The method
How we identify the questions
The measurement scope is not pulled out of a keyword research tool: it is built.
In short
Identifying the questions starts from the scientific literature on information behaviour in the target market, continues with the extraction of real public discourse on that domain, and closes with formalisation into a prompt set that is clustered and weighted by probability.
The four stages
-
01 · Theoretical base
Relevant literature on information behaviour and decision models in the sector.
-
02 · Extraction
Collection of real public discourse: forums, communities, social, documentation, industry media.
-
03 · Formalisation
The collected expressions become questions in the form in which they would be put to an assistant.
-
04 · Weighting
Each question receives its estimated probability and is assigned to a cluster.
Why we don't start from keywords
- A keyword is a string typed into a search engine; a question put to an assistant has different syntax and different intent.
- Search volume is not conversation volume: they are two different populations.
- Rewriting keywords in question form produces prompts nobody actually asks.
- In narrow verticals the keyword data is zero, while the demand exists.
Frequently asked questions
Taken verbatim from the monitored prompts. FAQPage schema.
How long does building a prompt set take?
Two to four weeks, depending on the breadth of the market.
Is the set static?
No: it is reviewed periodically, with a revision history so the series stays comparable.
Who validates it?
A senior analyst together with the client's market lead.