Expert report · Wave 1
Executive answer
They recognise the social barrier. We haven't shown that situational comms make anyone switch.
Keep push–pull. Open on a situation they live, then answer taste, satisfaction, cost or habit.
All deliverables
112
stimuli for 28 respondents
24 / 35
pairs won by product-led
28 / 13
cohort mean / SD
Finding 06
Two jobs: does the qual pattern hold up, and who's worth another interview.
Use case 1
Qual at scale
See whether what you heard in interviews holds up when you score it at scale.
Use case 2
Leads for IDIs
Who breaks the pattern, why, and what to ask them first — section 07.
Use case 3
Upload & compare
Put your own written findings next to this: what agrees, what conflicts, what's missing.
Not for
Stakeholder decks
Working material. Read it without the method and it will mislead you.
Finding 01 · Decision summary
Good territory to build on. Not proof anyone will move.
Recognised
Smell and social moments give people something real to react to.
Understood
Product-led wins 24 of 35 head-to-heads.
Not proven
44 up, 20 down against their own starting point. Intent was barely asked.
Finding 01b · Answers to the brief
Straight answers. Where this wave can't answer, we say so instead of guessing.
Core question
It names a struggle they recognise — just not the one that keeps them smoking. Keep the territory, but don't expect it to switch anyone on its own.
Evidence · Smell and awkward moments come up unprompted. When they tell us why they stay, it's taste, satisfaction, habit, cost, convenience.
Two of the four. Car and stuck-indoors-in-bad-weather feel lived. Waiting splits the room. The elevator doesn't land.
Evidence · Car 3 lived / 3 didn't get it · Home 4 / 2 · Waiting 2 lived, 1 said it feels like an ad, 1 didn't get it · Workplace 2 / 1 · Elevator 1, didn't get it.
Indoors when it's raining, and the car. The balcony works as an indoors-versus-outdoors feeling — nobody actually talked about balconies.
Evidence · Going out for a cigarette at work and coming back in; stubbing one out when a passenger complains.
Cost, the ritual and taste of a cigarette, the hassle of ashtrays and charging, and where you're allowed to smoke at all.
Evidence · Asked why they don't switch, unprompted: price, flavour, ritual, convenience, indoor rules.
Yes. Smell, and not bothering the people around you, get read back to us without any help.
Evidence · Stopping when someone complains; wanting nicotine without leaving the group.
They understand it's meant to fix the problem; we rarely pushed on whether they believed it. Understood isn't the same as believed.
Evidence · Believability wasn't coded in most sessions — that's a hole in our evidence, not a fail.
Inside one execution, yes. Across executions, no — the product-led one carries the story on its own, the situation-only one doesn't.
Evidence · Product-led beat situation-led in most head-to-heads.
We never ran the car executions as A/B/C copy variants.
Evidence · The coded executions are TST and FCTN only.
Neither was shown, so we can't separate the craft from the situation.
Evidence · No anime or copy-only stimulus in the coded material.
Where we asked: 5 try, 1 search, 2 notice, 2 ignore. We didn't ask in 18 of 28 sessions — the weakest part of the evidence.
Evidence · Intent only exists where we actually put the post-exposure question.
No. People agree the problem is real; almost nobody tells us what they'd do next.
Evidence · Some scores go up, some go down, no consistent movement.
Too thin to say. Droppers: 2 try / 2 notice / 1 ignore / 10 not asked. Dualists: 2 try / 1 search / 1 ignore / 5 not asked.
Evidence · 28 interviews: 15 droppers, 9 dualists, 4 cigarette-only.
From claims people can check against their own experience. The situation earns a nod; the product earns belief.
Evidence · Product-led wins most head-to-heads; people described the situational ones rather than believed them.
The trigger is social. The barrier is product performance.
Evidence · Why they stay: taste, habit, cost, ease of buying, no ashtray, indoor use.
Taste, ritual, satisfaction — not the social friction we're writing to.
Evidence · Other products get described as bitter or just not their taste.
It's differentiated space, but the wrong lever on its own. Open with the situation, close with product performance.
Evidence · No competitive comms audit in this material.
The output question
Right trigger, wrong lever. Nothing here shows that naming the friction moves anyone. The model gets us in the door; product performance has to do the persuading.
Evidence · Recognition is there across 28 interviews; we captured intent in 10.
Low confidenceWe never fielded copy variants, anime or copy-only versions, the balcony context, or a competitive audit — so none of that is in here.
Finding 01c · The visual system
Keep the situation. Remove the judgement.
The one rule from the data
Nobody on the left is being judged.
Alone with the cost: outside, waiting, walking, enduring. No grimaces. No coughing. No side-eye.
01 · Workplace · lead execution
席を外さない一服を。
No smoking room left inside the company. The highest-recognition sacrifice in the interviews.
Push
一服のたびに、席を外す。
Every smoke, I leave the room.
Pull
席を外さない一服を。
A smoke that doesn't leave the room.
77
Importance
56
Satisfaction
Keep the situation. Remove the judgement.
02 · Home · strongest pull
雨の日は、部屋で。
The home pull scored highest. The cost is being forced outside — not being alone.
Push
雨の日も、ベランダで。
Rainy days too — on the balcony.
Pull
雨の日は、部屋で。
Rainy days — in the room.
76
Importance
80
Satisfaction
Inside versus outside. Never lonely versus social.
03 · Waiting · friends
話の途中で、一服を。
The cost people named was missed time and exclusion — not keeping friends waiting.
Push
一服のあいだに、話は進んでる。
While you're out for a smoke, the story moves on.
Pull
話の途中で、一服を。
A smoke in the middle of the story.
75
Importance
55
Satisfaction
You do not delay them. You miss it.
04 · Car · drive-break version
停まらずに、一服を。
The pull had the strongest contextual fit. The push only works as a safe roadside stop.
Push
一服のために、一度停まる。
Stop once, for a smoke.
Pull
停まらずに、一服を。
A smoke without stopping.
70
Importance
46
Satisfaction
Never show smoking at the wheel.
05 · Taste proof · switching lever
旨さを、自分仕様に。
Smell-freedom is table stakes. Taste, customisation, battery and price are the reasons to switch.
Push
選べる4つの加熱モード
Four heating modes to choose from.
Pull
旨さを、自分仕様に。
Your taste, set to you.
8 respondents named taste unaided. “I can change my taste. This is the most attractive one to me.”
01 / Primary
Dualists & current heated-tobacco users
Run situations 1–4 as the backdrop. Lead with taste proof. Their social job is solved; they switch for taste, customisation, battery and price.
妥協してきた旨さを、自分仕様に。
The taste you've been compromising on, set to you.
02 / Secondary
RMC soloists
Run situations 1–3 as the lead, then taste in its coffee-morning form. Preserve the relaxation ritual and the robust taste cue.
紙巻きに一番近い味わい。手放すのは、我慢だけ。
The closest taste to a cigarette. Give up only the enduring.
01
No bystander reactions
02
No smoking at the wheel
03
No lonely framing
04
No anime or illustration
05
No device-only key visuals
06
No nature overlays
07
No health or harm language
08
No bold type or exclamation marks
Finding 02
Everything is measured against each person's own starting point, so one generous scorer can't skew the story.
Every score is compared to where that same person started.
We read lift-from-baseline first. The thirds (21 and 33 here) are a sense-check, nothing more.
IndicativeThese bands come from our own wave-1 data — there's no Japanese norm behind them.
Finding 03
1 PS and the pair-4 executions trail. On raw scores, the order mostly tracks who talked most.
| Stimulus | n | Mean score | Mean lift | Lifts | Depresses | |
|---|---|---|---|---|---|---|
 3 PLPL | 8 | 35 | +14 | 3 | 0 | |
 2 PLPL | 8 | 35 | +13 | 7 | 0 | |
FCTN AFCTN | 8 | 35 | +13 | 3 | 1 | |
 3 PSPS | 8 | 32 | +11 | 3 | 1 | |
 TST CTST | 7 | 31 | +10 | 3 | 0 | |
FCTN BFCTN | 6 | 37 | +9 | 3 | 1 | |
 4 PLPL | 2 | 31 | +5 | 1 | 0 | |
 2 PSPS | 8 | 26 | +5 | 5 | 2 | |
 FCTN CFCTN | 4 | 28 | +4 | 3 | 1 | |
 1 PLPL | 11 | 27 | +4 | 4 | 2 | |
TST BTST | 3 | 28 | +3 | 2 | 1 | |
 1 PSPS | 11 | 24 | +1 | 2 | 3 | |
 TST ATST | 13 | 27 | +1 | 4 | 4 | |
| 4 PL APL | 6 | 21 | -3 | 1 | 2 | |
 TST DTST | 1 | 22 | -4 | 0 | 0 | |
 4 PSPS | 8 | 19 | -6 | 0 | 2 |
Low confidenceOne to five reads per stimulus, and FCTN C is a single interview. Use this to shortlist for wave 2, not to call a winner.
Finding 04
Product-led won 24 of 35 head-to-heads across all 28 interviews. Same person both times, so talkativeness doesn't skew it.
Top bar problem-led, bottom bar product-led. Blue where product-led wins.
Indicative35 pairs across 28 people. The direction is real; the size of it isn't.
Appendix A1
The individual scores behind every claim above — including the thin cells.
| Resp. | Base | 3 PL | 2 PL | FCTN A | 3 PS | TST C | FCTN B | 4 PL | 2 PS | FCTN C | 1 PL | TST B | 1 PS | TST A | 4 PL A | TST D | 4 PS |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| R1 | 7 | +49 | +31 | +43 | +37 | ||||||||||||
| R2 | 42 | -4 | -1 | +3 | -2 | ||||||||||||
| R3 | 27 | +21 | 0 | -12 | -12 | ||||||||||||
| R4 | 6 | +2 | 0 | +10 | 0 | ||||||||||||
| R5 | 25 | +27 | +13 | +1 | +5 | ||||||||||||
| R6 | 44 | +11 | -23 | -11 | -12 | ||||||||||||
| R7 | 24 | +4 | -10 | +1 | -6 | ||||||||||||
| R8 | 10 | +1 | -3 | +5 | -2 | ||||||||||||
| R9 | 7 | +27 | +12 | +21 | +7 | ||||||||||||
| R10 | 23 | +2 | -2 | +5 | +3 | ||||||||||||
| R11 | 11 | 0 | +5 | 0 | 0 | ||||||||||||
| R12 | 15 | +53 | +6 | +10 | +10 | ||||||||||||
| R13 | 14 | +10 | +2 | +4 | +9 | ||||||||||||
| R14 | 24 | -5 | -14 | -20 | -18 | ||||||||||||
| R15 | 48 | -10 | -13 | -8 | -6 | ||||||||||||
| R16 | 24 | +46 | -9 | -1 | -14 | ||||||||||||
| R17 | 36 | -19 | -17 | -18 | -25 | ||||||||||||
| R18 | 33 | +10 | -17 | -9 | -2 | ||||||||||||
| R19 | 27 | -4 | +1 | +24 | -2 | ||||||||||||
| R20 | 22 | +19 | +12 | +8 | +14 | ||||||||||||
| R21 | 18 | +29 | +26 | +23 | +21 | ||||||||||||
| R22 | 31 | +9 | +12 | +5 | +13 | ||||||||||||
| R23 | 15 | +21 | +19 | +23 | +18 | ||||||||||||
| R24 | 26 | -5 | -2 | -4 | -7 | ||||||||||||
| R25 | 12 | +19 | +17 | +15 | +12 | ||||||||||||
| R26 | 35 | +12 | +9 | +10 | +7 | ||||||||||||
| R27 | 20 | +5 | +4 | +6 | +8 | ||||||||||||
| R28 | 28 | +5 | +3 | +8 | +2 |
Colours follow whichever yardstick you picked in section 02. Dashed cells mean we never showed it to them.
Appendix A2
R5 and R9 respond; R8 and R10 stay flat; R3 and R6 move down.
R1
baseline 7
4 of 4 stimuli lift, 0 depress. Mean stimulus score 47.
R2
baseline 42
0 of 4 stimuli lift, 0 depress. Mean stimulus score 41.
R3
baseline 27
1 of 4 stimuli lift, 2 depress. Mean stimulus score 26.
R4
baseline 6
1 of 4 stimuli lift, 0 depress. Mean stimulus score 9.
R5
baseline 25
2 of 4 stimuli lift, 0 depress. Mean stimulus score 37.
R6
baseline 44
1 of 4 stimuli lift, 3 depress. Mean stimulus score 35.
R7
baseline 24
0 of 4 stimuli lift, 1 depress. Mean stimulus score 21.
R8
baseline 10
0 of 4 stimuli lift, 0 depress. Mean stimulus score 10.
R9
baseline 7
3 of 4 stimuli lift, 0 depress. Mean stimulus score 24.
R10
baseline 23
0 of 4 stimuli lift, 0 depress. Mean stimulus score 25.
R11
baseline 11
0 of 4 stimuli lift, 0 depress. Mean stimulus score 12.
R12
baseline 15
3 of 4 stimuli lift, 0 depress. Mean stimulus score 35.
R13
baseline 14
2 of 4 stimuli lift, 0 depress. Mean stimulus score 20.
R14
baseline 24
0 of 4 stimuli lift, 3 depress. Mean stimulus score 10.
R15
baseline 48
0 of 4 stimuli lift, 3 depress. Mean stimulus score 39.
R16
baseline 24
1 of 4 stimuli lift, 2 depress. Mean stimulus score 30.
R17
baseline 36
0 of 4 stimuli lift, 4 depress. Mean stimulus score 16.
R18
baseline 33
1 of 4 stimuli lift, 2 depress. Mean stimulus score 29.
R19
baseline 27
1 of 4 stimuli lift, 0 depress. Mean stimulus score 32.
R20
baseline 22
4 of 4 stimuli lift, 0 depress. Mean stimulus score 35.
R21
baseline 18
4 of 4 stimuli lift, 0 depress. Mean stimulus score 43.
R22
baseline 31
3 of 4 stimuli lift, 0 depress. Mean stimulus score 41.
R23
baseline 15
4 of 4 stimuli lift, 0 depress. Mean stimulus score 35.
R24
baseline 26
0 of 4 stimuli lift, 0 depress. Mean stimulus score 22.
R25
baseline 12
4 of 4 stimuli lift, 0 depress. Mean stimulus score 28.
R26
baseline 35
3 of 4 stimuli lift, 0 depress. Mean stimulus score 45.
R27
baseline 20
1 of 4 stimuli lift, 0 depress. Mean stimulus score 26.
R28
baseline 28
1 of 4 stimuli lift, 0 depress. Mean stimulus score 33.
Score this any other way and a Japanese sample flattens even harder.
Finding 07
22 respondents don't fit the main pattern. For each: why, and the question we'd open with.
Insight bubbles
22 leadsEach bubble is someone who doesn't fit. Bigger bubble, more contradictions. Tap one for the reason and the opener we'd use when we call them back.
Why R1 stands out
Highly responsive4 of 4 stimuli lifted them, with none depressing — an unusually responsive read.
Open the interview with
“You reacted strongly to most of these. Which one would actually change what you buy, and why?”
R1
baseline 7
Why they stand out
4 of 4 stimuli lifted them, with none depressing — an unusually responsive read.
Open the interview with
You reacted strongly to most of these. Which one would actually change what you buy, and why?
R2
baseline 42
Why they stand out
Problem-led framing scored higher than product-led in 1 of their pairs — the reverse of the cohort pattern. Nothing shown to them moved the reading in either direction.
Open the interview with
Walk me through the last time the smell or the situation actually caused you a problem. What did you do?
R5
baseline 25
Why they stand out
2 of 4 stimuli lifted them, with none depressing — an unusually responsive read.
Open the interview with
You reacted strongly to most of these. Which one would actually change what you buy, and why?
R8
baseline 10
Why they stand out
Problem-led framing scored higher than product-led in 1 of their pairs — the reverse of the cohort pattern. Nothing shown to them moved the reading in either direction.
Open the interview with
Walk me through the last time the smell or the situation actually caused you a problem. What did you do?
R9
baseline 7
Why they stand out
3 of 4 stimuli lifted them, with none depressing — an unusually responsive read.
Open the interview with
You reacted strongly to most of these. Which one would actually change what you buy, and why?
R10
baseline 23
Why they stand out
Problem-led framing scored higher than product-led in 2 of their pairs — the reverse of the cohort pattern. Nothing shown to them moved the reading in either direction.
Open the interview with
Walk me through the last time the smell or the situation actually caused you a problem. What did you do?
R11
baseline 11
Why they stand out
Problem-led framing scored higher than product-led in 1 of their pairs — the reverse of the cohort pattern. Nothing shown to them moved the reading in either direction.
Open the interview with
Walk me through the last time the smell or the situation actually caused you a problem. What did you do?
R12
baseline 15
Why they stand out
Problem-led framing scored higher than product-led in 1 of their pairs — the reverse of the cohort pattern. 3 of 4 stimuli lifted them, with none depressing — an unusually responsive read.
Open the interview with
Walk me through the last time the smell or the situation actually caused you a problem. What did you do?
R13
baseline 14
Why they stand out
2 of 4 stimuli lifted them, with none depressing — an unusually responsive read.
Open the interview with
You reacted strongly to most of these. Which one would actually change what you buy, and why?
R14
baseline 24
Why they stand out
3 of 4 stimuli pushed them below their own opening score.
Open the interview with
What was it about those messages that put you off rather than pulled you in?
R15
baseline 48
Why they stand out
Problem-led framing scored higher than product-led in 1 of their pairs — the reverse of the cohort pattern. 3 of 4 stimuli pushed them below their own opening score.
Open the interview with
Walk me through the last time the smell or the situation actually caused you a problem. What did you do?
R16
baseline 24
Why they stand out
Problem-led framing scored higher than product-led in 1 of their pairs — the reverse of the cohort pattern.
Open the interview with
Walk me through the last time the smell or the situation actually caused you a problem. What did you do?
R17
baseline 36
Why they stand out
4 of 4 stimuli pushed them below their own opening score.
Open the interview with
What was it about those messages that put you off rather than pulled you in?
R19
baseline 27
Why they stand out
Problem-led framing scored higher than product-led in 1 of their pairs — the reverse of the cohort pattern.
Open the interview with
Walk me through the last time the smell or the situation actually caused you a problem. What did you do?
R20
baseline 22
Why they stand out
4 of 4 stimuli lifted them, with none depressing — an unusually responsive read.
Open the interview with
You reacted strongly to most of these. Which one would actually change what you buy, and why?
R21
baseline 18
Why they stand out
4 of 4 stimuli lifted them, with none depressing — an unusually responsive read.
Open the interview with
You reacted strongly to most of these. Which one would actually change what you buy, and why?
R22
baseline 31
Why they stand out
3 of 4 stimuli lifted them, with none depressing — an unusually responsive read.
Open the interview with
You reacted strongly to most of these. Which one would actually change what you buy, and why?
R23
baseline 15
Why they stand out
4 of 4 stimuli lifted them, with none depressing — an unusually responsive read.
Open the interview with
You reacted strongly to most of these. Which one would actually change what you buy, and why?
R24
baseline 26
Why they stand out
Nothing shown to them moved the reading in either direction.
Open the interview with
Nothing we showed changed how you felt. What would have to be true for it to matter?
R25
baseline 12
Why they stand out
4 of 4 stimuli lifted them, with none depressing — an unusually responsive read.
Open the interview with
You reacted strongly to most of these. Which one would actually change what you buy, and why?
R26
baseline 35
Why they stand out
3 of 4 stimuli lifted them, with none depressing — an unusually responsive read.
Open the interview with
You reacted strongly to most of these. Which one would actually change what you buy, and why?
R27
baseline 20
Why they stand out
Problem-led framing scored higher than product-led in 1 of their pairs — the reverse of the cohort pattern.
Open the interview with
Walk me through the last time the smell or the situation actually caused you a problem. What did you do?
IndicativeA lead is a hunch, not a segment. The value is knowing who to call back.
Appendix A3
Everything on this page comes from a 60-minute conversation that asks about the struggle before it shows a single stimulus.
We wanted three things: do they recognise the struggle, does the comms feel relevant, does it make switching feel doable. Conversational, anchored in what they actually do, neutral probes.
Introduction5 min
Frame: how you choose and use your packs.
Background5 min
Who they are; what they smoke now and used to smoke.
The job, unprompted10 min
The last time their current product let them down. Nothing shown yet — this becomes the baseline.
Context stimuli10 min
Four unbranded context images: what's going on here, is this you?
Execution evaluation20 min
Two executions each — the scores on this page.
Trade-off10 min
Visuals matched to whatever barrier they raised.
IndicativeStage 5 is where the scores come from; stage 3 is what we compare them against. This is working material for researchers, not a finished report.
Finding 05
Why product-led lands: odour, colleagues, and not annoying the people you drive with.
Social exclusion is the live wire
“I've been in situations where I have friends go away or be excluded from a conversation because they don't like the smell of smoke, so me and my smoke buddies just continued the conversation without them, which I feel very, very sad about.”
Inertia, not loyalty
“I'm using it because it's not broken. So it's kind of like a new phone — a new model comes out, yeah, maybe I'll buy it. I already know it works, so I don't really see a need to change it.”
Product-led framing lands as relief
“It feels quite relevant and appealing because I often drive with others and want to avoid annoying them with smoke or smell.”
Odour as a work-life risk
“It feels very relevant in a workplace setting because avoiding odour is important when talking to colleagues or clients after taking a break.”
Already living the behaviour
“It's definitely appealing to me, and I do it myself sometimes. I don't smoke in the car very often, but when I do want to, I always like to have my IQOS with me.”
Switching triggers are functional
“I would say cheaper products, or easier use — less need for charging.”
IndicativeCleaned up for filler words only. These were all said in English, so we're missing Japanese nuance.
Appendix A4
Five problems we flagged. Each is either fixed in this report or marked where it sits.
The scoring bands were imported, not derived
Was
Scores were read against imported cut-offs built on a different cohort in a different category, which called most of the wave one long suppression problem and separated nothing.
Now
Those thresholds are gone. Every chart is re-normed to this wave: within-respondent lift against each person's own baseline, with cohort tertiles at 17 and 29 as a cross-check.
Wave 1 is working material, not a market read
Was
The first read treated the wave as if it were a representative market result.
Now
No finding on this page is stated as a market result. Wave 1 is presented as a method demonstration and expert working material, and every cut carries an explicit confidence label.
Profile cuts were reported as if they were segments
Was
Age, gender and product-use splits were being read as differences, on cells of one to three people.
Now
Profile cuts are marked indicative and shown with their n. Nothing with n < 5 is described as a difference; it is described as something to test.
Sensitivity and intensity were treated as measures
Was
S and I sat alongside K as if they were three independent readings.
Now
43 of 54 S/I values are the default 50, so they carry almost no information. They are reported as a data-quality note, not as findings.
Stimulus rank was read off raw score
Was
A stimulus scoring 50 for a highly expressive respondent outranked one scoring 20 for a reserved one.
Now
The leaderboard ranks on mean lift from baseline, with raw score kept alongside so both are auditable.
One more: most sensitivity and intensity scores sit at the default 50 — the capture didn't work, so ignore those.
Appendix A5
Five fixes, in order of how much they'd change what we're allowed to claim.
Finding 08
Three questions: what agrees, what conflicts, what you missed.
What agrees?
Take each of your written insights and find it above. If it shows up here too, that's an independent second read.
What conflicts?
Start with sections 04 and 06 — reversals and flat respondents are where qual usually fails to reproduce.
What's missing?
Section 07 — patterns in the scores a 16-interview qual would never show you.
Quickest way: download the text file, give it to your AI alongside your own report, and ask where they disagree.