Problem-to-Pillar Down-Select
Cluster raw problems, flip them into opportunities, then rank on four criteria to pick exactly three pillars.
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 95%
The decision engine at the heart of the strategy sprint. It converts a free-flowing dump of user and business problems into a small, defensible set of strategic bets. The mechanism forces the choice that strategy is supposed to force, and produces an auditable scoring table you can point to whenever someone asks 'why these three?'.
Origin
Chandra Janakiraman's operationalization of the 'strategy forces choice' idea from Rumelt and Porter, executed live in a working-group session.
Core principles
- 01Start from problems, not solutions — generate them free-flow, then let clusters emerge organically.
- 02Flipping a problem cluster into a positive opportunity framing reframes what you're building toward (e.g. 'difficulty finding things' becomes 'Discovery').
- 03Down-select hard: aim for three focus areas because it creates clarity and focus.
- 04Qualitative high/medium/low ('t-shirt') scores are fine — the value is the group debating and colliding on the scores together.
How to run it
- 1
Generate and cluster problems
After the share-out day, run a free-flowing session where everyone throws out what's holding users and the business back into a shared sheet. Over about an hour, jointly cluster related problems — you'll typically land on 10–15 clusters, each retaining its known sub-problems.
Pro tip Because you generated it organically, each cluster already carries its sub-problems, so nothing is lost in aggregation.
- 2
Flip clusters into opportunity framings
Each cluster is currently named as a problem. Rename it as a positive opportunity: 'people can't find things' → Discovery; 'people see content they don't like' → Relevance; 'people can't find friends' → Social Connection.
Pro tip The positive framing sets up the 'how might we' questions later and reads better to leadership.
- 3
Score on four criteria
Rank each opportunity on: (1) Expected impact to what matters; (2) Certainty of impact — how concrete the evidence that it's a real problem; (3) Clarity of levers — do you actually have an idea of how to solve it; (4) Uniqueness/differentiation — could another team or company build it better, making it generic?
Pro tip If data is thin, use high/medium/low t-shirt scores — the debate matters more than precision.
Watch out Do the scoring together as a working group; the alignment and 'collision' during scoring is what makes the outcome durable.
- 4
Sum and select the top three
Add the scores, sort, and take the top three as your strategic pillars — the remaining ~7–12 become the explicit not-focus list. Keep the full table for the doc appendix.
Pro tip The scoring table is your defensibility — when anyone questions the pillars, return to the criteria and reasoning, and even a change becomes easy to justify.
- 5
Translate pillars into 'how might we'
For each pillar, write two or three 'how might we' questions (e.g. 'how might we surface the best content in the right place'). This takes about an hour and feeds the design sprint.
Pro tip 'How might we' framing (fertile questions) opens the mind to solutions in a way 'how do we' does not — designers are already fluent in it.
In the wild
Janakiraman illustrates the flip with a social product: a cluster of 'can't find things' problems becomes Discovery, 'seeing disliked content' becomes Relevance, and 'can't find friends / feeling lonely' becomes Social Connection.
→ Three concrete opportunity areas ready to be scored and down-selected into strategic pillars.
Common mistakes
Picking five instead of three
Some go three-to-five, but Janakiraman pushes very hard for three because more pillars dilute clarity and focus.
Ignoring the differentiation lever
An opportunity with high impact but no unique capability produces something generic that a competitor can build better once you launch it.
Is it for you?
Best for
A cross-functional working group that has already gathered inputs and needs to convert a long problem list into three defensible bets in a single week.
Not ideal for
Aspirational, greenfield future bets where there is no existing product problem set to mine.
From the transcript
“you sort of do a joint clustering of sort of related problems and and you sort of create these uh potentially like typically in my…”
“the next step is to flip it into an opportunity Framing”
“a lot of problems in that area so so difficulty finding things becomes the cluster the problem cluster and Discovery becomes the opportunity sort of…”
“the second dimension is certainty of impact certainty of impact is basically how concrete is the evidence that this is a problem”
“the third one is also very important which is Clarity of levers like do you actually have an idea of how you would solve it…”
“you have to down select from like those 10 to 15 you know opportunity areas into ideally three you know maybe five but I would…”
“the key is you're doing it together as a strategy working group and you're sort of like you're debating like the scores”
From the episode
An operator’s guide to product strategy
Chandra Janakiraman (CPO at VRChat, ex-Meta, Headspace, Zyng