The Framework Limit Test
Two tripwires that tell you your product framework (JTBD, OKRs, DAU) has become the enemy.
- Difficulty
- Advanced
- Time to result
- ~months to results
- Steps
- 5
- Confidence
- 92%
Beykpour's real critique of Jobs To Be Done is not the framework — it's that any framework followed religiously stops serving product judgement and starts replacing it. He offers two concrete, observable tripwires for detecting when you've crossed that line, plus a structural remedy: exempt speculative bets from the governing metric entirely and run them as seed bets.
Origin
Kayvon Beykpour, drawing on being charged with defending and rolling out Jobs To Be Done at Twitter (a framework popularised by Clayton Christensen and the milkshake study) while not believing in how it was applied — and on Twitter's decade-long over-indexing on DAU as the governing metric.
Core principles
- 01Every framework at its limit, followed religiously, is unhelpful — you follow process for the sake of process and lose the forest for the trees.
- 02The charitable core of JTBD is just healthy product thinking: see the world through the customer's lens and their true alternatives. You don't need a milkshake to do that.
- 03OKRs as the sole governing principle fail the same way: you may be optimising the wrong metric, or blind to features hostile to customers.
- 04The answer is never a framework. Sometimes it's good old-fashioned judgment and product taste.
- 05Metrics and customer benefit are sometimes aligned and sometimes not; leadership exists to make that trade-off explicitly.
How to run it
- 1
Name the governing framework
Identify what actually decides what gets built: JTBD, OKRs, a north-star metric like DAU. Whichever it is, it is now the thing under test.
Pro tip The governing framework is the one people cite to end an argument, not the one on the wiki.
- 2
Tripwire one — subjectively bad decisions
Ask: is the organisation being incentivised to make decisions that a person with good product taste would call bad? If a non-trivial share of shipped decisions are ones you can't stand as a user or builder, something has to change.
Pro tip Beykpour's own example of a customer-hostile, metric-positive decision: Amazon burying order details behind a click-and-authenticate in the order confirmation email.
Watch out This tripwire assumes the assessor has good product taste — which is itself subjective. Someone senior must own it explicitly.
- 3
Tripwire two — bold bets get disincentivised
Ask: when you plan a bet, does the organisation see itself as disincentivised from making it succeed? If yes, either the strategy is wrong or the framework must be adjusted to accommodate bets that hurt the primary metric short term.
Pro tip Spaces was the archetype: putting a live-Spaces bar at the top of the app pushes tweets and ads down, so DAU and revenue fall — even though it enabled a whole new vector of content creation.
- 4
Diagnose which failure you have
When a tripwire fires, sort it: either bad judgement was applied inside a sound process, or the process is wrong, or the framework didn't even produce the right debate. Each has a different fix.
Pro tip 'The framework didn't lead to the right debate' is the most common and most invisible case.
- 5
Exempt the speculative bets and run them as seed bets
For projects that can't survive the governing metric, structure them like a startup and explicitly release them from OKRs. Community Notes and Fleets were run this way: 'don't worry about the OKRs, we're not going to judge you on the basis of your OKRs.'
Watch out There are pros and cons: an unmeasured team can drift. Some projects are so intertwined with the core that separating them does more harm than good.
In the wild
Twitter's ranked-timeline toggle would flip you to reverse-chronological — then silently pull the rug and put you back on the ranked timeline after roughly 24 hours. The team defended it because it was better for metrics: even if a user wanted reverse-chron, they spent more time in the app on ranked. Power users repeatedly said they hated it.
→ A textbook tripwire-one hit: metric-positive, customer-hostile. Twitter eventually took the baby steps toward giving users real control, and a persistent following timeline exists today.
For Spaces to work, people had to join live rooms — which required push notifications and prime real estate at the top of the app, pushing tweets and ads down. Under a DAU-quarter-over-quarter regime, that reads as pure damage.
→ Twitter had to structure Spaces (originally 'Hydra') outside the metric regime, then made it the number one priority of the company — and it shipped.
Common mistakes
Blaming the framework instead of the religiosity
Beykpour is not a JTBD fan, but he insists the failure is applying any framework to its religious extreme. Swapping JTBD for OKRs reproduces exactly the same pathology at a different address.
Judging speculative bets on the core metric
Bets worth making often hurt the primary metric short term. Measured on it, they will be killed before their real payoff — an entirely new vector of behaviour — can be observed.
Letting the team's incentive stand in for the customer's interest
Teams do what they are incentivised to do. When a metric-positive decision is customer-hostile, that is an incentive problem, not a people problem — and fixing the people fixes nothing.
Is it for you?
Best for
Product leaders whose org has adopted a single governing framework or north-star metric and is starting to ship things the leaders themselves dislike as users
Not ideal for
Small teams with no formal framework yet, where the immediate need is more structure rather than less
From the transcript
“every framework at its limit if followed to such a religious extent is just unhelpful you need to have Nuance in how you leverage these…”
“one is if the result of your framework is that subjectively bad decisions are being made then something's got to change”
“when you start imagining and planning for a bunch of bets that the organization then sees is like disincentivized to make successful then something's got…”
“we intentionally structured that like a startup like it was literally like we made a seed bet on Keith and his team and we were…”
From the episode
Twitter’s former Head of Product opens up: being fired, meeting Elon, changing stagnant culture, building consumer product, more
Kayvon Beykpour