Target audience research: why your analytics can't tell you why people buy

Target audience research uses records and public data for who and where, and interviews for why people buy. Then write a description with its evidence.

Instant Expert EditorialPublished Updated 5 min read

Target audience research answers three questions: who your buyers are, what situation makes them look for a product like yours, and where you can find them. Records and public data cover the first and third. The second usually decides whether your marketing lands, and you mostly learn it by talking to people.

If you are still choosing between several customer groups, start with how to choose your first customer segment.

A made-up example: guest messaging for small hotels

Here is a made-up example used throughout the page. You sell a text-messaging tool that lets hotel front desks answer guest questions.

Your current audience description is "independent hotels." That names a market but not whom to interview or what an ad should say. Research turns it into something like: "general managers of independent hotels with roughly 40 to 120 rooms, who cover the front desk themselves on short-staffed nights and already answer guest texts from a personal phone." Every part of that sentence should trace back to evidence. Label any part that is still a guess.

1. Start with the people who already pay you

Your own records are the most reliable evidence you have. Pull the last 20 or so customers who paid. Note who signed, the business size, how they heard of you and what they used before. Do the same for people who tried the product and didn't buy. The differences between the lists often teach you the most.

2. Treat analytics as a partial view of visitors, not buyers

Google Analytics' demographic details report shows visitors' age bracket, gender, interests, location and language. It includes only aggregated data from users who consent to sharing demographics, and Google may withhold data when user counts are low. For a B2B product, the age and interests of site visitors rarely tell you who approves a purchase.

3. Use public data to count and locate the audience

The SBA's market research guide suggests gathering demographic information such as population data on age, wealth, family and interests, and lists free federal sources, including Census and Bureau of Labor Statistics data. For a business audience, the equivalent is counts of businesses by industry and location, which the same page links through NAICS and the Census Business Builder.

Public data tells you how many people or businesses match a description and where they are. It can't tell you which of them feel the problem. You could count hotels by state, but not which general managers are answering texts on their own phones at midnight (in bed, with the hotel's ice machine making that noise again).

4. Interview to learn the situation and the exact words

Interviews fill the gap data leaves. Nielsen Norman Group's guide to persona types recommends building audience descriptions from qualitative research, typically by interviewing 5 to 30 people in rolling groups of about 5 until each new interview adds little. It advises against descriptions built only from demographic or analytics data, because those lack the context of what people were trying to do and why.

Ask about a recent, specific event. For the hotel example:

Tell me about the last night a guest needed something and nobody was at the desk. How did they reach you, what did you do, and what happened next?

Listen for who handled it, what tools they used and the exact phrases they chose. Those phrases are often better ad copy than anything you would write yourself.

Interview three groups if you can: current customers, people who looked at your product and chose something else, and people who match your description but have never heard of you. The last group is the hardest to reach and the most likely to show that your description is too narrow or too broad. How to find customers to interview covers where to look, and customer interview questions has more question wording.

5. Write the description as a table, with evidence beside each line

Part of the descriptionHotel exampleEvidence
Who decidesGeneral manager; owner signs off above a set monthly costSales records and interviews
Situation that triggers the needGuest messages arrive when the desk is unstaffedInterviews
Current workaroundPersonal phone number on a card in the roomInterviews
Words they use"Late-night desk," "guest texts"Interview transcripts
Where to reach themState hotel associations, property-management software marketplacesInterviews; still to be tested
Who it excludesHotels with a 24-hour staffed deskInterviews with non-buyers

A line backed only by your team's opinion is an assumption. Keep it visibly separate from what people told you and what records show.

6. Count how common the description is last

Interviews tell you what the audience looks like, not what share of the market it is. Once the description is clear, a survey or a small ad test can estimate how many people match it. NN/g notes that turning survey data into statistically clustered personas needs at least 100 respondents, ideally 500 or more, plus someone who can do the statistical analysis, and that a qualitative approach is the right balance for most teams. Interviews vs. surveys explains when each fits, and market research methodologies covers the wider set of methods.

Your next step

Write your audience description in one sentence, then mark which parts come from evidence and which are guesses. Pick the biggest guess and schedule five conversations to test it.

If the people you need to hear from are professionals outside your network, describe them as specifically as your table does, such as "general managers of independent hotels who also cover the front desk." Instant Expert finds people who match, and you review them and approve the request before anything goes out. The directory pages for hospitality operations professionals and hospitality marketing professionals are one place to start.

Five calls are enough to test one guess. At a $60 offer for a 30-minute call, each costs you $75 with our fee added on top, so five cost $375, and you're charged only for the ones that get booked. A manager who covers the front desk may only be free between check-ins, so they book a time that suits them, within 7 days.