How to research Gen Z brand loyalty without skipping the basics

What to take away

Loyalty research fails in a predictable way. A team fields a survey, gets a warm score, and calls it loyalty. What it measured was familiarity. Gen Z brand loyalty is a behavior: the same person choosing you again when a cheaper or louder option sits on the shelf.

Start with the cohort. Pew Research Center defines Gen Z as births from 1997 to 2012, which puts a 2026 wave between ages 14 and 29. That span crosses a consent line, so minor assent and parental permission rules apply to the younger half of the sample.

Which loyalty metric counts, and who set it

Write the definition down first. Repeat purchase rate over a fixed window, share of category spend, switching rate to a named competitor, and post-purchase satisfaction are four different numbers. Each needs its own instrument. This guide does not supply a Gen Z loyalty benchmark. Use a published comparison only when its category, market, metric, period and sample align; otherwise establish a dated baseline of your own.

Name the metric and its origin. Net Promoter Score comes from Fred Reichheld's 2003 Harvard Business Review article, published with Bain & Company. CSAT and the Customer Effort Score are satisfaction scales. None of the three records a purchase.

Awareness surveys, ad recall and follower counts cannot answer a loyalty question. They measure exposure. If a stakeholder asks for brand love, ask which of the four numbers they would accept as proof.

The NIST AI Risk Management Framework (AI RMF 1.0, 2023) Playbook is a voluntary set of actions organized around govern, map, measure and manage.

It is useful when a model scores, segments or predicts your customers, because it forces you to record what the model was for and who owns it. It is not a certification and it does not rank vendors.

Seven research methods, and what each one can actually tell you

Run at least two. A survey wave and a receipt panel together cost more than one, and they disagree in ways that teach you something. Where they agree, you have a finding.

For the drivers behind a switching decision, the factors that move Gen Z buyers are worth reading before you write the questionnaire, because they shape which questions are worth asking.

Deloitte's annual Gen Z and Millennial Survey and GWI's Gen Z reports cover spending pressure and values, and they track platform habits. Read them as a source of hypotheses, then test those ideas in your own sample.

Interview schedules usually stop when new reasons stop appearing. Roughly 12 to 20 interviews per segment is a planning convention, not a guarantee; record the stopping rationale and date.

Build the evidence file

One file per study, versioned, with the date on every entry. It holds the population and how it was drawn, the questionnaire as fielded, raw counts before weighting, the weighting scheme, and the decision the study changed.

Record what you rejected. A hypothesis that failed is a result. Teams that keep only the confirming numbers repeat the same study next year and learn nothing.

Qualtrics, SurveyMonkey and Prolific each export raw response files. Keep the export and the codebook instead of the dashboard view, because a dashboard hides the weighting.

The NIST Privacy Framework version 1.0 (2020) getting started guide describes a voluntary process for identifying privacy risk, assigning owners and recording responses. Treat it as a management aid. It is not legal clearance for a marketing use, and privacy rules for under-18 audiences vary by state and country. For a voluntary research standard, consult the ICC/ESOMAR International Code on Market, Opinion and Social Research and Data Analytics (2016); it is a code, not a substitute for law or counsel.

Keep the records you will need later

Panel vendors, survey platforms and loyalty apps all produce event logs. Keep access, change and correction records for the systems that hold respondent data.

CISA's 2024 Best Practices for Event Logging and Threat Detection guidance explains why organizations retain and review those records. It is written for security work, not for proving a marketing result. A log shows who touched a record and when; it does not show that your campaign worked.

Five checks before you publish a number

If you publish to a UK public-service audience, the GOV.UK open standards guidance links open standards to interoperability, reuse and reduced supplier dependence. Read it as a portability prompt for your data formats, not as a rule for a commercial loyalty study.

Common questions

What is the first decision in a Gen Z loyalty study?

Define which of the four numbers counts as proof: repeat purchase rate, share of category spend, switching rate, or satisfaction. Everything else, including the questionnaire and the sample, follows from that choice.

How do I know if a loyalty score is real?

Check whether it came from observed behavior or a stated answer, and whether the comparison basis matches the previous wave. A score that moves without a matching change in repeat purchase is noise or a wording change.

What should a team avoid?

Calling awareness or engagement loyalty. If that figure has already gone out, correct it in the same channel it appeared in and note the correction in the evidence file.

Which authority sets the privacy rules?

There is no single authority for every study. In the United States, the FTC enforces COPPA; California's CCPA, as amended by the CPRA, is enforced by the California Privacy Protection Agency and the California Attorney General. If a study is subject to the Common Rule, an institutional review board applies 45 CFR part 46, subpart D to child assent and parental permission, including any permitted exception or waiver. There is no universal parental-consent form for every commercial survey of respondents ages 14 to 17. Check applicable state and platform rules with qualified counsel before collecting data.

How should a study define the cohort?

Use a published boundary. Pew Research Center's 2019 generational guidance uses births from 1997 to 2012 for Gen Z, after Millennials at 1981 to 1996. State the boundary in the report so the next wave compares like with like.

How many respondents does a wave need?

Enough for the smallest subgroup you will report. Quotas by age, gender and region are typical, and most commercial trackers field 500 to 1,000 respondents per market. That range is a convention, not a statistical rule, and does not establish adequate precision or power. Set the sample size using the metric, desired precision or power, and smallest planned subgroup; document those assumptions before fieldwork.