The Pivot: Change One Variable, Keep the Asset
Starting over is not a pivot. A pivot keeps everything you learned and moves exactly one thing.
Founders use the word to mean two opposite things. One is a disciplined change of a single business model variable based on evidence, with the team, technology and customer relationships carried forward. The other is abandoning eighteen months of work for an idea somebody had at the weekend.
The first is how Slack came out of a failing game company, keeping the internal chat tool the team had built for themselves. It is how Instagram came out of Burbn, keeping the one feature people used and deleting the rest. It is how Shopify came out of an online snowboard shop whose founders had to build their own storefront. Each kept an asset and changed the business around it.
Keep the asset, change the frame. If you are throwing away everything, you are not pivoting, you are restarting with less money.
Startup Genome's research found companies that pivoted once or twice raised significantly more capital, grew users substantially faster, and were far less likely to scale prematurely than those that never pivoted or pivoted repeatedly. That shape matters: one or two deliberate changes are healthy, continuous changes are a team that has not learned anything from any attempt.
Signals that justify a pivot
Not "it feels slow," which describes every startup. These are specific and checkable.
The retention curve never flattens
Cohort after cohort declines toward zero with no stable base. You have a leak nobody plugs, and it usually means the problem is real but not painful enough, or the people you sold to are not the people who feel it. This is the strongest signal on the list.
Sales cycles lengthen as you improve
You are getting better at selling and deals are taking longer. That is a market telling you the problem is not urgent, or that you are talking to someone without the authority to solve it. Additional selling skill does not fix either condition.
Customers use it for something else
The most valuable signal available, and the easiest to dismiss. When people repeatedly bend your product toward a different job, they are showing you a working business you have not launched yet. Both Slack and Instagram came from taking that seriously.
One tiny segment is thriving
Overall numbers are mediocre, but eleven customers of one type retain, expand and refer. That is not noise. That is a business hiding inside a broader failure, and the pivot is to become their company entirely.
The economics cannot work at any scale
Acquisition costs more than the lifetime value and no plausible improvement closes the gap. Run the arithmetic from Day 7 before deciding. If a 40 percent improvement in every input still leaves you underwater, the model, not the execution, is wrong.
0 to 1. No pivot case. Your problem is execution, and changing direction now would destroy the learning you are close to converting. 2 to 3. Worth investigating. Run twenty customer conversations before touching the roadmap. 4 to 5. A real case. Name the one variable and design the six-week test below. 6 or more. The current path is not working and you know it. The risk now is delay, not change.
Eight checkable signals. The last one is decisive: a team that cannot name the variable it wants to change is not ready to pivot, whatever the other seven say. Pivoting without a hypothesis burns runway and produces no learning.
The variables you can change
Eric Ries catalogued around ten pivot types in The Lean Startup. Four cover most real situations.
Customer segment pivot: same product, different buyer
You keep the product and change who you sell it to. The cheapest pivot available and the most frequently overlooked, because founders are attached to the audience they imagined.
- Keeps: the product, the technology, most of the team's knowledge.
- Changes: messaging, pricing, channel, and usually the sales motion.
- Test: ten conversations with the new segment before you change a line of code.
Customer need pivot: same buyer, different problem
You keep the customer relationship and solve a different problem for them, usually one they mentioned repeatedly while you were selling something else.
- Keeps: customers, credibility, distribution, domain knowledge.
- Changes: the product, sometimes substantially.
- Test: sell the new thing to five existing customers before building it.
Zoom-in pivot: one feature becomes the whole product
A single part of your product carries all the usage. You delete the rest and build the company around it. Instagram from Burbn is the textbook case.
- Keeps: the feature, its users, and the evidence that it works.
- Changes: scope, positioning, pricing, and everything you tell yourself about the roadmap.
- Test: ship the feature standalone to new users who never saw the old product.
Business model pivot: same product, different economics
Services to product, one-off to subscription, direct to channel, high touch to self-serve. The product barely changes; how you capture value changes entirely.
- Keeps: product, customers, brand.
- Changes: pricing, margins, cash cycle, and the shape of the team you need.
- Test: run the new model with three new customers while the old one continues.
Each type names what is preserved and what moves. Choosing the type explicitly is what stops a pivot from becoming a restart, because it forces you to write down what you are keeping. All four panels are readable with scripting off.
Run it in six weeks, not six months
A pivot is a hypothesis test with a deadline. Without the deadline it becomes a mood that hangs over the company for two quarters while nobody commits to anything.
Weeks 1–2: evidence
Twenty conversations in the new direction. No pitching. Their process, their spend, their last attempt to fix it.
Weeks 3–4: sell it
Offer it before it exists. A price, a start date, a deposit. Signatures are the test, not enthusiasm.
Weeks 5–6: deliver once
Deliver manually to the first buyer. Measure whether the outcome is real and what it cost you.
Decide
Pre-agree the bar in week zero: how many signed, at what price. Meet it and commit fully. Miss it and stop.
Six weeks, one variable, a written success bar set before you start. Setting the bar in advance is what stops a failed test from becoming three more months of hope, which is the most common way a pivot kills a company that had time to survive.
Run the test alongside the existing business rather than instead of it. Existing revenue funds the experiment, and killing the old line before the new one is proven turns a six-week test into a bet-the-company decision. If you cannot spare the capacity to do both, you are already too late for a considered pivot, which is a different situation covered on Day 10.
Before you decide anything, list what you actually own after all this work. Technology that took months to build. Customer relationships and the trust attached. Domain knowledge nobody outside your team has. A brand with some recognition in one niche. Data. Integrations. A distribution channel. Regulatory approval. Now write next to each one whether the pivot you are considering keeps it or destroys it. A pivot that preserves five of eight assets is a strong move. One that preserves one is a new company, and it deserves to be evaluated as a new company: from zero, on its own merits, against every other thing you could do with the next two years.
Telling everyone else
The communication decides how much of the team survives the change, and founders usually get it wrong by being vague to seem confident.
Your team. Tell them the evidence, not just the decision. "Cohorts one through nine never flattened; here is the chart" earns commitment. "We are going in a new direction" earns résumé updates. Name what stays the same, because people can absorb a large change when the boundaries are clear.
Existing customers. If they still get what they paid for, tell them what improves. If the pivot ends their product, say so with a date, a migration path or a refund, and do it early. The founders who handle this well keep the references. The ones who go quiet lose the industry's goodwill in a week, and industries are small.
Investors. They have seen more pivots than you have. Bring the evidence, the hypothesis, the test design and the success bar. Investors do not lose confidence over a change of direction. They lose it over a change of direction that arrives with no data and a new deck.
The pattern in Startup Genome's data: one or two deliberate pivots outperformed both never changing and changing constantly. A company that pivots every quarter is not adaptable, it is avoiding the part where you commit long enough to find out whether something works.
What a pivot cannot fix
Founders sometimes reach for a pivot because the current business is hard, and hard is not the same as wrong. Three problems follow you into the new direction unchanged.
A team that does not ship will not ship faster on a different idea. Cofounders who cannot make decisions together will deadlock on the new plan by month two. And a founder who has stopped talking to customers will misread the new market the same way they misread the last one.
Pivot when the evidence says the model is wrong. When the evidence says the execution is wrong, changing direction just gives the same execution a fresh surface to fail on, and costs you the eighteen months of learning that were about to become useful.
The takeaway
- A pivot changes one variable and keeps the rest. Changing everything is a restart with less money.
- Justify it with specific signals: flat-lining retention, lengthening sales cycles, off-label usage, one thriving segment, or economics that never close.
- Name the pivot type so you know what you are keeping.
- Run a six-week test with a success bar written before you start, alongside the existing business.
- Communicate with evidence. People commit to data, not to a new direction announced with confidence.
- One or two pivots correlate with the best outcomes. Continuous pivoting is avoidance.
Frequently asked questions
What is a startup pivot?
A structured change to one core element of the business model while keeping what you have learned and built. Eric Ries catalogued around ten types, including zoom-in, segment, need, platform and channel pivots. One variable changes; the rest is retained.
When should you pivot?
When evidence says the current path will not work and you can name the single variable you believe is wrong. Signals: retention that never flattens, sales cycles lengthening as you improve, customers using the product for another job, one small segment thriving. Without a named hypothesis you are thrashing.
Do pivots actually work?
Startup Genome found companies pivoting once or twice raised more, grew users faster and scaled prematurely far less often than those pivoting never or more than twice. One or two deliberate changes are healthy; continuous change means nothing is being learned.
What are famous examples of successful pivots?
Slack came from internal tooling inside a failing game company. Instagram was extracted from Burbn by keeping the photo feature people used. Shopify came from an online snowboard shop whose founders built their own storefront. Each kept an asset and changed the business around it.
Some companies pivot. Some just keep moving until the money is gone.
Kill My Startup separates the change of direction that saves a company from the one that ends it faster.
Buy on Amazon →Sources
- Eric Ries, The Lean Startup, on pivot taxonomy and the pivot or persevere decision.
- Startup Genome Report Extra, on pivot frequency, capital raised, user growth and premature scaling.
- Public accounts of the Slack, Instagram and Shopify origin stories.
- Rob Fitzpatrick, The Mom Test, on gathering evidence before committing to a direction.