EMPATHIZE → DEFINE → TEST · SURVEY v2 RESULTS · n = 134 kept of 150

What the survey says — need, content by state, wearables, product, price.

Every statistic below is produced by the pre-registered analysis pipeline: descriptives, Ulwick opportunity scores, the McNemar test on state-dependent content preference, TAM reliabilities and betas, Kano classification, Van Westendorp price points, purchase intent — and the likelihood ratios each pre-registered threshold triggers in the decision engine.

Exclusions: 16 attention-check fails · 0 speeders (< ⅓ median time; median 7.4 min) · wearable owners 63 (47%) · US/EU 80 · China 41 · other 13

/ 01Block B · the need side

Who needs it, when, and what they do today

GAD-2 positive (≥ 3)

38%

n = 51 · mean 2.1/6 — screener, not diagnosis

PSS-4 perceived stress

6.54

/16 · SD 2.66 · r(GAD-2, PSS-4) = 0.69 (convergent validity)

Need a calming moment ≥ weekly

87%

B7 · frequency of "want something to calm down"

Already use music/audio ≥ weekly

90%

C1 · the behaviour the product builds on

When it strikes (B3, multi)

bed / work / evening dominate → two jobs: acute-at-night and daytime wind-down.

What is hired today (B8, multi)

Music is the most common coping behaviour; scrolling is the real competitor.

Outcome opportunity scores (B9 · Ulwick: Imp + max(Imp − Sat, 0); ≥ 6 = under-served)

Under-served (≥ 6): Stop racing thoughts · near threshold (5.5–6): Calm down within minutes, Not make things worse, Know it is actually working · adequately served: Almost no effort, Fall asleep faster, Discreet — not 'therapy'.

Somatic / cognitive pattern (B6)

Racing thoughts 78% · sleep 59% · racing heart 39% · chest 34% · can't stand silence/noise 18%

Triggers (B5)

Work / always reachable 67% · uncertainty 46% · money 31% · family 22% · overstimulation 19% · no specific trigger 15%

Where (B4)

Home off-hours 51% · home working 40% · office 34% · transit 13%

/ 02Block C · content by arousal state · the A1 test

Real song vs neutral sound — the answer depends on the state

C3 · "which would you most want to hear?" (within-subject, order randomised)

Real song chosen

24% → 47%

acute [17%–31%] → wind-down [39%–56%] · McNemar χ² = 14.29, p = <0.001 (b = 16, c = 47)

Neutral sound chosen

46% → 19%

GAD-2-positive segment, acute: real song 16% · neutral 49%

Value of real artists (C5, top-2)

35%

matters "a lot / extremely" — the catalogue is an asset for the wind-down job, not the acute one.

C4 · reactions when very anxious (−2 makes it worse … +2 helps) — the acoustic spec

Negatives = what the neutral-sound variant must not contain (sharp transients, beat, lyrics, low hum). Positives = slow tempo and adaptivity itself.

/ 03Block D · wearables & measurement · A3

Would they connect the watch — and does the number matter?

Owners who would connect (D1 top-2)

57%

95% CI [45%–69%] · non-owners who'd consider buying one: 21%

Value of a measured result (D4 top-2)

54%

"your heart rate settled 92 → 68"

Trust in stress / HRV score

3.06

/5 among owners · daily/weekly checkers 54%

Privacy concerned (D5 ≥ 4)

18%

heart data to a music app

The action gap (D3, owners) — what happens after a "high stress" reading

Nothing / just note it 38% · breathe 30% · walk 18% · it makes me more anxious 18% · don't look 13%

A closed-loop product lives exactly in this gap: the score exists, the action doesn't.

/ 04Block E · why use · why not

Acceptance, feature classification, form and barriers

TAM (7-pt) · α PU 0.68 · PEOU 0.68 · BI 0.75

BI top-2 25% overall · 32% owners · BI ~ PU β 0.52 + PEOU β 0.34, R² 0.64

Kano map (Better / Worse)

Feature	A	O	M	I	R	Class	Better / Worse
K1 adapts to heart rate	38	31	25	31	4	A · GAD-2+: A	0.55 / 0.45
K2 shows before/after	29	43	28	20	7	O · GAD-2+: M	0.60 / 0.59
K3 real songs, calm versions	43	4	8	48	27	I · GAD-2+: I	0.46 / 0.12
K4 neutral artifact-free sound	48	12	10	55	8	I · GAD-2+: I	0.48 / 0.18
K5 breathing cue	28	7	11	68	15	I · GAD-2+: I	0.31 / 0.16
K6 offline / no account	11	13	59	44	4	M · GAD-2+: I	0.19 / 0.57

Which form would you actually use? (E3)

Relative advantage (E4 top-2): 60% say "better / much better than what I use now".

Barriers (E5, up to 3)

/ 05Block F · how much · A4

Price sensitivity, purchase intent, payment model

Van Westendorp · US / EU (USD per month)

OPP $6.28 · IPP $7.37 · acceptable range [$5.19, $8.46] · n = 80

Van Westendorp · China (RMB per month)

OPP ¥15.47 · IPP ¥21.38 · acceptable range [¥14.16, ¥28.6] · n = 41

Purchase intent top-2 · US/EU @ $6.99

35%

discounted ×0.5 for hypothetical bias → 18%

Purchase intent top-2 · China @ ¥30

15%

probe market only

Wearable owners

30%

top-2 purchase intent

Currently paying for any app (F2)

12%

cancelled 28% · free only 44%

Preferred payment model (F9)

Bundled + employer = 33% → the C2 / B2B2C route is not a fallback, it is a preference for a large minority.

/ 06Pre-registered thresholds → likelihood ratios

What this dataset would do to the beliefs

Each row is a threshold fixed before fieldwork (survey analysis plan). The LR column is what enters the decision engine — nothing else from this page does.

Item	Node	Pre-registered rule	Observed	LR
C3 sc1	A1a	real song ≥50% → 3 · ≤35% → 0.33	24%	0.33
C3 sc2	A1b	real song ≥50% → 2 · ≤35% → 0.5	47%	1.0
C5	A1b	top-2 ≥40% → 1.3	35%	1.0
D1	A3	owners top-2 ≥60% → 2 · ≤40% → 0.5	57%	1.0
D4	A3	top-2 ≥50% → 1.3	54%	1.3
D5	A3	≥4 for >50% → 0.8	18%	1.0
E1 BI	A3	≥40% overall or ≥55% owners → 1.8	25% / owners 32%	1.0
E2 K1/K2	A3	attractive or one-dimensional → 1.5	A/O	1.5
E2 K3	A1a	reverse in GAD-2+ segment → 0.7	I	1.0
E4	A3	top-2 ≥50% → 1.5	60%	1.5
E5	A4	'another subscription' >40% → 0.8	49%	0.8
F4–F7 PSM	A4	range ∋ $6.99 → 1.5 · OPP<$4 → 0.7	[5.19, 8.46] OPP 6.28	1.5
F8	A4	US/EU top-2 ≥30% → 2 · ≤15% → 0.5	35%	2.0
F9 / E3	C2 weight	bundled+employer ≥35% or C2 ≥ C1 → raise C2	33% · C2 45% vs C1 32%	raise C2

Combined survey LR per node (product of triggered rows)

A1a ×0.33 · A1b ×1.0 · A3 ×2.92 · A4 ×2.4

These outcomes are applied by default in the decision engine (scenario "Survey as analysed"); the efficacy pilot and the legal check remain pending and are what the gates now wait for.