Age, income and household size are the first facts a business learns about a customer. They are cheap to collect and easy to act on. They also explain less than most teams expect. This entry covers what demographic segmentation is, how generations are defined, where the evidence is thin and how a loyalty team can use it.
What is demographic segmentation?
Demographic segmentation sorts people by facts about them, such as age, income and job. The OpenStax textbook adds gender, schooling, race and ethnicity, religion and family (OpenStax). It calls this the who. Behavior is the how. Psychographics are the why.
What are the main demographic variables?
Seven families cover most uses. The table gives each with an example from the textbook or a loyalty use.
| Variable | Typical groups | Example |
|---|---|---|
| Age | Age range, life stage, generation | A student plan for people enrolled in college |
| Income | Income bands | Mercedes-Benz sells the A-Class, C-Class, E-Class and S-Class at rising price points |
| Gender | Men and women | Ad campaigns in different magazines |
| Education | Levels of schooling | A bookstore in a college town |
| Occupation | Job types | AllHeart makes scrubs for medical staff |
| Family structure | Household, children, marital status | Family plans and household pooling |
| Race, ethnicity, religion | Cultural groups | Zondervan, a Bible publisher |
The textbook adds two cautions. We tend to buy by the age we feel. The age on our ID matters less. And life-stage groups are tricky, because family patterns shift and stereotypes creep in (OpenStax).
Companies use these facts in plain ways. Mercedes-Benz prices its A, C, E and S classes at rising points for rising incomes. AllHeart makes scrubs for medical staff. Saf-Gard targets workers in construction and farming. Zondervan, a Bible publisher, segments by religion. IMAN Cosmetics is designed for women with darker skin tones (OpenStax). Each starts from a fact about the buyer.
How are the generations defined?
Marketers often split age by generation. OpenStax gives baby boomers as born from 1946 to 1964 and Gen X from 1965 to 1980. Pew Research Center sets Millennials as born from 1981 to 1996 and counts anyone born from 1997 onward as a new generation (Pew Research Center).
Pew adds a warning. Generational cutoffs are not an exact science, and they should be viewed mainly as tools. Those lines help a team compare groups. They do not make everyone in a group alike. Check any generation rule against your own member data before you build on it, and review it each year.
Does demographic segmentation work?
It works for some jobs and poorly for others. It is good at reach and eligibility. A student plan needs a rule about who qualifies. It is weaker as a predictor of what someone will buy. Russell Haley argued in 1968 that descriptive facts about buyers do not predict future buying well (Haley, 1968).
A large study backs this up. Hammond, Ehrenberg and Goodhardt checked buyers of rival brands in over 20 grocery categories. They found little brand segmentation. Buyers of rival brands looked alike on age, income and the like (Hammond et al., 1996). Our own segmentation guide says demographic data is cheap and stable but weakly predictive alone, since two customers with the same profile can behave in opposite ways.
Demographics fit three jobs. The first is eligibility. A student plan needs a rule on who qualifies. The second is reach. OpenStax notes that education can set which channels reach a group (OpenStax). The third is product fit, as when AllHeart sells scrubs to medical staff. Demographics fit prediction poorly. Use them to decide who can join and where to advertise. Use behavior to decide what to offer next.
What are the legal limits?
Some of these facts are protected. In credit, the Equal Credit Opportunity Act bars discrimination based on race, color, religion, national origin, sex, marital status, age or public assistance (FTC). A program tied to a card or loan needs a legal check before it targets or excludes on those facts. This describes the text of the law. It is not legal advice.
Income gives a simple example. With income data, a company can see how its base spends at the high and low ends (OpenStax). That tells a program where a premium tier can sell. It does not tell the program which members will buy it.
The best programs combine bases. OpenStax says a company can use all the bases or a mix, which it calls multi-segment marketing (OpenStax). Our segmentation guide puts it this way: the types stack. A working segment is usually a value band crossed with a behavior and a prediction. Age or household then sets who is eligible, and behavior sets what they see.
How do loyalty programs use demographics?
Mostly for eligibility and pricing. Spotify’s student offer costs $6.99 a month after a free month and now includes Hulu with ads. It is open only to students at eligible accredited colleges. The same page lists standard Premium at a normal $12.99 a month (Spotify). That is a demographic rule plus a check on who qualifies.
Starbucks uses age as a rule. Its terms say Starbucks Rewards is not aimed at anyone under 13 and that members from 13 to 18 take part under a parent’s supervision. Birthday data is optional, and members who skip it do not get the Birthday Reward (Starbucks Rewards Terms).
A household is a unit too. Some programs treat it as one customer. Members pool points and share status. Our financial services page says Member 360 uses household pooling to see the people around an account. Its flexible setup supports family, corporate and subscription plans.
How do you collect demographic data?
To collect these facts, ask for the least the offer needs. Our first-party data guide says zero-party data, such as preferences and birthday, is volunteered and the customer expects you to use it. Starbucks makes birthday data optional, and members who skip it get no Birthday Reward. The member gives a fact and gets something for it.
How can a loyalty team test a demographic segment?
Ask if the split changes what people do. The ADAMS list from OpenStax is a fair test: accessible, differentiable, actionable, measurable and substantial (OpenStax). Check differentiable first. If age bands act alike, the split adds nothing.
For illustration, a program sends one offer to 40,000 members. Redemption is 6.1% for ages 18 to 24, 5.8% for 25 to 34, 5.9% for 35 to 49 and 6.0% for 50 and over. Age explains almost none of the response. With about 10,000 members per band, the 0.3-point gap between the top and bottom bands is within chance. Now split the same members by behavior. Active members redeem at 9.1% and lapsing members at 3.0%. The behavioral split separates the groups, so the offer should follow behavior and use age only for eligibility.
How GRAVTY handles demographic segmentation
Demographics are member facts that rules can read. Our segmentation guide says segments sit in patented Visual Rules. The rules read member facts, behavior and model scores. So a team can cross age or household with behavior.
Frequently asked questions
What is an example of demographic segmentation?
A student discount is a clear example. Spotify offers Premium Student at $6.99 a month to eligible college students, while its page lists standard Premium at a normal $12.99 (Spotify). The group is defined by a fact about the customer. What they bought does not enter into it.
What are the main demographic variables?
The main variables are age, gender, income, education, job, family structure, and race, ethnicity or religion (OpenStax). Age is the most used. It comes as an age range, a life stage or a generation.
What years are Millennials and Gen Z?
Pew Research Center defines Millennials as born from 1981 to 1996 and Generation Z as born from 1997 onward. It calls these cutoffs tools and not an exact science (Pew Research Center). Other sources draw the lines a little differently.
Is demographic segmentation still effective?
They suit eligibility and reach. They predict purchases less well. In over 20 grocery categories, rival brands had buyers with much the same profile (Hammond et al., 1996). Most programs add behavior.
What is the difference between demographic and behavioral segmentation?
One uses who customers are. The other uses what they do. OpenStax calls them the who and the how (OpenStax). Actions are logged directly, so they predict future purchases better than a profile does.
