Prompts and topic clusters
A prompt is a question RankX AI asks AI assistants; a topic cluster is the subject prompts, keywords and content share. How they differ and how to set them up.
Two objects decide what RankX AI measures about you. A prompt is a question RankX AI asks AI assistants on your behalf. A topic cluster is the subject that a prompt, a keyword and a piece of content all belong to.
Get the clusters roughly right and every report becomes readable by subject. Get the prompts right and you are measuring the questions your buyers actually ask.
How is a prompt different from a keyword?
A keyword is what someone types into Google and is measured by its position. A prompt is what someone asks an AI assistant and is measured by whether the answer names your brand. They look similar and behave very differently.
| A keyword | A prompt | |
|---|---|---|
| Asked of | An AI assistant | |
| Shape | What someone types into a search box | What someone asks a person |
| Length | Short | Longer, with constraints |
| Measured by | Position, to the top 20 | Whether your brand is named in the answer |
| Charged | Per keyword, per check | Per prompt, per assistant, per run |
| Widens | Rank tracking and AI Overviews | AI visibility and AI Answer Citations |
"Running shoes" is a keyword. "Which running shoes are best for flat feet under £100?" is a prompt.
Putting a bare keyword into a prompt panel measures very little. An assistant asked a bare category term tends to return a general overview of the best-known brands, which tells you about the category rather than about you.
What makes a good prompt?
A good prompt is a question a buyer would actually ask, in the words they would use. Five tests:
- It does not name your brand. A prompt that names you is a question about you, and the assistant will name you in the answer. Whether you come up unprompted is the point.
- It has constraints. Use case, budget, size, location or compatibility. Constraints make the answer specific enough to name a real set of competitors.
- It asks one thing. A compound question produces a compound answer and an unclear verdict.
- It is one you want to win, not one you already win. A panel tuned to look good tells you nothing.
- It has a buyer stage. RankX AI tags each prompt as Researching, Comparing or Ready to buy. A panel that is all one stage tells you about one moment in a purchase.
What is a topic cluster?
A topic cluster is a subject your business wants to be known for, and in RankX AI it is the one structure that keywords, prompts and content all belong to. That makes three different measurements comparable by subject.
"We rank well on this subject, the assistants never name us on it, and we have published nothing about it" is a sentence you can only write when one structure spans all three.
Aim for roughly right rather than exactly right. Clusters can be edited and renamed without losing anything, and having imperfect clusters is far better than having none.
What does a good topic cluster look like?
- A subject, not a keyword. "Trail running shoes" is a cluster; "best trail running shoes for women" is a keyword inside it.
- A subject you sell into. A cluster you have no offer for produces prompts you cannot win and briefs nobody should write.
- Distinct from its neighbours. Two clusters that would take the same keywords and the same prompts are one cluster with two names, and having both makes every by-cluster reading less useful.
- Named the way you would say it aloud. Rename a suggested cluster if its name reads like jargon to your customers.
How do I set up prompts and clusters for a new Website?
RankX AI suggests topic clusters, prompts and keywords during setup, and you choose which to keep. A workable shape:
Keep six to ten clusters
Cover what you actually sell. Fewer than six is usually one cluster doing several jobs; more than ten is usually a taxonomy rather than a strategy.
Track two to four prompts per cluster
Spread them across buyer stages, weighted to the stage where you should be winning.
Attach keywords to clusters as you save them
The Add Keyword dialog has a topic cluster field. A keyword with no cluster is one that no brief, prompt or report by subject will ever find.
Prune after the first re-check
Every active prompt is a recurring cost, multiplied by its assistants and its cadence, and this is the cheapest moment to cut. See check cadence.
Who decides the cluster structure?
You do. RankX AI suggests clusters during setup, and you keep, rename, add or remove them on the Topic clusters screen. Every other surface, including MCP, works within the structure you set: an assistant connected over MCP can attach a keyword to an existing cluster, because that is bookkeeping, but it cannot create one.
Deciding what the clusters are is a decision about how your business describes itself. Every keyword, prompt and article filed under a cluster inherits that decision, so it stays with you.
Where to go next
- AI Prompts, for managing the prompt panel.
- Topic clusters, for the clusters screen.
- Keyword research, for finding keywords to fill them.
- AI prompt research, for finding the questions buyers ask AI assistants in your market.
Questions
Should my prompts include my brand name?
No. A prompt that names your brand is a question about you, and the assistant will name you in its answer. Leave the brand out so the check measures whether assistants recommend you unprompted.
How many prompts should I track per topic cluster?
Two to four per cluster is a workable start, spread across buyer stages. Every active prompt is checked on each of its assistants on every run, so prune after the first re-check rather than starting large.
Can an AI assistant connected over MCP create topic clusters?
No. Over MCP an assistant can list your clusters and attach a keyword to an existing one. Creating, renaming and removing clusters happens in RankX AI, where you decide the structure.
Last updated