What Is a Topical Map: The Data Structure, the Schema, and the Five Conditions

A topical map is a structured inventory of every entity-attribute pair a website intends to cover, with each pair classified, prioritized, and assigned to a URL before any content is written. A topical map contains 15 fields per row, covering the attribute itself, its classifications, its source queries, its filtration verdicts, its priority, and its position in the link graph. Every row carries three independent classifications: structural class as root, seed, or node; sectional placement as core or outer; and competitive class as root, rare, or unique. A topical map is not a keyword map, a content calendar, a site map, or a topic cluster, and it differs from each on a specific point. A document qualifies as a topical map only when it meets five conditions: it is built from a query network, it is organized by attribute rather than by keyword, it carries filtration verdicts, it assigns exactly one contextual vector per URL, and it defines a finite denominator. A topical map produces three outputs: a publication order, the input fields for every content brief, and the denominator against which topical coverage is measured. A worked example then shows the root row of Holistic Radar’s own topical map, field by field.

What a topical map is

A topical map is a structured inventory of every entity-attribute pair a website intends to cover, classified and assigned to a URL before any content is written.

The unit of a topical map is the tuple, not the keyword. One tuple is one entity paired with one attribute of that entity: semantic SEO paired with its definition, semantic SEO paired with its cost, semantic SEO paired with its measurement. Each tuple becomes a heading, a table row, or a page.

A topical map is finite by construction. It states the complete set of tuples that full coverage of a subject requires, which is what allows coverage to be expressed as a fraction rather than as an opinion. A document that cannot state its own total is not a map, because a map without edges cannot show a gap.

What a topical map contains

A topical map contains 15 fields per row. The schema below is the full production specification.

Nine of the 15 fields are decisions rather than data. Volume and source queries are harvested. Prominence, relevance, structural class, competitive class, section, predicate, contextual vector, priority, and bridge target are all judgments, and they are the fields that determine whether the map produces authority or produces a publication backlog.

The three classifications inside a topical map

Every row in a topical map carries three independent classifications, and the three run on separate axes.

Structural class records depth. A root attribute is inseparable from the central entity’s definition, and removing it destroys the entity. A seed attribute branches from a root and spawns its own cluster. A node attribute is terminal and spawns nothing. Structural class determines URL depth and internal link direction.

Section records the predicate. Core section rows serve the transactional predicate and carry the money. Outer section rows serve the informational predicate and capture the entry queries of every path. Section determines where the row sits relative to the contextual border and which direction its links run.

Competitive class records scarcity. A root attribute is well covered across competing properties. A rare attribute exists in the knowledge domain and is almost absent from competitor coverage. A unique attribute originates from the publisher’s own source and cannot exist elsewhere: proprietary frameworks, original data, named methods, internal results. Competitive class determines where information gain is available and therefore where authority is cheapest to acquire.

The three axes are independent. A row can be root, outer, and rare simultaneously, and the combination is what sets its priority.

What a topical map is not

A topical map is confused with four adjacent artifacts, and it differs from each on a specific point.

A topical map is not a keyword map. A keyword map assigns keywords to URLs. A topical map assigns attributes to URLs, and the attributes are derived from a query network rather than from a keyword tool export. The practical difference is that a keyword map cannot represent an attribute with no search volume, and low-volume attributes are precisely where prominence and information gain concentrate.

A topical map is not a content calendar. A content calendar records dates. A topical map records priority, and the priority is derived from prominence and structural class. A calendar built from a map is a valid output of the map; a calendar built without one is a schedule for publishing in an arbitrary order.

A topical map is not a site map. A site map records URLs that exist. A topical map records tuples, most of which do not yet have a URL. The map’s value sits almost entirely in the rows that are not yet built, which is the opposite of a site map’s function.

A topical map is not a topic cluster. A topic cluster is a hub page with supporting pages linked to it, which is a shape. A topical map is an inventory with classifications, verdicts, and a finite total, which is a data structure. A map produces clusters; a cluster does not produce a map.

The five conditions a document must meet to be a topical map

A document qualifies as a topical map only when it meets five conditions. A document failing any one of them is a keyword list with different column headings.

Condition one: it is built from a query network. The rows originate from harvested queries with their co-occurrence and sequence recorded, not from a keyword tool’s related-terms export. Sequence data is what determines internal link direction, and a map without it cannot specify the link graph.

Condition two: it is organized by attribute, not by keyword. Rows name entity-attribute pairs. If the rows are search phrases, the document cannot represent an attribute that people ask about in twelve different phrasings, and it will produce twelve thin pages instead of one complete one.

Condition three: it carries filtration verdicts. Every row records a prominence verdict and a relevance verdict, and rows that failed filtration are visible as rejections rather than silently absent. A map with no rejections was not filtered, and an unfiltered map is a harvest.

Condition four: it assigns exactly one contextual vector per URL. Two vectors on one URL produce a page that answers two different questions and ranks well for neither. A map that leaves the vector field empty has deferred the decision to the writer, which is where drift and cannibalization originate.

Condition five: it defines a finite denominator. The map states its total tuple count, so that coverage is computable as covered over total. A map with no total can never report a gap, which removes the only thing a map does that a list does not.

Most documents sold as topical maps satisfy conditions one and two and fail three, four, and five. The three failures are also the three that are invisible to a buyer who has not been told to check them.

What a topical map produces

A topical map produces three outputs, and the map is judged by them rather than by its own appearance.

A publication order. Rows sorted by prominence first and popularity second, with core section root attributes published before anything else, and each seed cluster completed before the next opens.

The input fields for every content brief. Ten of the 15 fields transfer directly into a brief: the attribute, its three classifications, the contextual vector, the source query cluster, the entity vector, the internal link targets, the priority, and the bridge target. A map that does not populate a brief has not reduced the work of writing; it has only described it.

The denominator for topical coverage. The tuple total becomes the divisor in every coverage measurement thereafter. This is the output with the longest life, because it converts every subsequent publication decision from a judgment into an arithmetic one.

A topical map root row from holisticradar.com

The root row of Holistic Radar’s own topical map is the Semantic SEO service page, and its fields show how a core section root differs from an outer article.

The row is live as semantic SEO services. Its two link targets are its service nodes, topical map production and query network research, and this article is one of the outer rows that bridges toward it.

Topical map architecture in the Six-Layer Pyramid

The five conditions above are checkable in about ten minutes against any document, and that is the reason to state them plainly rather than to keep them as trade knowledge. A buyer who knows to look for filtration verdicts, a vector per URL, and a finite denominator can tell a topical map from a generated keyword list before paying for it, and can tell afterward whether what arrived was what was described.

Holistic Radar produces topical maps against all five conditions, delivered as the full 15-field schema published above, with the rejected rows included so the filtration is auditable rather than asserted. The map arrives with its tuple total stated, which means coverage is measurable from the first article rather than from the hundredth.

Raw map versus processed map

A raw map is the list of entity and attribute pairs a query network produces, and a processed map is the same list turned into title tags, URLs and zone classes.

A raw concept map that mirrors the search results page produces overlapping URLs and cannibalization. A processed map resolves each pair into one page with one title, one slug and one place in the tree. The Holistic Radar map records both versions for every row, published in the topical map template.

Macro-context and micro-context on every page

Macro-context is the one entity and attribute pair a page exists for, and micro-context is the supplementary content that connects it to neighbouring pages.

Macro-context holds the main content, from the H1 to the last main H2. Micro-context sits below a grouper heading and holds bridges, antonym bridges and sibling links. Mixing the two blurs the page’s contextual vector, so the map assigns both before a brief is written.

→ The Six-Layer Pyramid: the topical map architecture behind every Holistic Radar build

Topical map size, build time, AI, and upkeep

How large a topical map should be

A commercial topical map typically resolves to several hundred tuples across a smaller number of URLs, because multiple tuples routinely share a page as headings or table rows. Holistic Radar engagements have produced builds at 216 pages and at over 400 pages, and the page count is an output of the tuple inventory rather than a target set in advance.

A map sized before the query network is harvested is a quota. Quotas produce thin pages, and thin pages raise cost of retrieval faster than they add coverage.

How long a topical map takes to build

The harvest and annotation stages are the fast half and scale with tooling. The filtration stage is the slow half, because prominence and relevance verdicts are judgments about a subject rather than operations on a spreadsheet, and they cannot be accelerated by adding rows per hour.

A map delivered faster than its filtration can be reasoned about has skipped filtration. The visible symptom is an absence of rejected rows.

Whether AI can build a topical map

AI can perform the harvest and the annotation, and it does both faster and more consistently than a person. Query expansion, entity extraction, question-type tagging, and n-gram analysis are all mechanical and all suited to automation.

AI cannot reliably perform the prominence test, because the test asks whether removing an attribute destroys the central entity’s meaning within a specific business context, and that judgment depends on how the business earns money from the entity. A generated map will typically include everything relevant and rank it by volume, which inverts the rule that prominence outranks popularity. The result is a document that satisfies conditions one and two and fails condition three.

The practical division is that automation belongs on the left half of the schema and judgment belongs on the right half.

Whether a topical map changes after it is built

A topical map changes when the query network changes, which happens when a subject acquires new entities or when demand shifts to new attributes. The classifications rarely change, because prominence is a property of the subject rather than of the season.

Re-harvest annually, or immediately when a material change occurs in the subject. Re-filtering more often than that produces churn in the publication order without adding coverage

→ Entity SEO: how Search Engines Resolve Things, Not Strings

→ How to build a topical map: the nine steps from harvest to publication

→ Topical map template: the fields every row carries