# Source context, central entity and central search intent

Before a topical map has a single row, three things have to be decided: why the site exists, what it is about, and what its visitors come to do. Koray Tuğberk Gübür calls them source context, central entity and central search intent. Here is each one in plain words, worked through for four kinds of business, with a worksheet you can fill in within an hour.

By Kevin Jeppesen, AI growth engineer. Founder, FrameworkSEO & Ascendily. Updated 2026-11-13. Source: https://frameworkseo.com/learn/semantic-seo/source-context-central-entity

## Key takeaways

- Source context is why a site exists and how it earns: Koray Tuğberk Gübür's 2023 definition ties it to "monetization, and existing purpose", and it decides the angle every page takes.
- Two sites can share a central entity and still need different maps: a print shop and a vintage reseller both sell graphic tees, yet only a few of their pages belong on both maps.
- Write the central search intent as a verb plus the central entity, and make it cover more than the sale, or the guides that bring buyers in fall off the map.
- The three decisions set the size of the map: our median growth plan was 159 pages on 5 November 2026, and maps run from about 50 pages to over 20,000.
- No engine publishes a source context score, but Google's leaked documentation stores a siteFocusScore and a siteRadius, and a Google engineer's 2025 notes describe sending narrow technical queries to a more technical site.

## The three decisions, in plain words

Source context is why your site exists and how it earns. The central entity is the one thing every page on the site is about. The central search intent is what your visitors come to do with that thing, written as a verb plus the entity. All three are decided before a map has its first row.

| Decision | The question it answers | Written as | Made-up salon example |
| --- | --- | --- | --- |
| Source context | Why does this site exist, and how does it earn? | One sentence | A Portland salon that earns from six grooming services |
| Central entity | What is every page about? | A noun | Dog grooming |
| Central search intent | What do visitors come to do with it? | A verb plus the central entity | Get a dog groomed in Portland and keep it groomed between visits |

The salon is the made-up business from [How to build a topical map](https://frameworkseo.com/learn/topical-maps/how-to-build-a-topical-map). Step 2 of that lesson asks for one sentence that names the subject. This lesson takes that sentence apart. If the word map itself is new to you, start with [What is a topical map?](https://frameworkseo.com/learn/topical-maps/what-is-a-topical-map)

![Source context and central entity combine into one central search intent, which then tests whether each page goes on the map.](https://frameworkseo.com/learn/source-context-central-entity/three-decisions.webp)

*Two inputs make one sentence, and every page on the map has to pass it.*

### Where the three terms come from

The terms are Koray Tuğberk Gübür's. He is the SEO consultant whose case studies, from February 2021 on, made the topical map a standard deliverable, and I learned this part of my work from him. [Koray Tuğberk Gübür's topical map framework](https://frameworkseo.com/learn/semantic-seo/koray-topical-map-framework) tells the whole story. Here I take only the first three parts.

In [a post from August 2025](https://www.linkedin.com/posts/koray-tugberk-gubur_outer-section-of-topical-maps-why-it-matters-activity-7361187658343006208-QnWP) he names "the 5 fundamental parts of a functioning topical map: Source Context, Central Entity, Central Search Intent, Core Section, and Outer Section." The first three are decisions. The last two are the halves of the map those decisions produce.

He has published a full definition of only one of the three. In [his 2023 article on expanding a topical map](https://www.holisticseo.digital/seo-research-study/topical-map) he writes: "Source Context is the perspective and function of the website and the brand on the open web that involves monetization, and existing purpose." The same article says the source context sets the angle of every page, the direction of the content briefs, and "the topical map's core and outer sections".

> **What is his, and what is second-hand:** Koray's own [2023 article on entities and attributes](https://www.holisticseo.digital/seo-research-study/entity-attribute-value) says central entities "are explained in the Semantic SEO Course", his paid course. So the next two definitions come from notes a student of that course published. I paraphrase them and do not quote them as his words.

His course, as those notes describe it, defines the central entity as what the website is mainly about: the topic that shows on every page and in the text repeated across the site, such as the header and the footer. It defines the central search intent as the source context and the central entity joined into one line, built from a verb and a noun.

### What the three are not

- **Not the search intent of one query.** Informational, commercial and transactional describe a single search. The central search intent describes the whole site, and every page is tested against it.
- **Not the main entity of one page.** Entity SEO guides use "central entity" for the subject of a single page. In Koray's framework it is the subject of the whole site, and each page's own topic sits inside it. [Semantic SEO in plain words](https://frameworkseo.com/learn/semantic-seo/what-is-semantic-seo) covers entities on a page.
- **Not a keyword.** "Dog grooming portland" is a search. Dog grooming is the thing the salon's site is about.

## Why two sites selling the same thing need different maps

Two sites can share a central entity and still need different maps, because the source context decides which pages belong and what angle each one takes. The product is the same. The reason each site exists is not.

Koray's own example is a single question. He writes that "what is an electronic bicycle" "should be answered differently according to the source context". An electric car site, a renewable energy site and an engineering site each need their own definition of the bike, their own internal links and their own angle (2023).

### Two shops that both sell graphic tees

Here is the same idea with something you can buy. Two shops, both made up, both sell graphic tees, so the central entity is the same for both.

- **The print shop** designs its own tees and prints each one to order. It earns from its designs and from custom orders.
- **The vintage reseller** sells one-of-a-kind secondhand tees. It earns from finding, dating and pricing rare pieces.

![Two made-up shops share the entity graphic tees but plan different pages: printing guides versus vintage checks.](https://frameworkseo.com/learn/source-context-central-entity/two-tee-shops.webp)

*Only two of the eight pages belong on both maps.*

| Page | Print shop | Vintage reseller |
| --- | --- | --- |
| Graphic tees collection | Yes | Yes |
| How to wash a graphic tee | Yes | Yes |
| Screen printing vs direct-to-garment printing | Yes: it explains what the buyer pays for | No: nothing it sells is printed to order |
| Custom tees for teams and events | Yes: custom orders are part of what it sells | No: one-of-a-kind stock cannot fill an order of 40 |
| How to tell if a vintage tee is real | No: a new print has no history to check | Yes: it is the reseller's whole promise |
| Single stitch vs double stitch tees | No: it has nothing to date | Yes: buyers use the stitching to date a tee |

Copy the reseller's map onto the print shop's site and you get pages that bring in readers it cannot sell to. That is the cost of a wrong source context: months of work on pages that never earn.

### Libraries made the same decision long before SEO

Library science has the same rule under another name. In the [ISKO Encyclopedia of Knowledge Organization](https://www.isko.org/cyclo/subject), Birger Hjørland (2016) describes a feminist library that buys a book about Napoleon. It indexes the book for what it says about women in Napoleon's time, because that is why it bought it. A history library would index the same book another way. His conclusion: "the purpose or perspective of a specific collection should ideally be reflected in its classification or indexing." He calls one form of this "policy based indexing".

Swap "collection" for "website" and you have source context. The same book, like the same tee, gets a different place depending on why the collection exists. [Where topical maps came from](https://frameworkseo.com/learn/semantic-seo/where-topical-maps-came-from) traces the other library ideas behind the map.

## Worked examples: a store, a service, a software company and a publisher

The table gives all three decisions for four kinds of business. The store and the restoration company are real client maps we delivered, anonymized, and the three decisions are written as I read them from each finished map. The software company and the magazine are made up.

| Business | Source context | Central entity | Central search intent |
| --- | --- | --- | --- |
| Online store for graphic tees (real, 391-page map) | Sells graphic tees and apparel online in the US | Graphic tees | Find a graphic tee that fits your style, and wear it well |
| Water and fire damage restoration company (real, 173-page map) | Paid per job to repair damage in homes, often through insurance claims | Water and fire damage in a home | Get the damage fixed and the insurance claim paid |
| Job software for restoration crews (made up) | A monthly subscription per company | The restoration job | Run a restoration job from first call to paid claim |
| Home repair magazine (made up) | Ads, and commissions on products it reviews | Home maintenance | Keep a home in good repair and know when to call a pro |

Look at the central entity column. The two businesses that sell something specific, tees and repairs, have narrow central entities. The magazine earns from attention, so its central entity is wide. That width comes back later as map size.

### One topic, three angles

The bottom three rows share a subject: water damage. Each of them needs a page about dehumidifiers. None of them can use the others' page.

![Three sites plan three different dehumidifier pages for the same water damage subject.](https://frameworkseo.com/learn/source-context-central-entity/water-damage-three-sites.webp)

*The first title is copied from a delivered client map; the other two sites are made up.*

The restoration company's real map has "Dehumidifier for Water Damage: How Long to Run It", a search with 6,600 a month in the US, and the page links up to the company's water mitigation service. The software company would write about logging dehumidifier readings in its app, because the reader is a crew lead. The magazine would review dehumidifiers and earn a commission on each sale. Same topic, three source contexts, three different pages.

The restoration map also has a guide called "Does Homeowners Insurance Cover Water Damage?" It is there because the central search intent includes getting the claim paid. Take the insurance part out of the intent and that row has no reason to be on the map.

> **Where this goes next:** Which of these pages earn and which support them is the split into [core section and outer section](https://frameworkseo.com/learn/semantic-seo/core-and-outer-sections). That lesson walks through a full branch of both real maps, row by row. For the business types themselves, see [topical maps for ecommerce](https://frameworkseo.com/learn/topical-maps/topical-map-for-ecommerce) and [topical maps for local and service businesses](https://frameworkseo.com/learn/topical-maps/topical-map-for-local-and-service-businesses).

## How to find your own three in an hour

Give each decision 15 minutes, then spend the last 15 testing all three against 20 pages you might build. Write each decision as one line. A page that fails the test stays off the map.

![A one-hour timeline: 15 minutes for each of the three decisions, then 15 minutes testing 20 pages.](https://frameworkseo.com/learn/source-context-central-entity/one-hour-worksheet.webp)

*The last block is the one most people skip, and the one that changes the map.*

### Minutes 0 to 15: the source context

1. List every way money reaches the business, largest first: services, products, subscriptions, ads, commissions.
2. Write what a searcher can check about you: who you are, where you are, what you have done.
3. Fill in one sentence: [who we are] earns from [how] by [doing what] for [whom]. The salon's: a Portland dog grooming salon that earns from six grooming services for local dog owners.

If your only honest answer is "traffic", stop and think again. A magazine can earn from ads, but its reason to exist is still the help it gives. A site with no reason beyond the visit has nothing to decide which pages belong.

### Minutes 15 to 30: the central entity

Read your menu and the titles of the pages that earn. Write the noun they share. Then test it three ways:

- **Too narrow** if a page you clearly need would not be about it. "Nail trims" is too narrow for the salon, which sells five other services.
- **Too broad** if whole parts of it are things you would never sell or advise on. "Dogs" is too broad: food, training and vet care are not the salon's business.
- **Known outside your site** if a search shows a knowledge panel or an encyclopedia article for it. Dog grooming has its own Wikipedia article, so engines already know it as a thing.

### Minutes 30 to 45: the central search intent

Write a verb plus the central entity. Add a second verb only if the source context earns from it. Then check that it covers more than the sale: "book a groom" on its own would drop every at-home care guide that brings a nail trim customer to the salon.

Then turn the sentence into a question. Joseph Novak and Alberto Cañas, who developed concept maps, write in [their 2008 paper](https://cmap.ihmc.us/Publications/ResearchPapers/TheoryUnderlyingConceptMaps.pdf) that "every concept map responds to a focus question". They also saw learners drift and build a map "that may be related to the domain, but which does not answer the question". Topical maps drift the same way. The salon's question: how do Portland dog owners keep their dog well groomed, and when do they need a groomer?

> **The verb changes the map:** Novak and Cañas found that the question "What are plants?" leads to a more classifying concept map than "Why do we need plants?" Same noun, different verb, different map. That is why the central search intent needs a verb, and why "electric bikes" alone is not one.

### Minutes 45 to 60: the test

List 20 pages you might build. For each one ask a single question: does it help someone do what the central search intent says? Keep the page if it does. Merge it into a page you already have if that page covers it. Cut it if neither. The salon keeps "Dog nail bleeding: what to do" and cuts "Best dog food brands".

[Download the one-hour worksheet](https://frameworkseo.com/learn/source-context-central-entity/source-context-worksheet.csv) (CSV, opens in Google Sheets or Excel. Every question, how to answer it, a filled-in salon example and 20 rows for your page test.)

### When your source context disagrees with Google's results

A reader asked Koray about this under his 2023 article. He runs an engineering blog, and Google answers "what is an electronic bicycle" from the angle of electric vehicles. Which angle should he take? The question has no reply on the page. My rule: answer the reading Google shows first, because that is what the searcher came for. Then add the angle only your site can give. If your angle cannot answer the main reading at all, the search belongs on someone else's map.

Deciding whether a search gets a page or only a section is its own call, covered in [Page or section](https://frameworkseo.com/learn/topical-maps/page-or-section).

## How the three decisions set the size of your map

The three decisions set how many pages a map needs. Our median growth plan was 159 pages on 5 November 2026, and maps run from about 50 pages to over 20,000. That spread comes from the central entity and the source context, not from how hard anyone works.

When a client asks me how big their map will be, my answer is that it depends on "the size of the company, what central entity they have that needs that sort of coverage, how many services they have, and so on." In practice three things multiply:

1. **How wide the central entity is.** Graphic tees has kinds, fits, styles and care questions. Home maintenance has every room, every system, every fault and every repair in a house.
2. **How many things the business sells, and where.** Every service, product line and place served adds pages that earn, plus the guides that feed them.
3. **How deep the source context lets you go.** A store goes deep into its own products. A magazine goes wide across its whole subject.

![Bar chart: median growth plan 159 pages, a restoration map 173 pages, a graphic tee store map 391 pages.](https://frameworkseo.com/learn/source-context-central-entity/map-sizes.webp)

*Two delivered maps beside our median on 5 November 2026; the full range runs from about 50 pages to over 20,000.*

The two real maps show it. The restoration company sells a handful of services to homeowners, and its map is 173 pages. The tee store sells many collections and supports each one with guides, and its map is 391 pages.

A wrong central entity changes the number before a single page is written. Call the salon's subject "dogs" and its map grows into hundreds of pages on food, training and health that the salon cannot earn from. Call it "nail trims" and the map loses the five services that pay most of the bills.

## What engines show about modeling a site's focus

No search engine publishes a "source context" score. But Google's guidance, a Google engineer's interview notes, Google's leaked documentation, two patents and Bing's guidelines all describe judging a site by what it is focused on. Here is the evidence, strongest first.

| Source | What it says | What it shows |
| --- | --- | --- |
| Google's helpful content guidance (updated October 2026) | Asks "Does your site have a primary purpose or focus?" and warns against "producing lots of content on many different topics" | Google tells site owners that focus counts |
| Notes of a call with Google engineer HJ Kim (court exhibit, 2025) | Quality is "largely related to the site rather than the query"; a narrow technical query can go to "a quality site that is more technical" | Google judges the site, and the kind of site, not only the page |
| Google's leaked API documentation (2024) | Stores a siteFocusScore and a siteRadius for a site | A focus measure exists; its weight is not public |
| Google patent application US20200050707A1 (filed 2018, abandoned) | Classifies a website by its distance from groups of websites in one knowledge domain | An idea Google wrote down, not a system known to run |
| Microsoft patent US11874882B2 (granted January 2024) | Ranks a page by "a topical authority score assigned to the website with respect to a topic" | Bing's owner patented a per-site, per-topic score |
| Bing Webmaster Guidelines (2026) | "Focus Each URL on a Single Topic" and "Define Entities Clearly and Consistently" | Another engine asks for the same discipline |

### The engineer's notes and the electric bicycle

The second row is where Koray's idea and Google's own words meet. Koray's 2023 rule says an engineering site and an energy site must answer the same electric bicycle question differently. HJ Kim, who started Google's page quality team, was interviewed for the US v. Google case in February 2025. The [notes of that call](https://www.justice.gov/atr/media/1398871/dl) say a site "may have high quality but general information so a query interpreted as seeking very narrow/technical information may be used to direct to a quality site that is more technical."

That is source context seen from Google's side: the kind of site you are decides which questions Google sends you. The notes are a summary of one call, made public in May 2025 in US v. Google, so read them as a snapshot, not as a description of today's ranking.

### The leaked fields

The 2024 leak of Google's internal API documentation describes [siteFocusScore](https://hexdocs.pm/google_api_content_warehouse/0.4.0/GoogleApi.ContentWarehouse.V1.Model.QualityAuthorityTopicEmbeddingsVersionedItem.html) as a "Number denoting how much a site is focused on one topic" and siteRadius as "The measure of how far page_embeddings deviate from the site_embedding". In plain words: Google stores a center for your site, and how far each page strays from it.

![A focused site with pages near its center beside a scattered site with off-topic pages far out.](https://frameworkseo.com/learn/source-context-central-entity/site-focus-radius.webp)

*An illustration of the two leaked fields, not real data: off-topic pages widen the radius.*

Google confirmed the documents were real and told [The Verge](https://www.theverge.com/2024/5/29/24167407/google-search-algorithm-documents-leak-confirmation): "We would caution against making inaccurate assumptions about Search based on out-of-context, outdated, or incomplete information." So the fields exist, and how much they count is not public. What they do show is that an off-topic page can be seen as off-topic. That is the page a clear source context keeps off the map. How focus turns into rankings over time is the subject of [topical authority](https://frameworkseo.com/learn/topical-maps/topical-authority).

> **Watch out:** A patent or a leaked field is not proof that a system ranks your site with it. Read this evidence as "engines can see a site's focus", never as a formula you can tune.

## The usual ways people get the three wrong

Most mistakes come from skipping the three decisions, or from writing them down and never using them. These are the seven we check for on every map.

| Mistake | How it shows | The fix |
| --- | --- | --- |
| 1. The source context is "get traffic" | Posts on anything with search volume | Name how the business earns, then cut what cannot lead there |
| 2. The central entity is a keyword | "Dog grooming portland" written as the subject | Use the thing, not the search: dog grooming |
| 3. The central search intent has no verb | "Electric bikes" written as the intent | A verb plus the entity: buy, run, fix, learn |
| 4. The intent only names the sale | Only product and price pages on the map | Add what buyers do before and after the sale |
| 5. A competitor's map copied | Pages that bring readers you cannot sell to | Check the competitor's source context before borrowing a row |
| 6. An off-context section bolted on | A news site starts reviewing products | A clearly separate section or sub-brand, or leave it out |
| 7. Written down, never used | Rows nobody tested; the site never says what it is | Test every row, and show the three on the site |

Mistake 6 is Koray's own warning. In [a February 2024 post](https://medium.com/@ktgubur/3-suggestions-about-topical-authority-1cd73963a9cb) he writes that a news site can open an affiliate section and rank for it, but "if products are irrelevant to their source context, they can lose their rankings after a Broad Core Algorithm Update." His way out for news sites is a sub-brand with its own authors and design, sometimes on its own subdomain.

Mistake 7 has two halves. The three decisions should filter every row, and they should show on the site. From the same post: "Reflect the overall topicality of your source site-wide with universal elements such as footer, header, sidebar, and company definitions."

The pages that fail the test are often already live. As I tell clients, "any website that's worked with a bad agency before has a large percentage of off-topic posts", and "if a page does not have a purpose, it does not deserve to be on the website." Before you delete one, check whether it can be merged into a page that is more relevant. [Pruning and merging](https://frameworkseo.com/learn/topical-maps/pruning-and-merging) covers how.

## Where the three decisions go next

With the three written down, the map can start. The central search intent splits the pages into the ones that earn and the ones that support them: that is [Core section and outer section](https://frameworkseo.com/learn/semantic-seo/core-and-outer-sections), the next lesson. The full build, step by step, is [How to build a topical map](https://frameworkseo.com/learn/topical-maps/how-to-build-a-topical-map).

[Koray Tuğberk Gübür's topical map framework](https://frameworkseo.com/learn/semantic-seo/koray-topical-map-framework) covers the rest of his method, and [Build a topical map with Claude](https://frameworkseo.com/learn/claude-for-seo/build-a-topical-map-with-claude) shows how to draft one with an AI model. Decide the three yourself first. A model can draft rows. It cannot know why your business exists.

If you would rather have the three decided and the whole map built for your site, that is what our [growth plan](https://frameworkseo.com/) is.

## Common questions

### Can a site have more than one central entity?

Yes, when the subject is two things that only make sense together. Notes from Koray Tuğberk Gübür's course give a German visa consultancy as an example, with Germany and visa as one paired central entity. A site that runs two unrelated businesses is different: it needs two maps, or a clearly separate section with its own identity.

### Can you change your source context later?

Yes, and the homepage is where it shows first. Koray describes a lifestyle publisher, TheCoolist.com, with a topical map that was focusing on whatever is searched most. He changed its homepage for what he calls Source Context Re-definition and reports about 9,000 extra clicks a day (2023, his own numbers).

### Should the three decisions appear on the website itself?

Yes. Put the source context on the home and about pages in one plain sentence, and the central entity in the parts every page shares: the header, the footer and the site's tagline. Koray's 2024 advice is to reflect the site's subject in the footer, header, sidebar and company definitions.

## Sources

- [How to Expand a Topical Map for Higher Topical Authority?](https://www.holisticseo.digital/seo-research-study/topical-map), Koray Tuğberk Gübür, Holistic SEO & Digital, 2023
- [Outer section of topical maps: why it matters](https://www.linkedin.com/posts/koray-tugberk-gubur_outer-section-of-topical-maps-why-it-matters-activity-7361187658343006208-QnWP), Koray Tuğberk Gübür, LinkedIn, 2025
- [Importance of Entity, Attribute, Value (EAV) Architecture for SEO](https://www.holisticseo.digital/seo-research-study/entity-attribute-value), Koray Tuğberk Gübür, Holistic SEO & Digital, 2023
- [3 suggestions about topical authority](https://medium.com/@ktgubur/3-suggestions-about-topical-authority-1cd73963a9cb), Koray Tuğberk Gübür, Medium, 2024
- [Subject (of documents)](https://www.isko.org/cyclo/subject), Birger Hjørland, ISKO Encyclopedia of Knowledge Organization, 2016
- [The Theory Underlying Concept Maps and How to Construct and Use Them](https://cmap.ihmc.us/Publications/ResearchPapers/TheoryUnderlyingConceptMaps.pdf), Joseph D. Novak and Alberto J. Cañas, IHMC, 2008
- [Creating helpful, reliable, people-first content](https://developers.google.com/search/docs/fundamentals/creating-helpful-content), Google Search Central, 2026
- [Interview with H.J. Kim notes, remedies exhibit PXR0356, US v. Google LLC](https://www.justice.gov/atr/media/1398871/dl), U.S. Department of Justice, 2025
- [QualityAuthorityTopicEmbeddingsVersionedItem (leaked Content Warehouse documentation)](https://hexdocs.pm/google_api_content_warehouse/0.4.0/GoogleApi.ContentWarehouse.V1.Model.QualityAuthorityTopicEmbeddingsVersionedItem.html), google_api_content_warehouse 0.4.0, hexdocs, 2024
- [Google confirms the leaked Search documents are real](https://www.theverge.com/2024/5/29/24167407/google-search-algorithm-documents-leak-confirmation), The Verge, 2024
- [Website representation vector to generate search results and classify website (US20200050707A1)](https://patents.google.com/patent/US20200050707A1/en), Google LLC, Google Patents, 2020
- [Extracting key phrase candidates from documents and producing topical authority ranking (US11874882B2)](https://patents.google.com/patent/US11874882B2/en), Microsoft Technology Licensing, Google Patents, 2024
- [Bing Webmaster Guidelines](https://www.bing.com/webmasters/help/webmaster-guidelines-30fba23a), Microsoft Bing, 2026