Persistence and stubbornness look identical from the inside. Deciding in advance what would change your mind is the only reliable defence.
12 min read · Level 2
Persistence and stubbornness feel identical from the inside, and your judgement is compromised precisely because you have spent two years on it. The only reliable defence is deciding in advance what evidence would change your mind — criteria written in January are honest, the same judgement made in June gets renegotiated because by then you know how hard everyone worked and how close it feels.
Changing direction — a pivot — means materially altering the product, the customer, or the problem, while keeping what you have learned.
The difficulty is that persistence and stubbornness feel identical from inside. Both involve continuing despite discouraging evidence.
Because founders are told constantly that persistence is the defining virtue, and it often is. But the same trait applied to a genuinely dead idea consumes years.
The defence is not better judgement in the moment — your judgement is compromised by having spent two years on it. The defence is deciding in advance what evidence would change your mind, while you can still think clearly.
Writing the criteria in advance:
'If by 30 June we do not have 20 paying customers, or week-8 retention is still under 20%, we reconsider the customer segment rather than the product.'
Written in January, that is a clear-headed judgement. Written in June, after five months of effort, it would be renegotiated — because in June you know how hard you worked and how close it feels.
Signals worth taking seriously:
Signals that are NOT reasons to pivot: a slow month, three investor passes, a competitor raising, boredom, or a hard week.
And a pivot keeps what you learned. The customer relationships, the domain knowledge and the team usually survive. Founders often frame it as starting over, and it very rarely is.
First-time founders often pivot too frequently, treating every discouraging month as a signal. Nothing gets enough time to work, and each restart discards the accumulated learning.
The opposite mistake is more costly: continuing for years on flat retention because stopping feels like failure. Sunk cost is the strongest force acting on this decision and it is entirely backwards-looking.
The third: pivoting the product when the evidence points at the customer. Often the product is fine and the segment is wrong — and that is a much smaller change than it feels.
Specific numbers and a deadline. Written while you can still think clearly about it.
Flat, low retention across cohorts despite real product changes is the clearest signal available.
Some organic signal — a referral, an unprompted signup, someone chasing you — matters more than volume.
Often the product is fine for a different segment. That is a far smaller change than a full pivot.
'We decide on 30 June.' Then actually decide, rather than extending because it feels close.
Relationships, domain knowledge and team survive a pivot. It is rarely starting over.
Write the criteria at the start of any significant push. Review them on the date you set, not when you feel like it.
Do not make this decision during a single bad week, immediately after a rejection, or while exhausted. Wait for a clear head and then use the criteria you wrote.
Several widely known companies emerged from something else: Slack from an internal tool built while making a game, Instagram from a check-in app whose photo feature was the only used part, Shopify from software built to run a snowboard shop. In each case the evidence for the change was already in usage data.
A pivot keeps the customers, the domain knowledge and the team. It is rarely starting over, and the signal is usually already in front of you.
The most common unnecessary pivot is rebuilding a product that was fine for a different segment. Testing a new customer group with the existing product is dramatically cheaper than rebuilding, and founders skip straight to the expensive option because it feels more decisive.
Before rebuilding, test whether the same product lands with a different segment. It is the cheaper experiment and it is frequently the right one.
Create a free account to read the rest and apply it to your own startup.