Find the people the product is actually for
Who exactly are you trying to reach, and what do they already believe and need?
This chapter shows how to research an audience through the job they need done, interview real customers without leading them, and segment by need instead of demographic stereotype. It also sets the data rules: consent, minimisation, and a flat ban on choosing people because they are vulnerable.
All stages
Start from the job, not the demographic
People do not buy products because they are 35 to 44 and live in a city. They buy because something in their life is not working and the product looks like a way to fix it. The jobs-to-be-done approach asks what the customer is trying to get done, in what circumstance, and what they use today instead. Write the job as a sentence in the customer's voice: "When [situation], I want to [progress], so that [outcome]."
Start with the data you already hold: order histories, cancellation reasons, support tickets, reviews. Look for the circumstance behind each purchase and each departure. This gives you hypotheses about the job. It does not yet give you the answer.
Run the interviews
Eight to twelve interviews are usually enough to hear the main patterns repeat. Recruit a mix: recent buyers, people who left, and people who solved the problem some other way. Tell each person who you are, what the conversation is for, and how the notes will be used. Ask before recording. Pay for their time whatever they say.
Ask about what happened, not what they think they would do. "Tell me about the last time you ran out" produces evidence; "would you use a flexible plan?" produces politeness. Walk through the timeline of the last purchase: what prompted it, what else they considered, what nearly stopped them, what happened afterwards. Write down their exact words. You will need them in stage 3.
Segment by need and choose one
Group what you heard by job and circumstance, not by age, sex, or income. Two people of the same age can be hiring the product for opposite reasons, and two people with nothing demographic in common can share the same need. For each need-based segment record four things: the job, the current workaround, the anxiety that holds them back, and the proof they would want before switching.
Size each segment with data you hold, then choose one primary segment for this campaign. The right choice is the group whose need the product meets best, that you can reach through channels where they expect to hear from you, and that is large enough to close the gap in your objective brief. Name the segments you are not serving this time, so nobody widens the audience later without a reason.
Collect less, and never segment on vulnerability
List every data field you plan to use for targeting. For each, write where it came from, whether the person knew it was collected, and why this campaign needs it. Drop any field without a purpose. Do not buy lists, and do not infer sensitive characteristics (health, finances, pregnancy, bereavement, emotional state) from behavior, even when a platform offers the option.
One rule has no exceptions: never define a segment by vulnerability. People who are grieving, ill, in debt, lonely, exhausted, very young, or cognitively impaired are easier to persuade because their judgment is under strain, and that is exactly why choosing them is manipulation. Check for proxies too, such as late-night browsing or a recent hardship cancellation. Finish with an exclusion list: people who asked not to be contacted, people under 18, and anyone whose records suggest hardship.
Techniques at this stage, and where the line is
Target on a need the person has shown or stated and on data they would expect you to use, such as past orders with you. Keep segments broad enough that the same honest message fits everyone in them.
The line: Targeting on inferred sensitive traits or moments of weakness, or showing different groups claims that could not all be true at once.
Use behavioral data to work out what people need and when a message is useful to them, and keep a written inventory of every field and its purpose.
The line: Building psychological profiles to find who is easiest to push, or using data the person never knew you held.
Expect interviewees to give the answer that sounds good. Ask about specific past behavior and treat stated intentions as weak evidence.
The line: Quoting flattering answers you prompted as if they were proof of demand or satisfaction.
Assume your team's tastes are not the customer's. Let interview transcripts, not internal opinion, decide what the audience cares about.
The line: Telling the market that "customers want" something only your own team wanted.
Guard against your own version first: interview people who left and look for evidence that the product does not fit the segment you hoped for.
The line: Choosing an audience because an existing belief makes them easy to tell what they want to hear, whether or not it is true.
People predict their own tired or stressed behavior badly. Learn when they really decide, then design the offer so the calm version of the customer would still approve of the choice.
The line: Scheduling contact for the moment judgment is weakest, such as late at night, straight after bad news, or on payday for someone in debt.
It is fair to speak to a shared practice or interest, such as people who brew at home, in the language that group uses.
The line: Defining the audience against a despised out-group, or implying that buying is what separates real members from outsiders.
Worked example
Kestrel Coffee has its signed brief (600 Flex starts by 31 March) and needs to know who Flex is for. It holds order histories, a year of cancellation survey answers, and an email list of 9,000 people who opted in to marketing, including lapsed subscribers.
- 1The cancellation survey shows that 41 percent of people who left the fixed plan chose "coffee was piling up" as the reason, ahead of price at 22 percent and taste at 14 percent.
- 2The team runs twelve thirty-minute interviews: six current subscribers, four lapsed, and two people who buy beans elsewhere. Each is told what the conversation is for, asked for permission to record, and sent a 25 dollar gift card whatever they say. Questions cover the last time they ran out or threw coffee away.
- 3Three need-based segments emerge: pile-up avoiders, whose consumption varies with travel and household routine; never-run-out routine drinkers; and gift buyers. Pile-up avoiders become the primary segment, with about 1,900 lapsed subscribers on the list matching by stated cancellation reason.
- 4The profile is written in customers' words. One lapsed subscriber said: "I felt guilty every time a new bag arrived and the last one was still half full." The proof this group wants is that skipping is easy and needs no phone call.
- 5Data decision: the segment uses order cadence and stated cancellation reason only. A proposal to buy a list of new parents, on the reasoning that exhausted people pay for convenience, is rejected and logged as a vulnerability-based segment. The date-of-birth field is removed from sign-up because nothing uses it.
Outcome. A one-page audience profile: pile-up avoiders as the primary segment, about 1,900 reachable on the owned list, three verbatim quotes, a data inventory of five fields, and an exclusion list that includes 41 people who cancelled citing financial hardship.
Audience profile
Complete after your interviews. Use the customer's words wherever you can. Saved in this browser only.
One sentence in the customer's voice: when [situation], I want to [progress], so that [outcome].
How many people, the mix of current, lapsed, and non-customers, how you asked consent, and what you paid.
For each segment: the job, the current workaround, the anxiety holding them back, and the proof they want.
The one segment this campaign serves, how many people you can reach, and which segments you are leaving out.
Three to five exact phrases from interviews that describe the problem as customers experience it.
Every field used for targeting, where it came from, whether the person knew, and why you need it.
Choose the option that honestly describes how you came to hold the targeting data.
Tick each group you have removed from every channel.
Ethics gate: audience
0/7If any of these is not true: Stop before writing any message. If the research is thin, do more interviews. If a field lacks consent or purpose, remove it and rebuild the segment without it. If a segment turns out to rest on vulnerability, delete it, record why, and choose the next best need-based segment even if it is smaller.
Common mistakes
- ×Writing personas from imagination, with a name and a stock photo, and never checking them against a real conversation.
- ×Asking "would you buy this?" and treating a polite yes as demand.
- ×Segmenting by age and sex because the advertising platform offers those boxes, not because the need differs.
- ×Keeping data "in case it is useful", which creates risk for customers and nothing for the campaign.
- ×Treating proxies as harmless: a segment built on late-night browsing or recent hardship is a vulnerability segment under another name.
Why am I seeing this, and what does the sender think they know about me?
Related
References
- Christensen, C. M., Hall, T., Dillon, K., & Duncan, D. S. (2016). Know your customers' "jobs to be done". Harvard Business Review, 94(9)Researching the job and circumstance behind a purchase instead of customer attributes.
- Yankelovich, D., & Meer, D. (2006). Rediscovering market segmentation. Harvard Business ReviewSegments built on needs and behavior predict purchase better than demographic or personality types.
- Portigal, S. (2013). Interviewing Users: How to Uncover Compelling Insights. Rosenfeld MediaInterview practice: asking about specific past behavior, consent, and recording.
- Kotler, P., & Keller, K. L. (2016). Marketing Management (15th ed.). PearsonSegmentation, targeting, and the criteria for choosing a primary segment.
- European Parliament and Council of the European Union (2016). Regulation (EU) 2016/679 (General Data Protection Regulation), Article 5(1)(c)The data minimisation principle: personal data limited to what is necessary for the purpose.
- Calo, R. (2014). Digital market manipulation. George Washington Law Review, 82Why targeting consumers at moments of vulnerability is a distinct harm.