You asked whether we bucket people into the four problems correctly, whether the diagnosis follows from the questions, and whether the reporting screens match what she told us. I ran two opposite people through the whole quiz to find out.
Nothing is bucketed. There is no scoring. Every diagnosis screen is a fixed mockup showing one invented persona, identical no matter what anyone answers.
I walked the quiz twice — once answering as badly as the options allow, once as well as they allow — and compared every screen that reports back. They are byte-for-byte identical.
| Reporting screen | Weakest possible answers | Strongest possible answers | Difference |
|---|---|---|---|
| 2.11 Your starting pointthe diagnostic reveal | Mobility 28% · Balance 50% · Strength 62% · Energy 58% | Mobility 28% · Balance 50% · Strength 62% · Energy 58% | NONE |
| 5.2 Your planpriorities + safety line | Mobility first, then strength | Mobility first, then strength | NONE |
| 5.3 What changes firstweek 2/4/8/12 milestones | identical 468 characters | identical 468 characters | NONE |
| 5.5 Paywallbefore/after on four measures | identical 4,816 characters | identical 4,816 characters | NONE |
So the honest status is: the quiz collects the right things and routes correctly — that part is verified — but it does not yet compute anything. The persona on screen belongs to a woman whose mobility is poor, balance middling, strength and energy fine. Everyone currently gets her.
This matters more than a missing feature, because six screens make first-person assertions about her. Until scoring exists, each one is a guess that is right for some people and wrong for the rest.
Answer “None of these” on 2.10 and it still says it is working around your back. Answer “Never” on 2.8 — the answer that correctly skips the “we won't ask you to jump” interstitial — and the plan still says it is removing jumping. The routing already respects those answers; the plan text does not. A woman who told us she has no back problem is being told her plan is built around one.
| Screen | What it asserts | Wrong when she… |
|---|---|---|
| 1.6 | “You said you want to be stronger for everyday things” | picked any of the other four goals on 1.4 — 4 of 5 people |
| 2.11 | “You said you want to be stronger… your answers say strength is holding up” | picked another goal, or strength is in fact her weakest axis |
| 2.11 | Verdict: “Mobility first, then strength” | is anyone whose weakest axis is not mobility |
| 3.6 | “what your answers pointed to” + a profile-matched statistic | has a different profile — the stat is fixed |
| 5.2 | Three priorities: stiffness, stairs, single-leg balance | reported none of those things |
| 5.5 | “Because you mentioned stiffness” + before/after on four measures | answered “Nowhere much” on 2.6 |
The design is not vague about this — the spec annotates what each answer is supposed to drive, with weights. Collected in one place, it is a coherent scoring model that simply has not been built.
| Question | Mobility | Balance | Strength | Energy | Also drives |
|---|---|---|---|---|---|
| 2.2 Up from the floor without hands | ●● | ● | ●● | — | starting difficulty |
| 2.3 Shopping up a flight of stairs | — | — | ● | ●● | — |
| 2.4 Stand on one leg for 10 seconds | — | ●● | — | — | supported vs free-standing work |
| 2.5 By mid-afternoon | — | — | — | ●● | — |
| 2.6 Where stiff or achy | ●● | — | — | — | region targeting · fires 2.7 |
| 2.1 How much you move in a week | modifier across all four axes | starting difficulty | |||
| 1.3 Age band | shifts the benchmark each axis is graded against | — | |||
1.4 offers five goals. Four map cleanly — steadier → Balance, stronger → Strength, without stiffness → Mobility, energy → Energy. The fifth, “Lose weight, especially around my middle”, maps to nothing we measure. You deliberately kept that option because it is what this audience actually searches for — so the funnel's most commercially important answer is the one the diagnosis cannot speak to. Decide now whether it routes to Strength (the honest mechanism for body composition) or gets its own reconciliation line.
On 2.4, options are Yes either leg / A few seconds / Not really / I've never tried. The first three are a performance scale; the fourth is missing data. Scoring it as the worst answer would mark a woman who simply never attempted a one-leg stand as having the worst balance in the funnel, and send her down the balance track. It should be treated as unknown and imputed from 2.2, or prompt her to try it.
2.2 — getting off the floor — feeds Mobility, Strength and Balance. It is a good clinical marker, but it means a single misread answer moves three quarters of the diagnosis at once, and it makes the axes correlate with each other rather than measure independently. Worth either adding a second strength item or lowering its balance weight.
Both 2.2 and 2.3 end in avoidance answers — “I avoid getting down there”, “I avoid stairs with anything heavy”. Avoidance usually does indicate inability, so scoring it low is defensible, but it is a judgement rather than a measurement, and it is the answer most likely to come from fear rather than capacity. It is also the answer where getting the diagnosis wrong costs the most trust.
This is the smallest thing that would make the screens honest. Every axis scores 0–100.
1 − (regions ÷ 6), with “Nowhere much” = 1.0.2.11's copy already does something no competitor does: it lets the measurement disagree with the stated goal — “you said you want to be stronger; your answers say strength is holding up, it's stiffness making everything feel harder.” That sentence only lands when it is true. Wire it as three cases: goal matches the weakest axis (confirm it), goal is a different axis (reconcile, as the copy does now), or the goal is weight (no axis — say what you can honestly say instead).
| Screen | Must become |
|---|---|
| 1.6 | Goal echoed from 1.4; which of bone/muscle/tendon/cartilage to lead on |
| 2.11 | Four scores + labels, the verdict headline, and the goal-vs-measurement reconciliation |
| 3.6 | The statistic matched to the winning axis — falls, walking speed or body composition |
| 5.2 | Three priorities from the two weakest axes + her stiffness regions; safety line assembled only from what she actually selected on 2.10 and 2.8 |
| 5.3 | Milestones ordered by axis rather than a fixed four |
| 5.5 | Before/after rows built from her own 2.2–2.5 answers; “because you mentioned…” from 2.6 |
Not the scoring — the safety line on 5.2. Everything else is a personalisation that is merely generic until it is built. That line is the only place the quiz tells a woman something factually untrue about her own body, and it does it in the two areas where this audience is most sensitive: continence and back pain. It can be fixed today by assembling the sentence from her 2.8 and 2.10 answers, without any scoring model at all.