All questions

Pivot, persevere, or shut down? How to decide

By Adi ShmorakUpdated 2026-10-08Read as markdown

Pivots are normal. Netflix, Slack and Airbnb all pivoted. So did most companies you've never heard of, because they pivoted too late.

The hard part isn't pivoting. It's deciding. Founders stay on the fence for months, half-committed to the old plan and half-flirting with a new one. That burns the two things a startup never has enough of: time and money.

Step 1: Set the thresholds before the data lands

This is the single most important rule, and almost nobody follows it.

Before you run an experiment, write down what result means GO, what means ITERATE, and what means PIVOT. It's the only way to protect yourself from your own optimism. Once the numbers arrive, every founder finds a reason why 12 is actually pretty good.

Some starting points I use for early validation:

Adjust them to your market. Just write them down first.

• GO: you hit the bar. Invest more.
• ITERATE: close, but the message, channel or scope is off. Adjust and rerun.
• PIVOT: far off, or the conversations reveal the wrong customer or the wrong promise. Revisit the ICP or the value proposition.

Step 2: Find which variable broke

"Pivot" is too vague to act on. Pivot what?

P/MF has four variables: ICP, value proposition, product, business model. They cascade. Change one, and everything after it has to change too:

• Change the ICP, and the value proposition, product and business model all move.
• Change the value proposition, and the product and business model move.
• Change the product, and the business model moves.

Most founders pivot one layer and don't see the blast radius. They break three things trying to fix one.

So diagnose before you decide. Sometimes the pivot is a new audience, not a new product. Viagra was the same formula, sold to a different market.

Step 3: Look inward, then outward

Inward: why do you want to pivot? Because the evidence says so, or because you're bored, scared, or chasing something shinier? Be honest. A pivot driven by emotion is just a restart.

Outward: go back to customers. Not to ask what you should build, but to understand what they're trying to get done and where your current offer falls short.

Step 4: Hold the paralysis, keep your insights

A pivot doesn't mean throwing everything away. You've learned things nobody else knows about this market. Keep them.

And don't pivot into consensus. If customers want it to taste like Coke, they should buy a Coke. Your edge is the insight others missed, not the idea everyone agrees with.

Step 5: Validate the new direction, then commit

Treat the new direction as a hypothesis. Set new thresholds, run the cheapest test that can prove you wrong, and decide.

Then commit. Give the decision a fixed period, review it as data comes in, and don't relitigate it every week.

A real case: RallyUp

RallyUp sold employee-generated LinkedIn content to B2B companies. Activation was low: employees didn't post.

A handful of customer interviews showed why. Employees had no intrinsic motivation to post, and often the wrong network. That's a big red flag, and no amount of nudging fixes it.

We laid out the options, experiment to rescue activation or pivot, and chose to pivot. The ICP and the promise stayed: B2B executive teams who want pipeline from LinkedIn. The product changed to building the CEO's own audience, and the business model followed, from SaaS subscription to agency retainer.

Small, precise pivot. RallyUp is now closing in on $1M ARR. Read the full case.

When to shut down

Almost never at the start. Here's how I think about it, stage by stage.

While you're finding the market: don't

A value proposition that doesn't click with an ICP doesn't mean you're wrong. It means you've learned something. So you pivot one thing. Maybe the ICP. Maybe the value proposition, because you still believe in the ICP. Maybe both, because a new ICP needs a new promise anyway.

And testing has never been cheaper. With $20 a day you can run campaigns that test different promises on different audiences. Cold DMs and outreach on LinkedIn and X cost you nothing but time. You'll get signals either way.

You'll also get a lot of no's. This doesn't work, and this doesn't work, and this doesn't work. It's frustrating. But every new attempt is something others most likely haven't tried yet. The more you try, the more likely you are to find the combination nobody else found.

At this stage the only real reason to stop is that you have nothing left to try with.

Once there's a product: only if the money runs out

Building costs more. You build it, host it, support it, and it starts to define your company.

But if you built it after validating the market, you already have an offer, an ICP, and people saying "yes, here's my money." That's traction. Traction makes it easier to raise, and it kills a lot of the self-doubt.

So at this stage, shut down only if you run out of resources and can't raise more. If the market is right and people are paying, there's no reason to stop. Sell more. Take a loan. Raise on a SAFE. Make it work.

When the business model won't work: pivot first

This is the other real reason to stop. You're creating value, but you can't keep enough of it. The product costs too much to deliver, acquisition costs too much, and the unit economics don't work at a price people will pay.

You can still pivot. Charge more. Sharpen the value proposition. Add what makes it worth more. Or change the ICP: from someone willing to pay $8 to someone willing to pay $80.

If you've tried, and you have no resources left, then yes, shut down. And don't sell your house to keep it alive.

The honest part

Most founders don't shut down because the market said no. They stop because they have no energy left. That's fair. This is hard, very hard, and most people never make it.

But you can almost always try a different approach. Luck favours the brave. Keep at it.

Common mistakes

• Deciding after the data. You'll always find a reason to keep going. Set the bar first.
• Pivoting everything at once. Change the variable that broke, then follow the cascade.
• Pivoting into consensus. The crowded idea is crowded for a reason.
• Waiting for certainty. You'll never have it. Small pivots now save you a huge one later.
• Quitting in market discovery. Testing an ICP and a promise costs almost nothing now. Run out of ideas before you run out of attempts.

What to do next

1. Write GO, ITERATE and PIVOT thresholds for your current experiment.
2. Name which of the four variables you think is broken, and what the evidence is.
3. Map what else moves if you change it.
4. Pick a decision date. Put it in the calendar.

Stuck on the fence? That's exactly the decision I help founders make. Tell me where you are.