# Topical map vs topic cluster vs keyword list

Three plans that sit inside each other. Here is what each one holds, what it leaves out, and what breaks when you use the smaller one as your whole plan, shown on one made-up store with real search data.

By Kevin Jeppesen, AI growth engineer. Founder, FrameworkSEO & Ascendily. Updated 2026-11-13. Source: https://frameworkseo.com/learn/topical-maps/topical-map-vs-topic-cluster

**Topic cluster:** A topic cluster is a group of pages on one subject: a pillar page that covers the subject broadly and cluster pages that each cover one part of it, linked to and from the pillar.

## Key takeaways

- A keyword list records searches and how often people make them, a topic cluster groups one subject's pages around a pillar page, and a topical map plans every page the site needs with its type, links and publishing order.
- The three nest inside each other: a topical map contains clusters, and a cluster is built from the searches on a keyword list.
- On a made-up cast iron store with real US search data, the same 12 searches became 12 rows on a list, a 7-page cluster that left out the two searches the store sells to, and an 8-page map with 2 pages that sell.
- Google's own keyword data gave "cast iron seasoning" and "seasoning cast iron" the same 74,000 searches a month, which fits how it counts close wordings as one, so a list that gives every row a page plans two rival pages for one need.
- A library thesaurus already names the three relations a map needs: same meaning means one page, broader and narrower means a parent page and the pages under it, and related means a link across groups.

## Which one you need: the short answer

Use a keyword list to see what people search and how often. Use a topic cluster to plan one subject's pages around a pillar page. Use a topical map to plan the whole site: every page it needs, what type each page is, how the pages link, and the order they go live.

The three are not rivals. They sit inside each other. A cluster is built from searches on a list, and a map is made of clusters plus the decisions a cluster leaves out. The trouble starts when the smaller plan is used as if it were the bigger one.

|  | Keyword list | Topic cluster | Topical map |
| --- | --- | --- | --- |
| **A list of** | Searches and their monthly numbers | Pages on one subject | Every page the site needs |
| **Decides** | Nothing about pages | A pillar page, one page per part, links to and from the pillar | Each page's type, its links across groups, the order, and what is out |
| **Leaves out** | Pages, links, order | The other subjects, the pages that sell, the order | The writing itself |
| **Where the name comes from** | No single origin | HubSpot, May 2017 | Popularized in SEO by Koray Tuğberk Gübür, February 2021 |
| **Breaks when used as the whole plan** | Rival pages for one need | Groups that never sell and never connect | Not applicable: it is the whole plan |

![A keyword list nested inside a topic cluster, nested inside a topical map, each adding its own decisions](https://frameworkseo.com/learn/topical-map-vs-topic-cluster/nested-plans.webp)

*Each plan keeps everything in the one inside it and adds the decisions that one leaves out.*

If you only want one of the three, start from the map. It is the only one of the three that tells you what to publish next.

## What a keyword list, a topic cluster and a topical map each hold

Each plan holds one more kind of decision than the one before it. Here is what each one is, where the name came from, and the one thing it is good for.

### A keyword list: searches and their numbers

A keyword list is a table of searches with a number next to each one, usually the searches a month in one country. It tells you how people word a need and how many of them do it. It says nothing about pages.

One detail changes how you read every list. Google's keyword data already folds close wordings together. The [Google Ads API documentation](https://developers.google.com/google-ads/api/docs/keyword-planning/generate-keyword-ideas) describes each keyword idea as a "canonical keyword idea text representing the suggestion and its combined close_variants" (2026). So when two rows show the exact same number, they are often one need counted twice.

### A keyword cluster: searches one page can answer

A keyword cluster is a small group of searches that one page can answer. "Cast iron seasoning" and "seasoning cast iron" are one keyword cluster. It is not the same thing as a topic cluster, and mixing the two names up is where much of the confusion starts. The test for whether two searches share a page is in [when a search deserves its own page](https://frameworkseo.com/learn/topical-maps/page-or-section).

### A topic cluster: a pillar page and the pages around it

A topic cluster is a group of pages on one subject. One pillar page covers the subject broadly. Each cluster page covers one part of it in depth. The cluster pages link to the pillar, and the pillar links back.

HubSpot named the model. Its research team published [Topic Clusters: The Next Evolution of SEO](https://web.archive.org/web/20180109215341/https://research.hubspot.com/topic-clusters-seo) on 10 May 2017, after experiments that two of its staff started in 2016. HubSpot's [current guide](https://blog.hubspot.com/marketing/topic-clusters-seo) says topic clusters are also known as "content clusters" or the "hub and spoke" model. The pillar page itself has its own lesson: [pillar pages](https://frameworkseo.com/learn/topical-maps/pillar-pages).

### A topical map: every page, its type, its links and its order

A topical map plans every page a site needs to cover its subject. Each page carries its type (guide, product, collection, comparison and so on), its main search, its links, including links to pages in other groups, and its place in the publishing order. A good map also writes down what is out. The full definition, with a real map, is in [what is a topical map](https://frameworkseo.com/learn/topical-maps/what-is-a-topical-map).

Koray Tuğberk Gübür is the person most often credited with making the topical map a standard step in SEO. The first step in his February 2021 case study for Oncrawl reads: "Create a Topical Map Before Starting Publishing the First Article." In 2024 he put the difference from a list plainly: ["creating a topical map is not merely listing keywords or entities"](https://medium.com/@ktgubur/korays-agents-generative-ai-agents-for-semantic-seo-and-topical-authority-d4b247fac72a). His whole framework, in plain words, is in [Koray's topical map framework](https://frameworkseo.com/learn/semantic-seo/koray-topical-map-framework). The name has an older cousin, the Topic Maps standard from library and data science, covered in [where the topical map came from](https://frameworkseo.com/learn/semantic-seo/where-topical-maps-came-from).

> **Flat, two layers, many layers:** Information science sorts these structures by shape. In Marcia Zeng's 2008 scheme, quoted in the [ISKO Encyclopedia of Knowledge Organization](https://www.isko.org/cyclo/ontologies), knowledge organization systems range "from flat to two-dimensional to multiple-dimensional". A keyword list is flat. A topic cluster has two layers: pillar and pages. A topical map has many, because every page also has a type, links across groups and a place in time.

## One subject three ways: a cast iron store as a list, a cluster and a map

The clearest way to see the difference is to plan the same subject three times. Here is a made-up online store that sells cast iron pans, with 12 real searches and their real US numbers from Google Ads data, pulled on 13 November 2026.

![Twelve cast iron searches as a 12-row keyword list, a 7-page topic cluster that leaves out the two searches the store sells to, and an 8-page topical map](https://frameworkseo.com/learn/topical-map-vs-topic-cluster/cast-iron-three-ways.webp)

*Same twelve searches, three plans: 12 rows, 7 pages, then 8 pages with types, links and an order.*

### As a keyword list: 12 rows

The list sorts the 12 searches by volume, from "cast iron skillet" at 165,000 a month down to "can you use soap on cast iron" at 1,300. The rows add up to 540,900 searches a month. That total is wrong: "cast iron seasoning" and "seasoning cast iron" both show exactly 74,000, so 74,000 of it is one need counted twice.

### As a topic cluster: 7 pages

Built the way most cluster guides teach, the core topic is cast iron care. A pillar page, "Cast iron care, the complete guide", links to six posts: seasoning, reseasoning, cleaning, soap, rust and the best oil. Every post links back.

Four searches fall outside the cluster because they are not about care: cast iron skillet, cast iron dutch oven, best cast iron skillet and cast iron vs carbon steel. Together they are 257,400 searches a month, including the two searches the store actually sells to.

### As a topical map: 8 pages, 2 that sell, 3 batches

| Page | Type | Searches it answers (US a month) | Main link to | Batch |
| --- | --- | --- | --- | --- |
| Cast iron skillets | Collection, sells | cast iron skillet (165,000) | How to season cast iron | 1 |
| Cast iron Dutch ovens | Collection, sells | cast iron dutch oven (74,000) | How to clean cast iron | 1 |
| How to season cast iron | Guide | how to season cast iron (60,500); cast iron seasoning and seasoning cast iron (74,000, one need); how to reseason cast iron (4,400, a section) | Cast iron skillets | 1 |
| How to clean cast iron | Guide | how to clean cast iron (49,500); can you use soap on cast iron (1,300, a section) | How to season cast iron | 2 |
| How to remove rust from cast iron | Guide | 9,900 | How to season cast iron | 2 |
| Best oil for seasoning cast iron | Guide with a comparison table | 9,900 | How to season cast iron | 2 |
| Best cast iron skillet | Buying guide | 14,800 | Cast iron skillets | 3 |
| Cast iron vs carbon steel | Comparison | 3,600 | Cast iron skillets | 3 |

Four searches do not get a page of their own. Two are extra wordings of the seasoning need. Two are smaller questions that a bigger page answers in a section. The map also writes down one thing that is out: recipes, such as "how to cook steak in cast iron" (3,600 a month), wait until the care and buying pages are live.

The order follows the links. The two collections and the seasoning guide go first, because most other pages link to them. The buying guide and the comparison go last, because they mostly link out. The full rule is in [publishing order](https://frameworkseo.com/learn/topical-maps/publishing-order).

[Download the cast iron example](https://frameworkseo.com/learn/topical-map-vs-topic-cluster/cast-iron-list-cluster-map.csv) (CSV, 13 rows: every search with its volume, where it sits on the list, in the cluster and on the map, its page type, main link and batch.)

## What breaks when a keyword list is your plan

A keyword list breaks the moment each row becomes a page. Three things go wrong, and all three showed up in the cast iron data.

### Two wordings of one need become two rival pages

"Cast iron seasoning" and "seasoning cast iron" are two rows with 74,000 searches each. Google's first page for the two wordings shared 3 of the same results on 13 November 2026. A list that gives each row a page plans two pages that answer one need and compete with each other. That problem has its own lesson: [keyword cannibalization](https://frameworkseo.com/learn/topical-maps/keyword-cannibalization).

The pages that win do the opposite. Lodge, which makes cast iron pans, ranked one page, its cleaning and care guide, in Google's top 4 for three wordings that day: first for "how to reseason cast iron", third for "seasoning cast iron" and fourth for "how to season cast iron". One strong page collected three rows of the list.

Bing's guidelines point the same way. The [Bing Webmaster Guidelines](https://www.bing.com/webmasters/help/webmaster-guidelines-30fba23a) tell sites to "Focus Each URL on a Single Topic" and warn that "Duplicate URLs dilute signals" (2026).

### The list hides what kind of page each search wants

A row says "cast iron skillet, 165,000". It does not say that 5 of the 9 results on Google's first page for that search were shop pages: four collection pages and one product page. A blog post written from that row would compete with stores on a page of stores. The page type is a column the list never had. How to read it from the results is step 6 of [how to build a topical map](https://frameworkseo.com/learn/topical-maps/how-to-build-a-topical-map).

### Nothing on the list knows about anything else

Every row is alone. There is no parent page, no link between the rust guide and the seasoning guide, and no reason to publish one before the other. So the order defaults to volume, and the biggest, hardest search gets written first. How the links should run once you have pages is in [internal links inside a topical map](https://frameworkseo.com/learn/topical-maps/internal-links-in-a-topical-map).

> **Read a list for what it is good at:** Keep the list. It is the best record you have of how people word a need. Just never let a row become a page until you have checked whether another row is the same need.

## What breaks when a topic cluster is your whole plan

A topic cluster fixes the list's worst problem: pages now belong to a group and link to a pillar. It breaks in a different place. A cluster is planned one subject at a time, so nothing decides what sits between clusters.

### The model was built for a blog, so the pages that sell sit outside it

HubSpot's 2017 model was a content model: a pillar page and the posts around it. Apply it to the cast iron store and the cluster covers care, while the skillet and Dutch oven searches, 239,000 a month between them, belong to no cluster at all. The visits arrive on guides that link to nothing a reader can buy.

A map starts from the other end. It begins with what the site exists to do, which Koray calls the source context, and builds the groups around [the core pages and the pages that support them](https://frameworkseo.com/learn/semantic-seo/core-and-outer-sections). The first decision is explained in [source context and central entity](https://frameworkseo.com/learn/semantic-seo/source-context-central-entity).

### Grouping by sameness misses that one search can need two kinds of page

Many cluster tools group searches by the results they share. By that rule, "cast iron skillet" and "best cast iron skillet" belong together: they shared 4 of their 9 first-page results on 13 November 2026. But the kinds of page Google showed were different. For "cast iron skillet", 5 of 9 were shop pages. For "best cast iron skillet", 1 of 9 was.

![Two cast iron searches share 4 of 9 Google results but show different kinds of page, so they get two pages](https://frameworkseo.com/learn/topical-map-vs-topic-cluster/same-results-two-pages.webp)

*Shared results say the needs overlap. The kinds of page Google shows say the store needs two shapes.*

A store needs both: a collection to buy from and a buying guide to choose with. Only a plan that knows what the site sells can make that call. Koray's name for the angle a page takes is the contextual vector. In his 2021 Oncrawl study he wrote that "a topic can be 'earthquake' and a context can be 'comparing earthquakes', 'guessing earthquakes', or 'chronology of earthquake'", and pointed out that Healthline had more than 265 articles about apples. A cluster groups by topic. A map gives every page its angle.

### Clusters have no borders and no order between them

"Best oil for seasoning cast iron" fits a care cluster and a buying cluster equally well. Plan two clusters separately and both can claim it. A map gives every search exactly one home and writes the other cluster's link to it as a link, not a second page.

Clusters also have no order between them. Which cluster goes first, and which pages in it, is a decision no single cluster can make.

> **What I see in maps buyers build themselves:** When buyers show me a map they made, three things come up. The pages "do not build on each other". There are "a lot of off-topic things that do not deserve to be there when they could have been part of another post". And "a query does not deserve its own page": many searches are a section of a page, not a page. All three are list or cluster habits carried into a map.

## How librarians sorted these three plans long before SEO

Libraries solved this problem decades before search engines. The standard for building controlled vocabularies, [ANSI/NISO Z39.19](https://www.niso.org/publications/ansiniso-z3919-2005-r2010), covers four structures: "lists, synonym rings, taxonomies, and thesauri" (2005, reaffirmed 2010). That is close to the same ladder. A keyword list is a list. A keyword cluster is a synonym ring. A topic cluster is a small taxonomy. A topical map is closest to a thesaurus.

### The three relations every thesaurus records

A thesaurus links its terms with three kinds of relation. In the [ISKO Encyclopedia of Knowledge Organization](https://www.isko.org/cyclo/thesaurus), Stella Dextre Clarke names them as equivalence, hierarchical and associative, tagged by long convention as USE and UF (use for), BT and NT (broader and narrower) and RT (related) (2019). Each one is a page decision.

| Relation | Tags | Page decision | Cast iron example |
| --- | --- | --- | --- |
| **Same meaning** | USE, UF | One page; the other wording goes on it | seasoning cast iron and cast iron seasoning: one guide |
| **Broader and narrower** | BT, NT | A parent page and the pages under it | Cast iron care above how to remove rust from cast iron |
| **Related** | RT | A link across groups | The skillet collection links to the seasoning guide |

![The three thesaurus relations and the page decision each one makes: one page, parent and child pages, a link](https://frameworkseo.com/learn/topical-map-vs-topic-cluster/thesaurus-to-pages.webp)

*A topic cluster records one of the three relations. A topical map records all three, then adds the order.*

This is the cleanest test I know for a plan. A keyword list records none of the three relations. A topic cluster records one: broader and narrower. A topical map records all three, then adds the one thing a thesaurus never needed, which is the order the pages go live.

The idea has a long track record. The Library of Congress has kept its [subject headings](https://id.loc.gov/authorities/subjects.html) since 1898, and its headings carry broader, narrower and related terms. Giving each book one preferred heading is close to what SEO later called keyword mapping.

## Content cluster, keyword map, content map, content hub, silo: what each word means

Five more words get used for one of the three plans. Some mean the same thing under another name. Two mean something else entirely, and one means two opposite things.

| Word | What it means | Closest of the three |
| --- | --- | --- |
| **Content cluster** | Another name for a topic cluster, per HubSpot's own guide. | Topic cluster |
| **Keyword map, keyword mapping** | Assigning each search to the page that should rank for it, usually in a spreadsheet. | A list with a page column; one step short of a cluster |
| **Content map** | Matching content to each kind of buyer and each stage before they buy. | None: a marketing plan, not a site plan |
| **Content hub** | A page that gathers and links a subject's pages; also HubSpot's product of that name. | The pillar of a topic cluster |
| **Silo** | A section of a site whose pages link mostly to each other; or, in other writing, content cut off from everything else. | Topic cluster, with links kept inside it |

### Content map: a different job with a similar name

"Content map" gets 1,000 US searches a month, more than "topical map" (590) or "topic cluster" (260). On 13 November 2026 every one of Google's top five results for it was about content mapping: deciding what content each stage of a customer's journey needs. That is useful, and it is not a plan of pages. A content map can sit on top of a topical map, never in place of it.

### Content hub: a page, and now a product

The "hub" in search goes back to Jon Kleinberg's 1999 paper [Authoritative sources in a hyperlinked environment](https://www.cs.cornell.edu/home/kleinber/auth.pdf), which describes hubs as pages that point to many good sources on a topic. In SEO, a content hub is a page that does that for one subject. It is the pillar of a cluster under another name.

The word has drifted. HubSpot renamed its website product Content Hub in 2024; its product page is first archived by the Internet Archive on 4 April 2024. On 13 November 2026, two of Google's top four US results for "content hub" were about HubSpot's product, and the first result was a piece of software. If you search the term to learn the idea, check which meaning you are reading.

### Silo: one word, two opposite meanings

In SEO architecture, a silo is deliberate. Bruce Clay's agency published "SEO Siloing: Building a Themed Website" by May 2010 (its first copy in the Internet Archive), whose current page describes a silo as "a group of themed or subject-specific content", like the chapters in a book. That is a topic cluster whose links stay inside it.

In content marketing, a silo is a failure. Moz's 2025 guide to topic clusters opens with "In content marketing, content should never exist in silos", meaning pages cut off from each other. Same word, opposite advice. When someone tells you to silo your site, ask which one they mean.

> **Where each word fits:** Keyword list and keyword map are the ingredients. Topic cluster, content cluster, content hub and the SEO silo are one group of pages. A topical map is the plan that holds all the groups. A content map answers a different question: who reads what, and when.

## Which plan do you have? Five questions to ask of any plan

Plans get labeled generously. A keyword export gets called a map, and an AI answer gets called a strategy. These five questions name what you actually have. Ask them in order. The first no is your answer.

![Five yes or no questions that tell a keyword list, keyword map, topic cluster and topical map apart](https://frameworkseo.com/learn/topical-map-vs-topic-cluster/which-plan-test.webp)

*Ask the five questions in order; the first no names the plan you have.*

| Question | If no, you have | What to add next |
| --- | --- | --- |
| 1. Does every search have a page it belongs to? | A keyword list | Merge same-need searches, then give each need a page |
| 2. Are the pages grouped under parent pages, with links to and from them? | A keyword map | Parent pages, and the links up and down |
| 3. Does it cover the whole subject, including the pages that sell? | One topic cluster, or a few | The selling pages, then the clusters that feed them |
| 4. Does every page have a type, and links across groups? | Clusters side by side | A type per page, read from Google's results; the related links |
| 5. Is there a publishing order and a written list of what is out? | A map without its last two decisions | The order, from the links; the boundary, in writing |

An AI answer can stop at question 1. Koray's warning from [February 2024](https://medium.com/@ktgubur/3-suggestions-about-topical-authority-1cd73963a9cb) still holds: "Asking ChatGPT for 20 types of X won't give you a topical map about X." You get a list with headings. How to get more than that out of an assistant is in [build a topical map with Claude](https://frameworkseo.com/learn/claude-for-seo/build-a-topical-map-with-claude), and what generators can and cannot decide is in [topical map tools and generators](https://frameworkseo.com/learn/topical-maps/topical-map-tools).

### After you publish, compare your groups with the engines' groups

Search engines group searches too, and you can now see some of their groups. Bing's [Topics report](https://blogs.bing.com/search/June-2026/New-AI-Visibility-Insights-in-Bing-Webmaster-Tools-Intents-Topics-Citation-Share-Compare), added to Bing Webmaster Tools on 16 June 2026, groups "related grounding queries into broader thematic clusters". Google has patented a way to [sort search results into themes](https://patents.google.com/patent/US12158907B1/en) by clustering summaries of passages (granted December 2024), though a patent does not show the method is in use. If the engine's groups and your map's groups disagree, look again at your borders.

To build the map yourself, follow [how to build a topical map](https://frameworkseo.com/learn/topical-maps/how-to-build-a-topical-map) and start from the [topical map template](https://frameworkseo.com/learn/topical-maps/topical-map-template). Or have it built: that is what our [growth plan](https://frameworkseo.com/) is, a full topical map of your site with every page's type, links and place in the order.

## Sources

- [Topic Clusters: The Next Evolution of SEO (archived copy)](https://web.archive.org/web/20180109215341/https://research.hubspot.com/topic-clusters-seo), HubSpot Research, via the Internet Archive, 2017
- [Topic clusters: The next evolution of SEO](https://blog.hubspot.com/marketing/topic-clusters-seo), HubSpot, 2026
- [Koray's Agents: Generative AI Agents for Semantic SEO and Topical Authority](https://medium.com/@ktgubur/korays-agents-generative-ai-agents-for-semantic-seo-and-topical-authority-d4b247fac72a), Koray Tuğberk Gübür, Medium, 2024
- [3 Suggestions about Topical Authority](https://medium.com/@ktgubur/3-suggestions-about-topical-authority-1cd73963a9cb), Koray Tuğberk Gübür, Medium, 2024
- [The Information Retrieval Thesaurus](https://www.isko.org/cyclo/thesaurus), ISKO Encyclopedia of Knowledge Organization (Stella G. Dextre Clarke), 2019
- [Ontologies as knowledge organization systems](https://www.isko.org/cyclo/ontologies), ISKO Encyclopedia of Knowledge Organization (Maria Teresa Biagetti), 2021
- [ANSI/NISO Z39.19-2005 (R2010) Guidelines for the Construction, Format, and Management of Monolingual Controlled Vocabularies](https://www.niso.org/publications/ansiniso-z3919-2005-r2010), National Information Standards Organization, 2010
- [Library of Congress Subject Headings](https://id.loc.gov/authorities/subjects.html), Library of Congress, 2026
- [Authoritative sources in a hyperlinked environment](https://www.cs.cornell.edu/home/kleinber/auth.pdf), Jon Kleinberg, Cornell University, 1999
- [Bing Webmaster Guidelines](https://www.bing.com/webmasters/help/webmaster-guidelines-30fba23a), Microsoft Bing, 2026
- [New AI Visibility Insights in Bing Webmaster Tools: Intents, Topics, Citation Share, Compare](https://blogs.bing.com/search/June-2026/New-AI-Visibility-Insights-in-Bing-Webmaster-Tools-Intents-Topics-Citation-Share-Compare), Microsoft Bing Blog, 2026
- [US12158907B1: Thematic search](https://patents.google.com/patent/US12158907B1/en), Google LLC, Google Patents, 2024
- [Generate keyword ideas](https://developers.google.com/google-ads/api/docs/keyword-planning/generate-keyword-ideas), Google Ads API documentation, 2026