PREMORTEM

Assume a future plan failed, work backward to plausible causes, warning signs, and preventive actions.

A Power Mode is an instruction pattern, not a feature of the AI. Nothing recognises the name — the wording underneath it does the work, and you can copy it from this page.

From the book. Chapter 12 / Part VII — PREMORTEM

What it actually does

PREMORTEM changes the starting position. Rather than asking what could go wrong, it fixes a failure that has already happened and asks what most plausibly caused it.

Assuming the failure removes the hedging. "What are the risks?" invites a balanced list; "it is February and the migration missed by eight weeks — what happened?" invites an explanation, and explanations name causes, warning signs, and the point at which the outcome was already decided.

A premortem does not predict the future. It generates plausible failure explanations to examine, which is why the instruction ties them to the project’s actual constraints and known risks rather than letting the model invent scenarios.

When to use it

Use PREMORTEM when you fix a point in the future, declare it a failure, and ask what caused it.

It is the primary mode for this job: pressure-test something before you commit to it.

When not to use it

This is the section worth reading. Using only what the work requires is the point; a mode applied to the wrong shape of problem costs you a worse answer than plain language would have given.

Use a different mode instead

REDTEAM fires when you want a plan or argument attacked but cannot yet say where it is weak.
BACKCAST fires when you name a concrete success you want by a date, and ask what has to be true between now and then.
ASSUMPTION CHECK fires when you already suspect the weakness is something the work takes for granted, and want those beliefs ranked by how much each would change the outcome.
COUNTEREVIDENCE fires when you have a conclusion you believe and want the credible case against it, not a critique of how it is argued.
BUYER OBJECTION LENS fires when you need someone specific to agree, and want their reasons for refusing rather than a critique of what you are showing them.

Use no mode at all

The suspected problem is already a specific weak assumption.
There is no plan or future outcome to examine.

There is no bonus for naming a mode when plain language is clearer. If you can already state the exact question, ask it.

Its failure mode

Treating hypothetical failure as prediction or inventing unsupported failure stories.

Starter instruction

Paste this above your task. It is the whole mode — there is nothing else to switch on.

Assume this plan failed in the future. Work backward and identify the most plausible causes, the earliest warning signs, and one preventive action for each. Use the actual constraints, dependencies, history, and known risks; avoid unsupported speculative causes.

The same task, run both ways

Blind run. Separate sessions per arm, Temporary Chat, no memory, no tools attached. Neither prompt named a mode. No arm searched.

The task: We are launching a paid online course in eleven weeks. - Six modules: recorded video plus a workbook for each - Audience: our 4,200-person email list, 38% open rate - Price: $349, target 80 seats - Two people building it, both part-time on it alongside client work - Nothing is recorded yet; the outline is done and approved - One partner has agreed to promote it to their list in week 9 Review this plan.

Baseline — "Review this plan"

I would put three hard checkpoints into the project. By the end of week 2, Module 1 should be completely finished—not merely recorded—so you know your true production time. By the end of week 5, most recording should be complete. By the beginning of week 9, the entire buying and delivery experience should be tested. If you're materially behind at one of those points, cut production complexity rather than borrowing time from the launch.

[…]

Overall, I would **keep the eleven-week launch date, freeze the course scope now, aim to have recording finished by week 6, and turn weeks 7–11 increasingly toward marketing and sales rather than production**. The audience is large enough that 80 seats is not inherently unreasonable; the bigger threat is arriving at week 9 with an unfinished course and an underprepared launch.

PREMORTEM — "Assume this plan failed in the future"

A useful pre-mortem is to assume that eleven weeks from now the launch missed the 80-seat target, the course shipped late, or both.

Plausible failure cause	Why it follows from the actual plan	Earliest warning sign	One preventive action
1. Production falls behind because six modules are too much for two part-time people.	Nothing is recorded yet, and each module requires both video and a workbook. Client work competes for the same time.	By the end of week 2, the first module is not fully recorded, edited, and paired with a usable workbook.	Treat module 1 as a production benchmark: finish it end-to-end in the first two weeks, measure the hours required, then immediately reduce scope if that pace does not fit the remaining capacity.
2. Client work repeatedly displaces course work.	Both builders are explicitly working on this alongside paying client commitments, so course work has no natural protection.	Course sessions get moved or cancelled twice in the first 2–3 weeks, or planned weekly deliverables roll into the following week.	Block fixed, non-movable course production periods for each partner through week 8 and agree that new client work cannot consume those blocks.
3. The team discovers too late that the list is engaged but not ready to buy a $349 course.	A 38% open rate measures email attention, not willingness to purchase this offer at this price. No prior course conversion or purchase history was provided.	Early launch-content emails get normal opens but weak clicks, replies, waitlist signups, or other purchase-intent signals.	Run a concrete demand test well before recording all six modules—for example, a waitlist or early-access offer with the course promise, price, curriculum, and launch date.
4. Too much of the selling is left until the course is nearly finished.	Production can easily consume the first 8–9 weeks, while the only stated external promotion is in week 9. That leaves little time to correct weak positioning.	By week 4–5, the audience has heard little about the problem, outcome, course, or why it is being created.	Separate the launch workstream from production now: begin audience education and demand-building while the course is being produced rather than waiting for completion.
5. Workbook creation becomes a hidden second product.	Six workbooks can add substantial writing, design, review, and formatting time beyond recording the videos.	A workbook is taking nearly as long to create as its corresponding video module, or design/polish becomes a bottleneck.	Define a fixed workbook template and maximum scope now—e.g. summary, exercises, templates, and action plan—and reuse it for all six modules.
6. Recording/editing standards are too ambitious for the schedule.	With 11 weeks and part-time capacity, reshoots and heavy editing can consume the production buffer quickly.	The first videos require multiple takes, extensive editing, custom graphics, or lengthy review cycles.	Set a minimum viable production standard before filming: acceptable audio/video quality, limited retakes, and a defined editing ceiling.
7. The partner promotion underperforms or arrives later than expected.	There is only one stated partner promotion, and it occurs in week 9, when there is little time to replace that demand source.	By week 6–7, there is no agreed send date, copy, offer, tracking link, or audience estimate from the partner.	Lock the promotion details now and treat partner sales as upside rather than sales required to reach 80 seats.
8. The sales target is discovered to be unrealistic only during launch week.	80 sales × $349 = $27,920. The plan gives a seat target but no intermediate funnel targets against which to judge whether the launch is on pace.	There is no quantified target for waitlist size, sales-page traffic, clicks, or early purchases by the middle of the launch.	Build a simple launch funnel backward from 80 purchases and track leading indicators weekly; revise tactics or scope as soon as the funnel materially misses plan.
9. Course completion and launch readiness become the same deadline.	If all six modules must be finished before selling, any production slip directly threatens the launch date.	Weeks 7–8 arrive with several modules still needing recording, editing, or workbook completion.	Create an internal completion deadline at least 2 weeks before launch, leaving weeks 9–11 for QA, sales activity, platform setup, and contingency.
10. The team gets the content finished but misses operational details needed to sell and deliver it.	The plan mentions content and promotion, but not checkout, hosting, onboarding, email sequences, support, testing, or fulfillment.	By week 7 there is still no end-to-end test purchase and no clear owner for launch operations.	Assign one launch-readiness checklist with owners and complete a test purchase → access → welcome-email → course-delivery flow before the promotional push.

[…]

**1. End of week 2: Can you actually produce at the required pace?**
One completed module is more informative than estimates. If the first module takes substantially longer than expected, six modules in eleven weeks becomes a scope problem, not a productivity problem.

**2. By weeks 4–5: Is there evidence of purchase intent?**
Your email list gives you distribution, but the critical unknown is conversion at $349. You want evidence stronger than opens before most of the course is sunk cost.

The baseline sets three checkpoints, and they are dates: by week 2, by week 5, by week 9. The mode arm builds ten rows, and every one of them runs cause → why it follows from this plan → earliest warning sign → one preventive action. What comes back are observations rather than dates — whether one finished module took longer than the estimate, whether the list clicks as well as opens. Its second signal has no baseline equivalent at all: the baseline never uses the word conversion, so it never asks whether a 38% open rate means anyone will pay $349.

The advice itself largely converges, and a reader comparing only the recommendations would find little between them. Four of the mode arm’s ten preventive actions are things the baseline also says — finish module 1 in week 2 as a pace benchmark, lock the partner details now, use one workbook template, start marketing before week 9 — and the baseline covers checkout, QA and tracking too. The word cause appears nowhere in the baseline; that, not the advice, is what separates them.

Extracts are verbatim from 06_Test_Lab/Experiments/TL-026_PREMORTEM_Raw_Outputs.md.

Related modes

Try it on your own work

Describe what you are actually doing and the Builder writes the full prompt — mode, context, output shape, and the verification step. It picks the mode from your description, which is why this link does not carry one.

Open the Prompt Builder →