Rules
Part of How to review Gen Z brand loyalty claims
How to vet Gen Z brand loyalty statistics
Use this checklist to vet Gen Z brand loyalty statistics: check the sample age band, repeat-purchase window, collection date, definition, and source table.
What to take away
- A loyalty percentage is only as good as its definition of a repeat purchase. Ask what counted before you quote the number.
- "Loyalty" in a vendor deck usually means retention, satisfaction or awareness. Those are three different measurements.
- Check the sampleage band, country, panel source and collection date change the answer more than the brand does.
- Trace any figure you publish back to its source table, then forward to every slide, ad and post that repeats it.
- Pair every loyalty proxy with an outcome you can verify, such as repeat purchase rate or refund rate.
Gen z brand loyalty gets quoted constantly in pitch decks, and the numbers rarely survive a close read. Two surveys can both report that most Gen Z shoppers stay with brands they trust, then disagree by thirty points once you see who they asked.
Three checks decide whether a Gen Z loyalty figure is usable: the age band of the sample, the repeat-purchase window and the collection date. Settle those three before you read the percentage. Get them wrong and the percentage tells you nothing.
Five measurement types you will actually be handed
Each answers a different question. The table describes measurement types, not verified Gen Z benchmarks; check the original report and source table for any figure you plan to quote.
Five loyalty figures compared
Figure
- Net Promoter Score
- Recommend
- Repeat purchase rate
- Bought again
- Retention rate
- Still active
- Share of wallet
- Spend share
- Aided brand awareness
- Recognition
Measures
- Net Promoter Score
- Repurchase
- Repeat purchase rate
- Why
- Retention rate
- Profitability
- Share of wallet
- Category growth
- Aided brand awareness
- Preference
Cannot tell
- Net Promoter Score
- Repeat purchase rate
- Retention rate
- Share of wallet
- Aided brand awareness
| Figure | What it measures | Where it comes from | What it cannot tell you |
|---|---|---|---|
| Net Promoter Score | Willingness to recommend | Survey question, 0 to 10 scale | Whether anyone repurchased |
| Repeat purchase rate | Share of buyers who bought again in a window | Transaction data | Why they returned |
| Retention rate | Share of a cohort still active after N months | Panel or CRM data | Whether they are profitable |
| Share of wallet | Your spend as a share of their category spend | Panel diary or card data | Whether the category is growing |
| Aided brand awareness | Recognition when prompted | Survey with brand list | Preference or purchase intent |
If you are using Nielsen, Kantar, GWI or YouGov, check the specific report's age band rather than relying on the firm name. A report may define an age group as 18 to 24, 16 to 24 or 25 to 34; those samples are not interchangeable. Align the bands before comparing results.
Use the report's methodology notes and source table to verify the sample size, age band and collection date. Cite the firm, report title, age band and date, and keep a link to the source table. If no source table is available, do not present the number as a verifiable statistic.
Typical ranges show why the window matters more than the brand. Repeat purchase rate for one brand inside a US Gen Z cohort aged 18 to 24 commonly lands between 15 and 25 percent on a 90 day window. The 25 to 34 band usually runs 10 to 15 points higher on the same window.
A longer window can give consumers more opportunity to buy again, so a higher rate is not evidence of stronger loyalty by itself. When a source gives no window, ask how it defines a repeat purchase before interpreting the rate.
Three checks before you quote a number
Who was asked. Panel surveys can skew toward people who join panels. In a multi-country sample, the US subgroup may be much smaller than the total, making subgroup estimates less certain. Check the country-level sample size before quoting a US finding.
Three checks before quoting
- Who was askedpanel skew, sample size
- What countedrepeat purchase window
- When collecteddate every statistic
- Retire anything older than three years
What counted. A repeat purchase within 90 days is a different claim from a repeat purchase within two years. Ask for the window. If the source does not state it, treat the figure as unusable.
When it was collected. A 2021 figure reflects pandemic purchasing. A 2023 figure reflects a different price environment. Date every statistic you put in a deck, and retire anything older than three years unless you are deliberately showing a trend.
For a worked example, the brand loyalty factors breakdown shows how the same cohort splits once you separate price sensitivity from habit.
How to trace a questionable figure to its source
Start with the figure and citation on the slide, then open the source table. Compare the measure definition, sample age band, repeat-purchase window and collection date with what the slide claims. Note where they differ, then correct every slide, ad or post that repeats the figure.
Trace a bad figure
Review field
- Definition drift
- Restore versioned definition
- Claim gap
- Correct affected assets
- Handoff break
- Assign owner, recheck
- False success
- Pair proxy with outcome
Acceptance test
- Definition drift
- Claim gap
- Handoff break
- False success
| Review field | What to record | Acceptance test |
|---|---|---|
| Definition drift | The unit or population changed between source and slide | Restore the versioned definition |
| Claim gap | The slide promises more than the survey measured | Correct every affected asset |
| Handoff break | No one owns the source table | Assign an owner and recheck |
| False success | Retention looks fine but refunds are climbing | Pair the proxy with outcome quality |
When a panel or analytics vendor supplies a loyalty statistic, ask what data you can access, who owns it and how you can leave the service. The GOV.UK technology selection guidance is written for public-sector buyers, but these procurement questions can also inform a vendor review.
Test the finding before you scale it
Before using a loyalty statistic to justify a program, test whether its source meets your evidence requirements and whether the finding holds under the conditions you expect to use it.
- Confirm the source, date and sample of every figure in the deck
- Align age bands across sources before comparing them
- State the repeat-purchase window in the same sentence as the rate
- Check whether the panel skews toward heavy category buyers
- Pair the loyalty proxy with refund rate or repeat purchase rate
- Set a review date and a stop rule
For broader evaluation planning, the GAO evaluation design guide connects evaluation questions with evidence needs and design choices.
If AI tools summarize survey data or score respondents, the NIST AI RMF Playbook lists voluntary actions under govern, map, measure and manage.
When the number is wrong
A loyalty statistic can fail because its definition, population, time window or collection date changed. The table above names each failure and the test that catches it. Keep the source date, sample and correction history beside the number.
For a sense of how loyalty plays out in practice, the brand loyalty examples collection shows where programmes held and where they did not.
Common questions
What is the single most important thing to check in a loyalty statistic?
The definition of a repeat purchase, including the time window. Without it, the percentage is not a measurement, it is a mood.
Can I compare figures from two different research firms?
Only if the age bands, countries, collection dates and question wording match. In practice they rarely do, so quote each figure with its own source and date rather than averaging them.
What should I do if I already published a figure that turns out to be wrong?
Trace it back to the source table, correct every asset that repeated it, and note the correction date. Do not quietly delete the slide, because the same figure is probably in someone else's deck.
