Before hope rewrites the bet
Startup Pivot Criteria — When to Change the Bet Instead of Killing the Idea
Stop is one contract. Continue is another. The messy middle is Pivot: the pain is real, the current shape is wrong, and you still have not named which part of the bet to rewrite. Pivot criteria are that page — tripwires you write before the week so “change the segment / problem / solution / channel and re-test” is executable, not a Sunday mood.
Last updated: September 29, 2026
Direct answer
What are startup pivot criteria?
A pivot criterion is a pre-written tripwire with the same four fields as a kill criterion — an observable signal, a threshold, a calendar window, and a named action — except the action is “change this dimension of the bet and re-test,” not “stop.” Eric Ries, in The Lean Startup, treats a pivot as a structured change of strategy while keeping the vision: you do not throw the company away; you rewrite a fundamental hypothesis about the customer, the problem, the solution, or the channel and run a new test. You write the tripwire before outreach so Sunday cannot stretch mixed evidence into “hope harder.” Scores do not decide pivots. A failed current-shape threshold plus adjacent, repeated pain does.
Key takeaways
What to remember
- Same four fields as kill criteria. Different action: change one named dimension of the bet, then re-test.
- Write the tripwire before recruitment. A pivot invented after compliments is a recap, not a criterion.
- Pivot is one rewrite — segment, problem, solution, or channel — not a feature bolted onto the same sentence.
- Pain without a working offer is a pivot. No pain, and no adjacent segment or problem showing the same pain, is a kill.
- After a pivot criterion fires, the cheapest next move is a field test on the rewritten sentence — not a new factory.
Why it matters
“We should pivot” is not a decision
I have watched founders treat “we should pivot” as a feeling that arrives after a disappointing week. Feelings do not name which part of the bet to change. They rename the same product and call it learning. CB Insights’ 2026 analysis of 431 VC-backed companies that shut down since 2023 is useful here as a warning about symptoms versus roots — not as a personal forecast. “Ran out of capital” leads their list at 70 percent, and they are explicit that this is often the final cause of death. Among the 385 companies with identifiable reasons, more telling patterns include poor product-market fit (43 percent), bad timing (29 percent), and unsustainable unit economics (19 percent). Many shutdowns cited more than one reason, so those shares can add to more than 100 percent. Poor fit is often a wrong bet shape — the wrong person, the wrong job, the wrong offer — not proof that no problem exists. Ries’s public methodology page is blunt about the fork: measure whether you are moving the drivers of the business; if not, it is time to pivot or make a structural course correction to test a new fundamental hypothesis about the product, strategy, and engine of growth. That is a method. It is not a success rate, and this page does not invent one.
Vocabulary
Four terms this page uses on purpose
These match how the rest of Yibud talks. If a sentence here uses one of these words, it means this.
Pivot criterion
A pre-written tripwire: signal + threshold + window + action. The action names which dimension of the bet you will change and what you will re-test. A risk list without a dated rewrite is not a criterion.
The bet
The current sentence of person + job + offer + channel. A pivot rewrites one of those. Adding a feature to the same sentence is not a pivot. Renaming the landing page is not a pivot.
Adjacent evidence
Repeated, unprompted pain in a named neighboring segment or problem that you wrote down before the window opened. “Someone else might want this” is not adjacent evidence. “Segment B, named in advance, described the weekly incident without being asked” is.
Window and action
The window is calendar dates, not “until we know.” The action is the rewrite you will make when the window closes: change the person, the job, the offer, or the channel, then write a new tripwire. If the action is “keep building and see,” you have written Continue twice.
What you are allowed to change
Four dimensions of the bet (Ries as framing, not as a scoreboard)
Ries names common pivot types as methodology — customer segment, customer need (problem), zoom-in / solution, channel, and others. This page uses four of them as the only rewrites a tripwire may name. It does not claim any type “works more often.” Fill the brackets with your own counts. Do not treat the examples as industry benchmarks.
Segment
The named person is wrong; the job may still be real. Template: if fewer than N of M interviews in segment A show unprompted weekly pain by day D, and segment B — named before outreach — does, rewrite the person the same day.
Problem
The person is real; the job you named is not the one they already spend time on. Template: if they describe a different weekly incident than the one on your page, rewrite the job to that incident and drop the original sentence.
Solution / offer
Pain is real; the current offer or price fails. Template: if a waitlist converts and a priced checkout does not, inside the same window, rewrite the offer or the price assumption — not the marketing copy — and re-test.
Channel
The person and the job are real; you cannot reach them where you planned. Template: if you cannot find M qualified people in the named channel by day D, and a second channel you named in advance can, rewrite the channel. Silence in one inbox is evidence about reach, not about the problem.
Not the same page
Pivot criteria vs kill, pre-mortem, assumption, and the week-after decision
Five Yibud pages sit next to each other. Mixing them up turns a rewrite into a slogan, or a Stop into another sprint.
Kill criteria — the Stop contract
Kill criteria are the pre-written Stop: what evidence, by which date, means you halt the current one-liner. This page is the Pivot contract: what mixed pattern means you change one dimension of the bet and re-test, instead of stopping or hoping harder. Write both. If every mixed outcome still says “keep the same product,” you have not written Pivot.
Startup kill criteria (write Stop before the week) →Pre-mortem — invent the failure history
A pre-mortem treats the one-liner as already dead and ranks the ways it died. That hour produces failure modes. It does not yet say which mixed pattern, by which date, means you rewrite the person rather than stop. Pivot criteria are tripwires you attach after the ranking.
Startup idea pre-mortem (30–45 minutes) →Critical assumption — kill one claim
That page names the single claim that would make the rest of the plan irrelevant if false, then runs a cheap week against Continue / Refine / Stop. This page is the Pivot branch of those lines: which dimension of the bet to change when Refine is the honest word, and how you would have known before anyone was polite.
Critical assumption first — name, falsify, decide →Validation decision — after a week of evidence
That page is the call once interviews, a smoke test, or a payment log already exist. This page is written before that emotional investment, so “pivot” on Sunday is a tripwire you already named — not a new story.
When to continue, pivot, or kill after evidence →A mid-band overall score is a flashlight, not a pivot switch. It can point at a weak dimension. It does not name the rewrite. What a validation score actually means →
The sitting
How to write pivot criteria in one sitting
Do this after a pre-mortem ranking, or after a report names a weak dimension — not after you have already spent the week explaining the vision. Set a timer. Leave the build closed. If you have a co-founder, each of you writes a draft alone, then you keep the stricter rewrite.
- 1
Write the current bet as one sentence
Person + job + offer + channel. “A $39/month inbox that drafts Sunday-night parent emails for elementary teachers, reached through a Facebook group.” If a field still says “users” or “content,” you do not have a bet yet. Tighten until a no is possible on each field.
- 2
Name the dimension that would change if the week is mixed
Pick one in advance: segment, problem, solution, or channel. If you cannot name it until Sunday, you will change whichever field feels least embarrassing. Write the adjacent alternative now — segment B, a different job, a different price, a second channel.
- 3
Write the signal that means “pain is real, this shape is wrong”
Two observables, not one vibe. Example pair: unprompted weekly incident in the interview, and failure of the current offer or current segment to hit its threshold. Polite interest is not a signal. Rob Fitzpatrick’s Mom Test is the filter: ask about last week’s life, not whether they like the idea.
- 4
Write two thresholds a stranger would accept
Current-shape threshold: “fewer than N of M in segment A do the action.” Adjacent threshold: “N of M in segment B, named before outreach, do.” Or: “waitlist yes, priced checkout no.” Fill N and M yourself. These numbers are yours. They are not a benchmark copied from this page.
- 5
Date the window and write the rewrite
Start date and end date. Fourteen days from first outreach is a window. “Give it the summer” is not. The rewrite is one sentence: change [dimension] to [named alternative], write a new tripwire the same day. If the rewrite still contains the old person and the old job, you have not pivoted.
- 6
Date the page, then leave the building on the new sentence
Same sitting. Sign it. If you later want a kinder rewrite, that is a new page with a reason — not an edit after the first compliment. When the criterion fires, Steve Blank’s Customer Development is the posture: the cheapest next test is a field hypothesis you leave the building to check. It is not a reason to open a new repo.
What you leave with
The Pivot Contract (copy this)
If the sitting produced a mood, you did not finish. The page should have one filled tripwire, one named dimension, and a rewrite you would actually run. Fill the brackets. Leave nothing as “users” or “soon.”
The tripwire
Current bet: [person] will [action] via [offer] in [channel]. Signal: [observable pain] AND [current-shape miss]. Threshold: [N of M current] / [N of M adjacent]. Window: [start]–[end]. Dimension to change: [segment / problem / solution / channel].
If any bracket still says “people,” “traction,” or “later,” the contract is not written. The AND matters: pain without a miss is Continue. A miss without pain is closer to Kill.
The rewrite and the re-test
Pivot if [named mixed pattern]: rewrite [one dimension] to [named alternative] the same day. New tripwire: [signal + threshold + window]. Next cheapest test: [conversation / one-page / manual job] on the new sentence — not a new factory.
One dimension. If you change person and offer and channel in the same sitting, you have started a different company without a contract.
Example tripwires (templates — fill your own N)
Segment: if fewer than N of 12 interviews in A show unprompted weekly pain by day 14, but B (named now) does → pivot person. Offer: if waitlist converts and priced checkout does not, same window → pivot offer or price. Channel: if you cannot find M qualified people in channel 1 by day 14, and channel 2 (named now) can → pivot channel.
The counts are blanks for you to fill. They are not conversion rates, not Yibud results, and not a claim about what “usually” happens.
The fork
Kill, Pivot, or Continue — written before the evidence
The contract does not replace the field week. It decides what mixed evidence is allowed to mean. The kill-criteria page is the Stop contract. The critical-assumption page is how you run the week. The validation-decision page is how you read a pile of evidence you already have. This page is why “pivot” cannot be invented after the compliments.
Kill
The critical assumption fails and no adjacent segment or problem — named in advance — shows repeated pain. Or payment / unit-economics evidence fails and there is no salvageable buyer. Stop the current one-liner. A different sentence gets its own contract. Do not “kill” by quietly renaming the same product.
Pivot
Pain is real in interviews or behavior. The current offer, segment, or channel misses its threshold. You already named which dimension changes. Rewrite that one field the same day. Write a new tripwire. Adding a feature is not a pivot.
Continue
The current-shape threshold was met on a signal a stranger could log. The next move is the next cheapest test, not a feature backlog. Continue is not “we are right.” It is “this sentence survived this window.”
When Pivot beats Kill
People describe the weekly incident without being prompted. Files exist. Time is already spent. The current offer or the current person fails the threshold you wrote. Adjacent evidence you named before outreach shows the same pain in a different shape. That is a wrong bet, not “no problem.” Refine / Re-test language on a Yibud report sits near this fork when the weak dimension is segment, problem, or solution — not when the honest word is Stop.
When Kill beats Pivot
Qualified people do not have the incident. Adjacent segments you actually named do not have it either. Or they will not pay any salvageable price, and there is no other buyer in the notes. Pivoting then is how a founder spends another quarter on a sentence that already failed. Write Stop. Keep the notes.
Worked example
One pivot contract you can copy (illustrative)
The week below is illustrative — names, prices, and counts are invented so you can see the setup. It is not a conversion benchmark and not a Yibud report. Copy the method, not the numbers.
The bet, written before outreach
“A $39/month inbox that drafts Sunday-night parent emails for elementary teachers, reached through one Facebook group.” The tempting next step is a template library. The current bet assumes the teacher is the person, the Sunday pile is the job, email drafts are the offer, and the group is the channel.
Two tripwires, written the same sitting
Segment tripwire: if fewer than 3 of 12 qualified elementary teachers describe an unprompted weekly parent-email pile by day 14, and 3 or more school-office coordinators — named as segment B before the first message — do, rewrite the person to the coordinator. Offer tripwire: if 5 or more join a waitlist and fewer than 2 complete a $39 checkout in the same 14 days, rewrite the offer toward a one-time $15 Sunday job or a live 20-minute call. Qualified means: they did the job last Sunday. Compliments do not count.
What Pivot looks like on day 14
Illustrative close: two teachers compliment the idea; none describe last Sunday’s pile. Four coordinators describe a Friday newsletter they already send and want the draft done for them. Waitlist: six emails. Checkout: zero. That is a segment pivot plus an offer pivot you already named — rewrite toward coordinators and a one-time job, write a new contract, and test by hand. It is not a reason to start the template library “because some people were warm.”
What must not happen in the window
Do not invent segment B on day 13 because A was quiet. Do not count “I would use that” as a weekly pile. Do not extend day 14 because the twelfth teacher has not replied — silence is a no on reach, and reach may be the dimension to change if you named a second channel in advance. If you cannot find twelve qualified people in fourteen days and you named no second channel, that can be Kill, or a channel tripwire you failed to write.
Write both tripwires before the first DM. If day 14 arrives and the pain showed up in B, not A, honor the segment line. If pain never showed up in anyone you named, honor Kill. The method is the dated page, not the story you tell after.
Where Yibud fits
Use the free validator as a flashlight, then write the pivot contract yourself
Yibud is a free, no-signup startup idea validator. Scores come from a deterministic rule engine, not from a language model guessing success. Optional AI text, when it is used, only polishes prose. It does not invent the number. A Startup MRI report can name a weak dimension and a critical assumption, then a Day-7 Continue / Refine / Re-test / Stop signal. Pivot sits next to Refine and Re-test when the weak dimension is segment, problem, or solution — rewrite the claim, do not hope harder. Stop is still Stop when there is no adjacent pain. If you already have a report, start the contract with the named assumption and the weak dimension — not with the overall score. Then run the week yourself. The method stands alone if you never open the analyzer.
Analyze my idea →Common mistakes
What usually wastes the pivot contract
Calling a feature a pivot
A new button on the same sentence is Continue in costume. If the person, the job, the offer, and the channel are unchanged, you did not pivot. You shipped.
Pivoting on polite interest
“Sounds useful” cannot fire a tripwire. Fitzpatrick’s Mom Test is the filter: last week’s incident, last week’s workaround, last week’s spend. If the only evidence is warmth, you do not have adjacent pain. You have manners.
Inventing segment B after A went quiet
Adjacent evidence is named before the window. A new audience discovered on day 13 is a new bet, not a pivot you planned. Write B now or you will always find a friendlier room.
Using the score as a pivot switch
A low distribution score does not pivot you into a new channel. A low monetization score does not pick a price. The score ranks relative risk in the inputs you typed. A dated threshold on a named dimension is what rewrites the sentence.
Pivoting when Kill is the honest word
No weekly incident. No adjacent segment you actually named. No salvageable buyer. Another rewrite is how the bar moves. Write Stop. Keep the notes. A different sentence gets its own sitting.
Sources
Where these ideas come from
- CB Insights, “The top 9 reasons startups fail” (report analyzing 431 VC-backed shutdowns since 2023; updated March 5, 2026) — Used as a warning about symptoms versus roots, not as a personal success forecast. CB Insights reports “ran out of capital” at 70 percent and calls it often the final cause of death. Among the 385 companies with identifiable reasons, more telling patterns include poor product-market fit (43 percent), bad timing (29 percent), and unsustainable unit economics (19 percent). Many companies cited more than one reason. This page does not invent other percentages, and it does not claim a pivot-success rate.
- Eric Ries, The Lean Startup — public methodology at theleanstartup.com/principles — Used for the definition of a pivot: a structured course correction to test a new fundamental hypothesis about the product, strategy, and engine of growth, while keeping the vision. Common types (customer segment, problem / customer need, zoom-in / solution, channel) are named only as methodology framing. This page does not invent success rates for any type.
- Steve Blank, Customer Development — Used only after a pivot criterion fires: the cheapest next test is a field hypothesis you leave the building to check. Blank is not the source of the tripwire fields, and this page does not treat Customer Development as a sample-size rule.
- Gary Klein, “Performing a Project Premortem” (Harvard Business Review, September 2007) — Cited for contrast. A pre-mortem invents a failure history and ranks modes. Pivot criteria are tripwires you evaluate later against evidence. Klein does not claim the exercise predicts startup outcomes.
- Rob Fitzpatrick, The Mom Test (2013) — Used as a filter on evidence: polite interest is not a pivot signal. Ask about last week’s life, not whether they would use the idea. This page does not treat the book as a conversion benchmark.
In one paragraph
Summary you can quote
Startup pivot criteria are a pre-written contract for changing the bet: an observable signal, a threshold, a calendar window, and a named rewrite of segment, problem, solution, or channel, then a re-test. Kill criteria are the Stop contract. A pre-mortem invents the failure history. A critical-assumption week tests one claim. The validation-decision page reads evidence you already have. This page makes Pivot executable so mixed evidence cannot be stretched into “keep building the same thing.” Pivot beats Kill when pain is real and the current shape fails a threshold you wrote. Kill beats Pivot when the critical assumption fails and no adjacent segment or problem — named in advance — shows repeated pain, or when payment evidence fails with no salvageable buyer. Scores do not decide pivots. A Yibud report can point at a weak dimension and a Refine / Re-test / Stop line. The tripwire still has to be written by the founder.
FAQ
Questions founders actually ask
How is a pivot criterion different from a kill criterion?
Same four fields: signal, threshold, window, action. Kill’s action is Stop. Pivot’s action is “change this named dimension of the bet and re-test.” Write both. If every mixed week still says “keep the same product,” you wrote Continue twice.
Can a Yibud score tell me to pivot?
No. The score is a structured read of the inputs you typed. It does not interview anyone or collect a payment. A failed current-shape threshold plus adjacent pain you named before the window is what rewrites the sentence. A report’s Refine / Re-test line is a flashlight. The contract is still yours.
What if every week says Pivot?
Then the sentence is still too wide, or you are changing the bar instead of one dimension. Two pivots in a row should narrow the person or the job, not invent a new factory. A third pivot with no narrower bet is a Kill you are avoiding.
How do I know I am not just avoiding Stop?
Look at adjacent evidence you wrote before outreach. If no named neighbor showed repeated pain, and no salvageable buyer appears in the log, Pivot is a story. Honor Kill. Keep the notes. A different one-liner gets a new sitting.
Do I need an account, and where does Yibud fit?
You do not need an account. The method on this page stands alone. Run the free analyzer if you want a flashlight on weak dimensions before you write the contract. The engine is deterministic. Optional language-model text only polishes prose. Bring the named assumption and the weak dimension back here and fill the brackets.
Write the Pivot contract before the next sprint
Yibud scores weak dimensions with a deterministic engine. You write signal, threshold, window, and which part of the bet to change while you are still honest — then Pivot is a tripwire, not a Sunday story.
Generate a Yibud report →