One feature set, six generations.
I benchmarked 168 competitor features and read every one against how six age cohorts actually think, then asked users whether the mapping was right.
Same features for everyone.
Digital health platforms ship the same features to everyone and hope different generations find a reason to stay.
Feature lists are easy to copy from competitors. What is hard is knowing which features fit which people. Age is a blunt instrument, so the question became sharper: does a feature match the mental model of a 22-year-old, a 38-year-old and a 48-year-old, and by how much?
I needed a method that could compare an identical set of features across cohorts, show where the fit was strong or thin, and be checked by real users rather than by my own judgement.
Qualitative attributes, mapped.
For each cohort I derived qualitative influential attributes (QIAs) from demographics. I then mapped the 168 benchmarked features to those QIAs and classified each match as part of a mental model.
- Benchmark168 features collected from competitors, identical across every cohort.
- Derive32 to 36 unique QIAs per cohort, built from demographic research.
- MapEach feature read against each cohort's QIAs: 973 of 1,008 cohort × feature pairs mapped.
- ClassifyEvery mapped feature sorted into adoption (why people start) or engagement (why they return).
- ValidateA user survey tested whether each cohort agreed with the mapping.
The four classes of mental model every mapped feature was filed under.
Almost everything mapped.
96.5% of all cohort × feature pairs found a home in a mental model. Coverage was high and strikingly even across ages.
| Age range | Mapped | Unique QIAs | Adoption | Engagement | QIA % | Mental model % | Agreed % |
|---|---|---|---|---|---|---|---|
| 18–24 | 161 / 168 | 32 | 89 | 72 | 80.0% | 95.8% | 90.1% |
| 25–30 | 165 / 168 | 35 | 93 | 72 | 87.5% | 98.2% | 100.0% |
| 31–35 | 165 / 168 | 34 | 92 | 73 | 85.0% | 98.2% | 97.6% |
| 36–40 | 163 / 168 | 32 | 93 | 70 | 80.0% | 97.0% | 95.1% |
| 41–45 | 159 / 168 | 36 | 93 | 66 | 90.0% | 94.6% | 98.1% |
| 46–50 | 160 / 168 | 33 | 90 | 70 | 82.5% | 95.2% | 88.8% |
A partition, not two readings.
In every cohort the adoption and engagement counts sum exactly to the mapped total (89 + 72 = 161 for 18–24, and so on). Adoption carries roughly 55–58% of the set everywhere.
Different ingredients, same coverage.
Mental-model coverage looks alike across ages, but its makeup does not. Habits dominate the 36–40 cohort at 40.5%; aptitudes lead at 41–45 with 31.5%; the youngest cohort leans hardest on attitudes at 26.8%.
Cohorts differ in endorsement, not reach.
Mental-model coverage sits inside a 3.6-point band (94.6–98.2%). User agreement spans 11.2 points (88.8–100%). The feature set reaches every cohort about equally; what changes is how much of that mapping people recognise as theirs.
Mean user agreement across cohorts: 95%. Agreed at 50% and above.
The rationale, for each age.
Each cohort was handed three business cases and a survey. Here is how the mapping held up.
18–24
90.1% agreed80% of demographic-derived QIAs mapped accurately to 96% of the mental models in the benchmarked features.
55% adoption to 89, 45% engagement to 72 · 32 QIAs25–30
100% agreed88% of QIAs mapped accurately to 98% of the mental models represented.
56% adoption to 93, 44% engagement to 72 · 35 QIAs31–35
97.6% agreed85% of QIAs mapped accurately to 98% of the mental models represented.
56% adoption to 92, 44% engagement to 73 · 34 QIAs36–40
95.1% agreed80% of QIAs mapped accurately to 97% of the mental models represented.
57% adoption to 93, 43% engagement to 70 · 32 QIAs41–45
98.1% agreed90% of QIAs mapped accurately to 95% of the mental models represented.
58% adoption to 93, 42% engagement to 66 · 36 QIAs46–50
88.8% agreed83% of QIAs mapped accurately to 95% of the mental models represented.
56% adoption to 90, 44% engagement to 70 · 33 QIAsRead the full study.
The complete matrix, all six cohorts and the validation survey conclusions, as delivered.
Open the PDF (opens in a new tab)