Do Gen Z shopping habits hold up once you test them properly?

What to take away

Which habits hold up, and which fall apart

Most Gen Z shopping claims come from one platform and one product category. Someone surveys 400 students about sneakers, finds TikTok drove the decision, and the deck says Gen Z discovers everything on TikTok.

That claim fails at the second category. Groceries, cosmetics and electronics run different discovery paths, and the same person behaves differently in each. The habit was never general. The sample was.

Public data show where the split lands. Pew Research Center's 2023 teen survey found YouTube used by 93% of US teens and TikTok by 63%, so the platform story is never one platform.

Piper Sandler's Taking Stock With Teens survey has put food near 23% of teen spending and e.l.f. Beauty at the top of cosmetics.

Returns break the story further. The National Retail Federation and Appriss Retail counted US returns in 2024 at about 17% of sales, roughly $890 billion, with apparel well above that figure and groceries near zero.

The test is simple. Repeat the study in a second category and a second channel. If the finding moves, it was a category fact wearing a generation label.

Pew Research Center's May 2023 note on generational reporting adds the statistical warning: one cross-section mixes age, period and cohort effects.

Four stages, four different questions

Discovery asks where the product first entered the buyer's view. Evaluation asks what the buyer compared. Purchase asks what closed the sale, including price, shipping and returns policy. Returns and repeat use test whether the decision held.

Each stage needs its own measure. Completed journeys, return rate and repeat purchase rate are the four numbers worth carrying through the study.

A team that tracks only conversion is reading the third row and guessing at the other three.

Category moves every number in that table. Apparel and footwear return rates run well above the all-retail average in NRF and Appriss Retail data, while grocery returns stay near zero.

Pick the design before you collect

Design choice starts with the objective and the practical constraints, not with the tool. The NIST experimental design selection guidance puts that order first, and it draws a line between descriptive reporting and a controlled effect estimate.

Descriptive work counts what happened. A controlled comparison estimates what caused it. Most youth marketing studies claim the second while doing the first.

Piper Sandler's Taking Stock With Teens is descriptive: a self-reported survey of several thousand US teens, run twice a year. A holdout group in a live campaign estimates an effect.

The GAO evaluation design guide makes the same connection between the evaluation question and the evidence that can answer it. Borrow the discipline, not the federal-program context.

Practical study design

Sample size follows the decision, not the budget. For a survey, about 1,000 completed responses per group gives a margin near plus or minus three points at 95% confidence. A channel test with 300 to 500 completed journeys per group is a typical planning range.

State the smallest effect you would act on before you collect. If a five point shift would change the campaign, a study that can only resolve a fifteen point shift is the wrong study.

Recruit where the purchase happens. Shelf intercepts, receipt invites and panel quotas by age, geography and store access beat a single campus. Keep the entry route for every respondent, because a convenience sample cannot carry a national claim.

Instrument returns at the transaction. Attach a study ID to the order, match the receipt to the loyalty record, and count returns at 30, 60 and 90 days. Measure repeat purchase inside a fixed window.

Self-reported return intent misses how many items come back. Treat the matched transaction as the measure and the survey answer as the reason.

Keep the scope visible

Write the approved scope, exclusions, version, owner and decision date into the working file. Age cohort, geography, life stage, platform access and purchasing power belong there, because each one changes what the result can support.

A scope that reads "US teens 13 to 17 on TikTok" cannot support a claim about Gen Z adults buying groceries. Those are different populations, and smartphone ownership sits near 96% among US adults aged 18 to 29 in Pew Research Center's mobile data.

The GOV.UK technology selection guidance covers adaptable choices, data control, security review and ownership cost. Those questions shape a pilot without endorsing a vendor.

Substantiate the claim before it runs

An objective advertising claim needs a reasonable basis before it goes out. The FTC advertising substantiation policy sets that standard.

Brand rankings and cost-of-living findings move between survey waves. A 2023 figure cannot carry a 2026 campaign without a fresh read.

A worked grocery re-test

The template is the easy part. Running it means filling in the same four rows for a second category and comparing the result with the first.

Discovery is the row that moved. In the sneaker pilot the first touch came almost entirely from social video. In the grocery re-test the shelf and the display carried nearly the same share.

The numbers above are worked example inputs used to show the arithmetic. They are not from a published survey.

The factors behind those habits belong in the same file, because a shift in one of them moves the baseline.

Common questions

Do the habits hold up?

Partly. Video discovery and price sensitivity repeat across surveys. A single platform, a single category and a generation-wide brand loyalty do not survive a second wave. Set the threshold at 300 completed journeys per group before the claim travels.

What is the first decision?

Name the owner, the audience, the outcome and the evidence that would stop the study. Everything else follows from those four. The NIST experimental design handbook starts from the objective and the constraints, not the tool.

How should the result be reviewed?

Review completed journeys, returns and repeat use against cost, quality, exclusions and source limits. Date the decision. A 90-day repeat purchase window is a typical starting point for groceries.

What should a team avoid?

Avoid generalizing one channel or category to all purchases. Preserve the affected record and correct the output where the error appeared. A campus sample of 400 students is a common route into that error.