Audience research
How to run AI marketing tools audience research that guides campaigns
A practical guide to AI marketing tools audience research for UK teams: lawful data collection, ONS context, personas, interviews and clearer campaign briefs.
What to take away
- Draft the campaign decision, and the threshold that would change it, before you open any tool.
- Choose the tool by jobtranscript coding, ticket clustering, search query clustering or survey analysis.
- Record a lawful basis before you collect, enrich or store personal data about buyers.
- Segment by need rather than job title, and test each cluster against revenue or retention.
Set the decision, then choose the tool category
Write the decision your campaign must support, name the evidence that would change it, then choose the tool that produces that evidence.
- Write the decision as one sentence, for example finance managers in Manchester paying for a monthly reporting add-on.
- Name the threshold that would change the plan.
- List what you already holdcall notes, tickets, search terms, churn reasons.
- Choose the tool by job rather than by brand.
- Set a closing date, because this work fills whatever time you allow.
Set the research decision first
- Write the decision as one sentence
- Note what would change your mind
- Ask for ten sceptical questions
- Delete anything you cannot act on
- List evidence you already hold
Choose the tool by job
| Job | Category | Named examples |
|---|---|---|
| Coding interview transcripts | Research repository | Dovetail, Otter.ai |
| Grouping support tickets | Text clustering | MonkeyLearn, Zendesk Explore |
| Grouping search queries | Keyword clustering | Google Search Console, Semrush, AlsoAsked |
An illustrative £400 a month on tooling still leaves someone to read the output. Search query clustering only pays when you compare the clusters with rival messaging, which is the method in this guide to AI marketing tools competitor research in England.
Record the lawful basis before you collect
Audience research usually touches personal data, even when you only read public comments. UK GDPR applies across the United Kingdom, and each activity needs a recorded lawful basis before collection starts. The ICO's guidance on lawful basis explains consent, contract and legitimate interests.
Legitimate interests suits much market research, but it is not automatic. You must show the processing is necessary and balance it against the rights of the people you hold data on. A tool that guesses income band from postcodes is processing buyer data.
Size the audience with national statistics and first-party records
Published statistics and your own records answer different questions, so use both and label which is which. The ONS internet access reference tables give a neutral baseline on connection type, device ownership and online activity.
Those tables can rule out a plan that assumes every buyer researches on a smartphone. Check whether a figure covers England, Great Britain or the United Kingdom before you quote it.
Size the audience
| Source | Question it answers | How to use it |
|---|---|---|
| ONS statistics | How many people are online, and how | Sanity-check a plan |
| Your analytics | Which channels bring enquiries | Set a baseline |
| Sales call notes | Why buyers chose you | Write the brief |
| Support tickets | Where the product fails expectations | Find unmet needs |
Public vs first-party evidence
ONS tables
- What it answers
- Neutral baseline
- Can rule out
- Wrong assumptions
- Cannot tell you
- Why buyers switch
- Coverage check
- GB vs UK
- Use it to
- Size and segment
Your own records
- What it answers
- Your buyers
- Can rule out
- Weak channels
- Cannot tell you
- Why they churn
- Coverage check
- Sample and period
- Use it to
- Convert and value
Example: one persona line traced to its source
Label every persona line, because mixing evidence, inference and assumption in one paragraph is how research loses credibility.
One persona line
- Need
- monthly reporting a finance manager can forward without editing. Evidence: call note, 12 March.
- Trigger
- board pack deadline in the first week of the month. Inference: three of five interviews.
- Budget authority
- up to £2,000 a year. Assumption: two buyers, untested.
That structure is the one set out in the guide to AI marketing tools buyer personas in England, which keeps the summary short and the sources in an appendix.
Google's responsible AI glossary is a useful reference for fairness and bias when you check a model's output. Give one named person the job of checking personas and briefs, and keep that record in one dated file.
Five buyer calls, eight questions
Five conversations beat fifty generated profiles.
Five buyer calls
- Walk me through the last time you bought something like this.
- What triggered the search?
- Which options did you compare, and how did you find them?
- What nearly stopped you buying?
- Who else had to agree, and what did they ask?
- What did you expect it to cost before you saw a price?
- What had you tried before, and why did it stop working?
- What would make you switch away from your current supplier?
The same eight questions go to five buyers inside a fortnight. This guide to AI marketing tools customer interviews in England shows how to keep those calls easy to code.
Common questions
Which AI tools do I actually need for audience research?
Use a research repository such as Dovetail or Otter.ai for interviews, text clustering for tickets, and keyword clustering for search queries.
Can AI replace customer interviews?
No. A model can group and summarise what buyers said, but it cannot ask the follow-up question that explains a purchase.
How often should I refresh the research?
Quarterly suits most campaigns and twice a year suits personas. Delete anything the decision no longer needs.
Do I need a lawful basis for reading public comments?
Yes, if you store or combine anything that identifies a person. Record the basis and a deletion date first.
In this guide
- Audience research for AI marketing tools: fix the briefUse audience research to scope an AI marketing tools target audience: set a decision, test segment rules, score candidates and check UK data duties.
- How to build AI marketing tools buyer personas from first-party dataHow to build AI marketing tools buyer personas from evidence: the fields that matter, a before and after table, and the compliance checks to run first.
- Plan audience research interviews for AI marketing toolsPlan audience research interviews for AI marketing tools: define UK segments, recruit decision-makers, probe tool use and protect records.
- AI marketing tools for UK survey researchExplore AI marketing tools for UK survey research, from question drafting and open-text coding to panel options and practical data-handling considerations.
- Competitor research with free AI marketing toolsA step-by-step checklist for using free AI marketing tools and public records to run competitor research in the UK without a paid data subscription.
