scores a site’s coverage and names the layer that limits it.
→ The Radar Scan: coverage scored and the binding constraint named in five business days.
→ Topical authority vs domain authority: what relevance becomes with coverage and history.
→ What is source context: the filter that keeps off-topic rows out.
→ What is an entity gap: the page-level check of relevance.