All 29 modules are open from the start — nothing here is
locked, and nothing costs anything. Sign in so your progress, titles and
credentials stay with you, on every device you use.
Free forever, with your Google account. No password, no payment.
Slice a huge, fuzzy audience into groups you can actually build for
"Everyone" is not a target audience. Segmentation is how product teams split a crowd of
users into meaningful groups, so features, pricing, and messaging can fit real people
instead of an imaginary average.
Ready?
1
Users vs. Customers
The user is the person who uses the product. The customer is the person who pays for it and decides whether to keep paying. Sometimes they are the same person. In business software, in family products, in anything bought by an organisation, they usually are not — and they want different things.
A concrete case: a parent enters billing details to upgrade a family plan so their six-year-old can stream audio stories. The child is the user and wants dinosaurs and simplicity. The parent is the customer and wants safety controls and a fair price. Delight only the child and the subscription gets cancelled; delight only the parent and the child stops opening the app, and the subscription gets cancelled slightly later.
This matters because product decisions optimise for whoever is in the room, and the person in the room is usually the one who talks to you — which is the user, because they are the one with the product open.
The practical habit is to name both explicitly for every decision, and to ask which one each feature serves. A roadmap that serves only one of them will look successful on one set of metrics right up until the renewal conversation.
Read the last line of each panel together: the one who talks to you most cannot renew, and the one who decides whether it continues is rarely in your research. That asymmetry is the whole problem — the feedback arrives from the side of the drawing that has no say in the outcome.
Everyday example, dog food
Who eats the dog food? The dog (the user). Who chooses the brand and pays
for it? The owner (the customer). The dog cares about taste; the owner
cares about price, healthy ingredients, and not lugging a heavy bag. That's why dog-food
ads talk about your pet's shiny coat and your peace of mind, they're selling to the
owner about the dog. Miss this split and you might build a product the user loves but the
buyer won't pay for, or vice versa.
Uses it
The User
Interacts with the product daily. Cares about ease, speed, and getting their task done.
Pays for it
The Customer
Holds the budget and makes the buying decision. Cares about price, safety, and value for money.
Building for the wrong one
A classroom tool was designed entirely around what teachers asked for. Renewals were decided by school administrators, who cared about safeguarding reports the product did not produce. Teachers loved it and schools did not renew — the user was delighted and the customer was never served.
Quick check
Sarah enters her billing details to upgrade the family plan so her 6-year-old son Leo
can safely stream dinosaur audio stories on his iPad. Who is who?
2
The Four Ways to Slice an Audience
Segmentation is dividing your audience into groups that behave differently enough to deserve different treatment. The last clause is the whole test: if two groups should be served identically, splitting them is a slide, not a segmentation.
There are four common ways to slice, and they are not equally useful.
Demographic — age, gender, income, location. Firmographic is the business equivalent: company size, industry, revenue. Both are easy to collect and frequently predict nothing.
Behavioural — what people actually did: how often they log in, which features they use, whether they imported data on day one. Actions do not lie, and this is usually the most immediately actionable cut.
Psychographic — attitudes, values, how someone sees themselves. Harder to collect, and it explains motivation that behaviour alone cannot.
Marketing teams lean on demographics because ad platforms are bought that way. Product teams should lean on behaviour, because behaviour is the thing your product can actually change.
The top two rows are attributes you can buy or filter for, and they are the two that predict least. Only the highlighted row describes something your product can change — which is why a segmentation assembled from the top two tells marketing who to advertise to and tells the team nothing.
Everyday example, sorting party guests
Imagine describing the people at a big party four different ways.
Demographics: "mostly 20-to-40-year-olds." Geographics:
"half came from across town, half are neighbors." Psychographics: "the
adventurous, outgoing crowd." Behavioral: "the ones who actually got up
and danced all night." For a product team, that last lens, what people actually
do, is usually the most useful, because someone might call themselves a
party animal but stand by the wall all night. Actions reveal more than labels.
The Who
Demographics
Fixed facts about people: age, income, job title, industry.
The Where
Geographics
Location: country, city size, climate, urban vs. rural.
What people actually do: usage frequency, features touched, purchase patterns.
A rule worth remembering
Marketing teams lean on demographics to plan ad spend. Product teams live in
behavioral data, because what people say about themselves in surveys
rarely matches what the usage logs show they actually do.
Demographics that predicted nothing
A team built two personas on age and job title and could not explain why one converted at triple the rate of the other. Re-cut by whether the account had an existing dataset to import on day one, the difference was obvious and immediately actionable. The demographic split had been describing the customers, not their behaviour.
Quick check
In a fitness app you find a clean cluster of users who log in five times a week,
connect external devices, and average 45-minute sessions. Which slicing method is this?
3
Needs-Based Segmentation
The most useful segments are frequently invisible to every other cut, because they are defined by why someone is buying rather than by anything about who they are.
This is needs-based segmentation. Instead of grouping by age, industry or feature usage, you group by the job people are hiring the product to do — the progress they are trying to make.
The defining property is that these segments cut across demographics and firmographics, which is exactly what makes them useful and exactly what makes them hard to find. Two customers identical on every observable attribute can be in the segment for completely different reasons, and two customers with nothing observable in common can belong to the same one.
The cost is real: you cannot buy this from a data provider or filter it out of your database. It comes from talking to people and asking what they were trying to accomplish, which is slower than running a query and cannot be automated.
The payoff is that it predicts behaviour when nothing else does, and — more importantly — it tells you what to build.
A demographic segment tells you who to advertise to. A needs-based segment tells you what the product should actually do, and only one of those is a product decision.
Both highlighted cells are invisible to a demographic cut — one holds people who look identical on paper and are buying for different reasons, the other holds people with nothing observable in common who want the same thing. Neither can be filtered out of a database, and the segment worth having is the second one.
Everyday example, the hardware store
A smart hardware store doesn't really care whether you're young or old, rich or poor. It
groups you by the project you walked in for: "fixing a leaky tap,"
"building a deck," "painting a room." A retired grandmother and a 25-year-old flat-sharer
fixing the same leak need the exact same aisle. Grouping by shared need, not shared
traits, points you straight at what to build (or stock), which is why it's the product
manager's most powerful lens.
The milkshake principle, again
When researchers helped a restaurant sell more milkshakes, sorting customers by income
or age revealed nothing. Grouping by situation revealed a hidden segment:
people buying milkshakes at 7 AM.
Their shared need: a boring 30-minute commute, one free hand, and a stomach that
should stay full until noon. Age and income were irrelevant.
The opportunity appeared only when the grouping was based on shared needs, not shared traits.
The interview tactic: 5 Whys
In user conversations, ask "why?" repeatedly to move past surface complaints and uncover the deeper progress people are seeking.
Two segments, same demographics
A meal-kit service found its two best segments were identical on age, income and household size. What separated them was the need: one group was buying time back on weeknights, the other was learning to cook. The same box, the same price, and two completely different reasons to stay subscribed — which meant two different reasons to cancel, and only one of them was fixable with better recipes.
Quick check
A team finds a group of customers who all bought for the same reason — "I need to prove to my board that spend is under control" — despite differing wildly in company size, industry and usage. What kind of segment is this?
Drill what you learned
Scenario 1
easy
You're designing a wearable health tracker for senior citizens. The senior wears it; their adult child pays the subscription and monitors the dashboard.
How should you classify them — and what follows from it?
Scenario 2
easy
You manage a workplace tool like Slack and need to decide next quarter's feature priorities.
Which segmentation will guide those decisions best?
Scenario 3
easy
A teammate proposes splitting your audience into two segments: 'young users' and 'power users'.
What's wrong with this scheme?
Scenario 4
easy
You're a small startup with limited resources and five plausible customer segments, all somewhat attractive. Leadership wants to serve all five at once.
What's the wiser segmentation-driven strategy?
Scenario 5
easy
A survey says 80% of users "care deeply about privacy." But your usage logs show almost no one changes the default privacy settings or reads the policy.
Which signal should a product team trust for segmentation, and why?
Scenario 6
medium
A B2B PM segments purely by company size (SMB, mid-market, enterprise) and assumes everyone within each tier wants the same thing.
What's the limitation?
Scenario 7
medium
Your team defines a segment as "users who would benefit from our advanced analytics." When you try to build a campaign for them, no one can pull a list.
Which segment-quality test does this fail?
Scenario 8
medium
Analysis reveals a genuinely distinct, well-defined segment with a unique need — but it contains only 40 users, and building for them would take months.
Which test fails, and what does it mean?
Scenario 9
medium
A design team creates one 'average user' by blending all segments together: someone who is 38, uses the app 3x a week, and wants both simplicity and power features.
What's the danger of designing for this 'average'?
Scenario 10
medium
In a kids' education app, the child uses it, a parent pays, and a teacher recommends it to the whole class.
How does segmentation handle three parties?
Scenario 11
hard
Two behavioral segments emerge: 'power users' (daily, deep feature use) and 'dormant users' (signed up, barely return). A PM wants to build advanced features to please the power users.
What should segmentation prompt you to consider first?
Scenario 12
hard
A marketer proposes segments: "Millennials," "People who like coffee," and "Users in cold climates." Your PM instinct says something's off.
What's structurally wrong with this set?
Scenario 13
hard
You discover a needs-based segment — "users who need to collaborate with clients outside their company" — that cuts across every age, region, and company size.
Why is this often more valuable than a demographic segment?
Scenario 14
hard
A dating app's user (the single person swiping) and its 'customer' (also the single person, who pays for premium) are the same individual. A teammate says "so the user/customer split is useless here."
Is that right?
Scenario 15
hard
Your team has clean, MECE, measurable segments. Leadership asks: "Great — so which one do we actually target first?"
What factors should decide the target segment?
Drilled it. Now apply it to a real situation.
Put it to work
From lesson 1
The person who pays is not the person who uses it
A school homework platform
The product is bought by school IT directors on annual contracts and used
every day by teachers and by children aged 9 to 14. Renewal rates are
strong. Daily use per pupil has fallen for three terms running.
The roadmap for the year is built entirely from IT directors' requests:
SSO, audit logs, a data-retention dashboard.
What is the risk in a roadmap shaped this way?
Select all that apply — there are 3 to find.
What actually happened
They split the roadmap explicitly: a buyer track and a user track, each
with its own outcome. The buyer track kept the contract; the user track
moved daily pupil use for the first time in three terms.
Name the customer and the user separately. When one pays and another
uses, a roadmap built from one voice will quietly stop serving the other.
From lesson 2
Four cuts of the same audience
A photo-printing service
The same customer base, sliced four ways. Each slice answers a different
question, and picking the wrong one produces a confident wrong answer.
Match each question to the way of slicing that answers it, in the order the questions are listed. 1. Which customers should the sales team call first? 2. Why do some people order every month and others once a year? 3. Which ad creative should run on which channel? 4. Which feature request should we build?
Drag the rows, or use the arrows, then check.
Firmographic or demographic — company size, age, location, incomeQuestion 1. Sales needs to know who to call, and those attributes are visible before any relationship exists. It is the only cut you can act on with a stranger.
Behavioural — what people actually do in the product, and how oftenQuestion 2. Ordering frequency is behaviour, and behaviour is explained by other behaviour far more reliably than by who someone is.
Psychographic — attitudes, motivations, what they care aboutQuestion 3. Creative works on motivation. Someone printing a wedding album and someone printing 200 holiday snaps want different things said to them.
Needs-based — the job they are trying to get doneQuestion 4. Roadmap decisions need the underlying need, because that is the only cut that tells you what to build rather than who to talk to or how.
What actually happened
The team had been using demographics for all four. It had worked
acceptably for sales targeting and produced three years of roadmap
decisions based on the observation that older customers order more —
which is true, and says nothing whatsoever about what to build.
Demographic for reach, behavioural for explanation, psychographic for
message, needs-based for what to build. Using one cut for every question
is the most common segmentation mistake.
From lesson 3
The segment that was not a segment
A budgeting app
The team has segmented its users into four groups and built a roadmap
around them.
The four proposed segments
Segment
Defined by
Roadmap implication
Students
Age 18-24
Simpler onboarding
Debt-fighters
Trying to clear a specific debt by a date
Payoff projections and progress
High earners
Income over £80k
Investment tracking
Couples
Two people sharing one budget
Shared categories, separate spends
Work out which of these will actually inform what to build.
Step 1 of 3
Two of these four are needs-based and two are not. Which pair is which?
"Simpler onboarding for students" assumes young people need simpler things. That is an assumption smuggled in as a segment.
Step 2 of 3
Data shows 31% of the "high earners" segment are also debt-fighters. What does that overlap tell you?
So the two demographic segments move to marketing and pricing, where they belong, and the two needs-based ones drive the roadmap.
Step 3 of 3
What is the test for whether a segment belongs on a roadmap?
Applied to the four, only two survived as roadmap segments — and the roadmap got noticeably shorter.
What actually happened
The debt-fighter features shipped first and were adopted across every
income band and age group in the base. Had the roadmap stayed
demographic, they would have been built for students and quietly
under-served the people who actually needed them.
A segment earns a place on the roadmap by sharing a need that implies
different work. If two segments would get the same build, they are one
segment.