Query network research is the Holistic Radar service that maps how a market actually searches: which queries appear together, which follow each other, and which sequences end in an action. This page defines what query network research produces and the three signals it records, correlative queries, sequential queries, and query paths, then shows the structural decision each signal sets. It covers the annotation applied to every query, the eight contextual vectors and ten question types, the four data sources and what has to be approximated, and a worked example from this site’s own network. It then explains how the research selects the central entity, how attributes are filtered from it, how it compares to keyword research, what is delivered, where it sits in the Six-Layer Pyramid, and the questions buyers ask.
What query network research produces
Query network research produces an annotated query dataset that records three relationships between queries: co-occurrence, sequence, and path.
A keyword list records demand as isolated phrases with volumes. A query network records demand as a graph, in which each query is a node and each relationship between two queries is an edge. The graph shows that searchers who ask one question also ask a second, that one question is usually followed by another, and that certain sequences end in a purchase, a booking, or an enquiry. A site’s structure has to mirror those relationships, which is why the network is built before the topical map.
Three signals from query network research
Query network research records three signals a keyword list cannot give: correlative queries, sequential queries, and query paths.
| Signal | What it records | Structural decision it sets |
|---|---|---|
| Correlative queries | Queries that appear together for the same need | Page scope: which attributes share one URL |
| Sequential queries | The query a searcher tends to ask next | Link direction and supplementary content |
| Query paths | Sequences that end in an action | Where contextual bridges into the core section are placed |
Without these three signals, a writer decides page scope, link placement, and forward direction on instinct at the moment of writing. Made hundreds of times across a network, those instinctive decisions produce near-duplicate pages, orphaned pages, and links that run in the wrong direction.
Correlative queries and page scope
Correlative queries set page scope: attributes whose queries co-occur for the same need share one URL, and attributes whose queries never co-occur get separate URLs.
Two phrasings of one question belong on one page. Two different questions that happen to share a word do not. Correlation data is how the difference is decided with evidence rather than by resemblance, and it is the main defense against the contextual overlap that causes content cannibalization.
Sequential queries and link direction
Sequential queries set link direction: a page links forward to the page that answers the query its readers search next.
Sequence data also decides supplementary content. When the next query is close enough to answer on the same page, it becomes a supplementary section; when it belongs to another attribute, it becomes a forward link placed at the point in the page where the question arises.
Query paths and bridge placement
Query paths set bridge placement: a contextual bridge from the outer section into the core section is placed where real paths cross from learning to buying.
A path such as “what is a topical map”, then “how to build a topical map”, then “topical map service” shows the reader moving from the knowing predicate to the ranking predicate. The bridge is placed on the page where that move happens, with a justification passage before the link.
Query network research annotation
Query network research annotates every query with its question type, its contextual vector, and its entities.
The contextual vector is one of eight: definitional, procedural, comparative, evaluative, quantitative, causal, temporal, or problem-solution. The question type is one of ten, and it decides the answer shape a page needs, from a boolean answer in the first sentence to a numbered procedure. The entity annotation records every entity in the query so that entities can be rolled up by type. The ten types and their answer shapes are set out in the ten question types.
Query network research data sources
Query network research combines four data sources: search suggestions, related searches, the site’s own Search Console queries, and competitor ranking data.
| Source | What it contributes | Limit |
|---|---|---|
| Search suggestions | Phrasings and suggestion chains that approximate sequence | No volumes, and suggestions are personalized |
| Related searches | Co-occurring and follow-on queries | A small sample per query |
| Search Console | Real queries the site already receives, with impressions | Only queries the site already appears for |
| Competitor rankings | Attributes competitors are rewarded for | Reflects their choices, not the market’s full demand |
Search engines do not publish sequence data directly, so sequential signals are approximated from suggestion chains, related searches, and the order in which a site’s own visitors search. The delivered dataset marks which signals were observed and which were approximated, because a dataset that hides its approximations cannot be audited.
A worked query network example
This site’s own query network centers on the entity semantic SEO, and a small slice of it shows all three signals.
| Query | Signal | Decision on this site |
|---|---|---|
| what is a topical map / topical map meaning | Correlative | One URL: what is a topical map |
| what is a topical map, then how to build a topical map | Sequential | Forward link between the two articles |
| how to build a topical map, then topical map service | Path | Bridge into topical map production |
| topical authority vs domain authority | Correlative with topical authority meaning | One comparative URL, not two pages |
Each decision in the right-hand column is visible on the live pages, which is how a network can be checked against the site built from it.
How query network research selects the central entity
Query network research selects the candidate central entity by rolling every annotated entity up to its type and taking the most frequent type as the candidate.
Frequency proposes; the business decides. The candidate is confirmed only when it matches how the business earns money, and the source context is then distilled to a predicate pair, one informational and one transactional, that every page on the site will serve.
How attributes are filtered from the query network
Attributes mined from the query network pass three filters before they enter the topical map: prominence, popularity, and relevance.
Prominence is a removal test: remove the attribute and check whether the central entity still means the same thing. Popularity measures demand. Relevance measures distance from the central entity for this business. Prominence outranks popularity, so a low-volume attribute that defines the subject is kept ahead of a high-volume attribute that does not. The filters are run with Semantic Instruments.
Query network research compared to keyword research
Query network research differs from keyword research in what it records: keyword research records phrases and volumes, and query network research records the relationships between queries.
| Question | Keyword research | Query network research |
|---|---|---|
| What do people search? | Yes | Yes |
| Which searches belong on one page? | No | Yes, from correlation |
| Which page should link to which? | No | Yes, from sequence |
| Where does learning turn into buying? | No | Yes, from paths |
| Which attribute defines the subject? | No | Yes, after filtration |
What query network research delivers
Query network research delivers the annotated dataset, a correlation grouping, a sequence and path summary, the candidate central entity with its entity-type roll-up, and a list of attributes ready for filtration.
The dataset is the first input of topical map production and of every content brief, and it can be delivered on its own to teams that build their own maps.
Query network research in the Six-Layer Pyramid
Query network research is layer two of the Six-Layer Pyramid, and it is the first phase of every semantic SEO services engagement.
Layer one, the source context and central entity, is confirmed from the network. Layer three, the topical map, is built from it. A weak network caps both. The reasoning behind the three signals is set out in what is a query network.
Query network research questions
How many queries does a query network contain?
A query network contains as many queries as the subject produces before new harvests stop adding new attributes. The stopping rule is attribute saturation, not a fixed count.
Is Search Console enough for query network research?
Search Console is not enough on its own, because it only records queries the site already appears for, and the attributes a site is missing are by definition absent from its own data.
How often is query network research repeated?
Query network research is repeated annually, or when the subject gains new entities. Re-harvesting more often changes the publication order without adding coverage.
Does query network research work for local businesses?
Query network research works for local businesses by recording location as an entity on the query, so location becomes a field on the service row rather than a separate page for every city.