How to Run a Pre-Mortem: A Launch Facilitation Script and Template
A pre-mortem assumes your launch already failed and asks why. Here is Gary Klein's method, Shreyas Doshi's Tigers framework, a word-for-word script and a template.
Share
A pre-mortem is a short meeting in which the team imagines a launch has already failed and writes down why. Here is how to run a pre-mortem in 60 minutes: brief the plan, have everyone silently list failure reasons as Tigers, Paper Tigers and Elephants, share them round the room, rank the top three to five threats, and give each one an owner, a mitigation and a deadline.
Key takeaways
Psychologist Gary Klein developed the pre-mortem about 30 years before his 2021 write-up and described it in Harvard Business Review in September 2007 (HBR, 2007).
Klein's basic version takes as little as 20 to 30 minutes, with two minutes of silent writing (Gary Klein, 2021).
In a 2010 study of 178 participants, the pre-mortem reduced overconfidence in a plan more than simple critiques or pros-and-cons lists (Veinott, Klein and Wiggins, as reported by Klein).
Shreyas Doshi's Stripe format runs one hour, one to three months before launch, with at least two Tigers per person (Shreyas Doshi).
The meeting is the cheap part. The value is the ranked, owned list of threats you track until launch day.
What is a pre-mortem?
A pre-mortem is a planning exercise that starts from failure. Instead of asking "is this plan good?", the facilitator says: it is three months after launch, and the launch was a disaster; write down why. Klein's HBR article describes team members assuming the project they are planning "has just failed" and working back to the causes.
Klein writes that he developed the method about 30 years before his 2021 post, to help his own company reduce problems on projects. The underlying idea is prospective hindsight: imagining an outcome as certain makes it easier to explain than imagining it as merely possible.
A pre-mortem is not a launch checklist. A feature launch checklist confirms you did the known tasks; a pre-mortem looks for threats nobody has written down.
Why does a pre-mortem catch risks planning meetings miss?
Klein's diagnosis is social, not analytical. Projects fail partly because people with reservations are reluctant to speak up during planning. In a normal review, raising a risk feels like not being a team player. In a pre-mortem, finding the failure is the assignment, so dissent becomes the cooperative move.
Three design choices do the work:
Silent, independent writing first. Everyone writes before anyone speaks, so the most senior or loudest person can't anchor the room.
Certainty of failure. "It failed" removes the debate about whether failure is likely and goes straight to how.
A shared vocabulary. Doshi argues that the Tigers, Paper Tigers and Elephants metaphors give people more psychological safety to raise concerns, in the meeting and afterwards in Slack, stand-ups and one-to-ones.
The evidence is modest but real. Klein reports that Veinott, Klein and Wiggins (2010) compared the pre-mortem with other critique methods among 178 college students reviewing a plan for a hypothetical H1N1 outbreak. The pre-mortem group showed the greatest reduction in overconfidence. That is a lab study with students, not a field trial with product teams, so treat it as support for the mechanism rather than proof of launch outcomes.
What are Tigers, Paper Tigers and Elephants?
Shreyas Doshi, who used the format as a product manager at Stripe and discussed it on Lenny's Podcast in 2022 (Lenny's Newsletter, 2022), sorts every threat into one of three types:
Type
Doshi's definition
Launch example
Tiger
A clear threat that will hurt us if we don't act
The billing migration double-charges annual customers
Paper Tiger
A threat you are personally not worried about, though others might be
"A competitor will copy this within a month"
Elephant
Something you worry the team is not talking about
Nobody owns support questions after launch day
Paper Tigers matter because one person's Paper Tiger is another person's Tiger. Discussing them out loud either reassures the worried person or reveals that the confident person is missing something.
Tigers then get an urgency label. The Product Compass write-up of the method uses three (Product Compass, 2024):
Launch-blocking: must be fixed before launch. Gets an owner and a date.
Fast-follow: fix in the next iteration after launch.
Track: monitor, and escalate if it grows.
How to run a pre-mortem, step by step
Before the meeting
Pick the timing. Doshi's rule of thumb is one to three months before launch: late enough that the plan is concrete, early enough to change it. Klein runs it at the kickoff, once the team knows the plan.
Invite across functions. Everyone who will build, sell, support, market or be on call for the launch. Leaders attend but speak last.
Send a one-page plan summary a day ahead. Scope, target date, success metric and the riskiest decisions. A technical spec engineers actually read works well as the pre-read for engineering-heavy launches.
Prepare the board. Three columns (Tigers, Paper Tigers, Elephants) on a whiteboard or a shared doc.
During the meeting
Kick off (10 minutes). Read the plan summary aloud, then deliver the failure prompt (script below).
Silent writing (10 minutes). Each person writes at least two Tigers, plus any Paper Tigers and Elephants. No talking.
Share round the room (15 minutes). Go round one item at a time until every item is on the board. No debate yet. Klein starts with the team leader, which signals that criticism is welcome; if your leader's view tends to anchor the room, start with a random person and put the leader last.
Discuss and vote (15 minutes). Discuss Paper Tigers first, then cluster duplicate Tigers. Everyone gets three votes for the Tigers and Elephants that worry them most.
Draft the action plan (10 minutes). Take the top three to five threats. For each: urgency label, owner, first mitigation step and a due date.
After the meeting
Send the action plan within 24 hours, with the verbatim threats, the themes, and each owner's mitigation. Doshi's version keeps the exact wording so people can see their concern was recorded.
Review the list weekly until launch. A threat without a status update is a threat nobody is handling.
Pre-mortem facilitation script (copy and adapt)
Use these words or close to them. Timings assume a one-hour slot with five to ten people.
[0:00] KICKOFF
"Thanks for coming. For the next hour we're going to try to kill
this launch before it kills us. Quick recap of the plan: [2-minute
summary: scope, date, success metric, riskiest decisions]."
[0:05] THE PROMPT
"Imagine it's [date, ~3 months after launch]. The launch failed.
Not a small miss: [success metric] is nowhere near target, and
we're writing the post-mortem. Your job is to tell me why it failed."
[0:10] SILENT WRITING (10 min)
"Write at least two Tigers: clear threats that will hurt us if we
do nothing. Add any Paper Tigers, things others may worry about
that you don't, and any Elephants, things you think we aren't
talking about. One per sticky or line. No talking yet."
[0:20] ROUND ROBIN (15 min)
"We'll go round, one item at a time, starting with [random person].
Just read it out; no discussing or defending yet. [Leader], you're last."
[0:35] PAPER TIGERS, THEN VOTE (15 min)
"Let's start with the Paper Tigers. Does anyone see one of these as
a real Tiger?" ... "Now everyone gets three votes for the Tigers
and Elephants that worry you most."
[0:50] ACTION PLAN (10 min)
"For our top [3-5]: is it launch-blocking, fast-follow, or track?
Who owns it? What's the first step, and by when?"
[0:58] CLOSE
"Thank you. Raising these now is the most useful thing anyone did
for this launch this week. Action plan in your inbox by tomorrow."
Learning how to run a pre-mortem remotely changes little: do the silent writing in a shared doc with comments turned off, and vote with emoji reactions. Doshi notes the vocabulary should carry on afterwards: "that sounds like an Elephant" is an easy way to raise a worry in a channel.
Pre-mortem template
Copy this table into your launch doc. Score likelihood and impact from 1 to 3, and multiply. A simple decision rule: a score of 6 or 9 is launch-blocking, 3 or 4 is fast-follow, and 1 or 2 is track. Klein also suggests rating how easy each problem is to prevent, which helps order the work.
#
Threat (verbatim)
Type
Likelihood (1–3)
Impact (1–3)
Score
Urgency
Owner
First mitigation
Due
Status
1
Tiger / Elephant
Launch-blocking / Fast-follow / Track
Open / Mitigated / Accepted
Worked example (illustrative)
Say an eight-person team is launching usage-based pricing in six weeks. Eight participants write 31 items in ten minutes: 19 Tigers, 7 Paper Tigers and 5 Elephants. They cluster into nine themes. The top four:
#
Threat
Type
L
I
Score
Urgency
Owner
First mitigation
1
Billing migration double-charges annual customers
Tiger
2
3
6
Launch-blocking
Billing engineer
Dry-run the migration on a copy of last month's invoices, two weeks before launch
2
Sales can't explain the new pricing on calls
Tiger
3
2
6
Launch-blocking
Product marketing
One-page pricing FAQ and a 30-minute sales session
3
Usage meter lags by hours, so customers dispute invoices
Tiger
2
2
4
Fast-follow
Tech lead
Show "usage updated at" time on the billing page
4
Nobody owns support macros after launch
Elephant
3
1
3
Fast-follow
Support lead
Draft five macros before launch week
The loudest Paper Tiger, "a competitor will copy us within a month", was discussed for two minutes and parked: the team agreed it would not change the launch either way. Total cost: one hour for eight people, eight person-hours, against a double-charging incident that would have hit every annual customer.
What should happen after a pre-mortem?
Knowing how to run a pre-mortem is the easy part; the follow-through is what prevents the failure. Three habits keep it honest:
Turn assumptions into tests. Many Tigers are untested assumptions ("customers will understand the new invoice"). Move them onto an assumption map and test the riskiest ones before launch. The guide to assumption mapping covers how.
Turn launch-blocking Tigers into tickets. Mitigations that need engineering work belong in the backlog with acceptance criteria, not in meeting notes. The PRD-to-user-stories guide shows how to write them.
Run the post-mortem against the pre-mortem. After launch, compare what happened with what the team predicted. Google's SRE book recommends blameless post-mortems that look for contributing causes without blaming individuals, and warns that "an unreviewed postmortem might as well never have existed" (Google SRE). The risks you foresaw and mitigated are wins. The ones that surprised you show where the team's imagination of failure has gaps.
Pre-mortem vs post-mortem vs risk register
Pre-mortem
Post-mortem
Risk register
When
Before launch (1–3 months out)
After an incident or launch
Throughout the project
Question
"It failed. Why?"
"It failed. What actually caused it?"
"What could go wrong, and how likely is it?"
Main output
Ranked, owned threats
Contributing causes and action items
A maintained list of risks
Main strength
Surfaces unspoken doubts
Learns from real events
Keeps risks visible over time
Main weakness
Only as good as the follow-up
Too late for this launch
Tends to list only the obvious risks
Used together, the pre-mortem fills the register with unspoken risks, and the post-mortem checks how good the pre-mortem was.
What are the common mistakes when running a pre-mortem?
Running it too late. A pre-mortem the week before launch can only produce worry. Doshi's one-to-three-month window leaves time to act.
Letting the leader speak first. If the VP names their top risk first, the room converges on it. Leaders go last.
Debating during the share-out. Defending the plan while people are still sharing teaches them to stop sharing.
Leaving with a list but no owners. A risk without an owner is just a worry. Every top threat leaves the room with a name and a date.
Treating it as a one-off. The list should be reviewed weekly until launch, then compared with what actually happened.
Using it for every change. A one-line bug fix does not need a pre-mortem. Save it for launches, migrations, pricing changes and anything hard to reverse.
Where does Specky fit?
Specky's Strategy Canvas has a Run Pre-mortem action. You describe the initiative, and AI imagines it failed a year from now, drawing on your Product Graph, then drafts up to 11 cards: three wrong assumptions, three risks, three failure modes and two mitigations, each with a severity. You choose which to keep as tracked cards, and assumptions can be tested with linked experiments. Use the draft as a pre-read or as one more voice in the room, not a substitute for the team's silent writing. The Product Graph post explains where the context comes from.
Frequently asked questions
How long should a pre-mortem take?
A pre-mortem takes 20 to 60 minutes. Gary Klein's basic version runs in as little as 20 to 30 minutes, with two minutes of silent writing per person. Shreyas Doshi's format uses a full hour: 10 minutes to kick off, 10 to brainstorm, 30 to vote and discuss, and 10 to draft the action plan. Use the longer format for launches with many teams involved.
When should you run a pre-mortem?
Run a pre-mortem one to three months before launch, which is Shreyas Doshi's rule of thumb, or at the project kickoff, as Gary Klein recommends. The plan needs to be concrete enough to critique, with enough time left to change it. If you only have time for one, run it once scope and the launch date are agreed. Block it in a quiet week; the example PM weekly schedule shows where.
Who should attend a pre-mortem?
Invite everyone who will build, sell, market, support or be on call for the launch, typically five to ten people. Cross-functional attendance matters because support and sales often see threats engineering cannot. Senior leaders should attend but speak last, so their concerns do not anchor the room. If more than about 12 people need a voice, split into groups and merge the lists.
What is the difference between a pre-mortem and a post-mortem?
A pre-mortem happens before launch and imagines failure to prevent it; a post-mortem happens after an incident or launch and analyses what actually went wrong. The pre-mortem's output is a ranked list of threats with owners. The post-mortem's output is contributing causes and action items. Comparing the two after each launch shows how well your team predicts failure.
What are Tigers, Paper Tigers and Elephants in a pre-mortem?
They are Shreyas Doshi's three categories of threat. A Tiger is a clear threat that will hurt the launch if nobody acts. A Paper Tiger is a threat you are personally not worried about, even if others are. An Elephant is something you think the team is avoiding talking about. The labels make it easier to raise uncomfortable concerns.
Can you run a pre-mortem remotely?
Yes. Run the silent writing in a shared document or whiteboard with comments hidden, so nobody anchors on others' answers. Share items round-robin over video, then vote with reactions or dot-voting tools. Send the action plan within 24 hours, exactly as you would in person.
GDPR-aligned. EU workspace data is stored in Frankfurt, Germany. GDPR and EU data Built in Vienna, Austria. A product of influence solutions fm GmbH (FN 675374y). Imprint 14-day trial. Card required, and nothing is charged until day 14. 30-day refund after subscribing. See pricing