When should a startup pivot?

The evidence framework leaves one decision: refine the current direction or change it. Iterate when the core problem is validated and the solution needs work. Pivot when the problem, the market or the resource row is red. Each choice has a plan: what you keep, what you change, how you will validate the change, and the criterion that would make you exit. The one thing not allowed is neither.

What is the difference between iterating and pivoting?

The evidence framework leaves one decision: refine the current direction or change it. Iterate when the core problem is validated and the solution needs work. Pivot when the problem, the market or the resource row is red. Each choice has a plan: what you keep, what you change, how you will validate the change, and the criterion that would make you exit. The one thing not allowed is neither.

How do you plan a pivot?

The evidence framework leaves one decision: refine the current direction or change it. Iterate when the core problem is validated and the solution needs work. Pivot when the problem, the market or the resource row is red. Each choice has a plan: what you keep, what you change, how you will validate the change, and the criterion that would make you exit. The one thing not allowed is neither.

Iterate or pivot · Product Strategy

§The question: given the rows, do we refine this or change it?

§Eric Ries gave the decision its name, or persevere, and the name has done some damage, because it made the pivot sound dramatic and the persevere sound passive. Neither is true. Both are plans, and both are chosen from the same grid.

§ 2.1Triggers#

§Iterate when: the core problem is validated, the solution needs refinement rather than replacement, market timing is right, the team can build it, and the resources are sufficient or nearly so. The reds, if any, are in solution fit, and they are specific: a feature that failed, a gesture nobody understood, a unit of value that did not land.

§Pivot when: problem validation is weak, market barriers are too high, the solution is indistinguishable from what exists, the capability gap is large, or the resources required exceed what can be raised. A single red in problem validation or resource match is usually enough. Two reds anywhere almost always are.

Weak reasoning	Strong reasoning
"The feedback was mixed, so let's keep going and see."	"Problem row is green, solution row is red on three specific features that traced to no question. Iterate: keep the problem and the profile, drop the token unit, redesign check-in, test again."
"The market is huge, we just need to execute better."	"Resource row is red: the first version needs twice the budget and exceeds runway. Pivot to a version that depends on no one and ships in a month, or pivot the segment to one that pays sooner."
"We've come too far to change now."	"Sunk cost is not a row. The grid today reads the same as it would for a team starting fresh, and a team starting fresh would not choose this."

§ 2.2What a pivot is and is not#

§A pivot keeps something. Ries's catalog is useful precisely because it names what: a zoom-in pivot keeps one feature and makes it the product; a customer segment pivot keeps the product and changes who it is for; a platform pivot keeps the capability and changes the form; a value-capture pivot keeps everything and changes how money moves. A pivot that keeps nothing is not a pivot. It is a new company, and should be treated with the honesty that deserves.

§ 2.3The iteration plan#

§If the decision is to iterate, write four things.

§What we keep. The validated assumptions, the differentiators strangers named, the segment, the capabilities that showed. In the hospitality : the profile, the tier, reservations, the operation's appetite for anticipation.

§What we change. The solution elements that failed, the feature priorities, the sequence. The loyalty unit rebuilt on visits rather than tokens. Check-in redesigned before it enters. The card screen removed.

§How we validate the change. The metric per change, the success criterion, the time you will give it, the resources it needs. Not "test again"; "six new strangers, five of six complete check-in unaided, or it does not ship."

§How we would exit. The obstacles you expect, the fallback if they appear, the decision points, and the criterion that would make you stop. Annie Duke calls these kill criteria and argues that they are only trustworthy when written before the attempt, because after the attempt every criterion looks negotiable.

§ 2.4The pivot plan#

§If the decision is to pivot, write four different things.

§The new direction. The core , the target, the value proposition, the differentiators. In one paragraph, in the customer's words, using everything the sprint taught.

§What we still need to learn. The research gaps the new direction opens, what has to be validated before building, which experts to return to, what the market row now looks like.

§What it costs. Team implications, budget, timeline, capability gaps. A pivot that requires a person you do not have is a hiring plan first.

§How we tell people. The team first, in a conversation and not a memo. Then the stakeholders who backed the old direction, with the that changed it. Then, if the old direction had a public face, the market. A pivot communicated badly is read as panic. A pivot communicated with the grid attached is read as learning.

§ 2.5The decision, then the movement#

§The grid says where the argument is. The triggers say which way it points. The plan says what you will do about it. What remains is to decide, on a date, and move with the whole team behind the decision, including the people who argued for the other one. The next chapter is about writing it down so that the reasoning survives.

2. Iterate or pivot