The Speech Tax
Why your team already knows what will kill the project, and why they will never tell you in the meeting
The board date is fixed, the integration is not, and half the room is pretending those facts can coexist. That is your kickoff meeting. Marketing does X, engineering hits Y, sales lands Z. Coffee is warm. Nobody says anything useful for longer than anyone is paying attention.
The meeting is theater. Your team has already run the real meeting. They ran it in Slack DMs late on Thursday. They ran it again in the parking lot after the sprint review. The two senior engineers ran it over dinner on Friday when nobody from leadership was there to park the conversation.
You were not in any of those chats.
The findings are quite good. They named the vendor dependency that will slip. They named the compliance lead who has been flagging the same AML gap for three monthly reviews. They named the Q3 deadline as the thing that makes the rest of the plan dishonest. The information exists, it is specific, it is correct. It is not going to reach you.
This is not a failure of your team. They are doing the informal version because the formal version is not safe.
The Pre-Mortem That Already Happened
Annie Duke’s How to Decide walks through the formal pre-mortem. Frame the decision as an autopsy of a failed future. Ask the team to list what went wrong. Specificity rises compared to a standard risk review. The research supports it. The discipline works, when people in the room agree to play.
The problem Duke does not describe is what happens before the meeting. The team has already done this exercise. They did it without the prompt. They did it weeks before you thought to schedule it. In every fintech rollout I have seen, there are two parallel risk conversations running at all times. The formal one lives in the status deck, colour-coded amber at worst, mostly green. The informal one lives in DMs, hallways, after-work drinks, and the ride home. The formal one describes the project leadership wants. The informal one describes the project that will ship.
By the time you schedule the official pre-mortem, the findings are already known. Your challenge is not getting the information. The information is sitting in a group chat your direct reports started weeks ago. Your challenge is that the people who have it have already calculated what saying it in your meeting will cost them.
The accurate pre-mortem is not the one you schedule. It is the one that has been running continuously since the project was approved.
The Speech Tax
Naming a risk inside an organization is not a free action. In functioning teams the cost is low, sometimes even positive - you come across as thoughtful, maybe valuable. In most teams, the cost is real. You are seen as negative. Not a team player. Insufficiently committed. You become the person associated with the problem, which is different from the person who solves the problem. The second one gets promoted. The first one gets reassigned.
This is the Speech Tax. It is paid by whoever names the risk first, and it is paid in career capital. The higher the tax, the quieter the experts. The quieter the experts, the less accurate the information that reaches you. Decisions get made on inertia and optimism. Outcomes get worse. The penalty for being associated with those outcomes rises. The tax is self-reinforcing.
Robert Kegan and Lisa Lahey’s Immunity to Change documents the individual side. People and organizations know what they need to change and systematically cannot, because changing violates a “big assumption” about what will happen if they behave differently. The big assumption behind corporate silence is simple. “If I name this in the meeting, it will cost me.” Usually the assumption is correct. The organization has taught the assumption by rewarding silence and punishing visibility, through thousands of small observed outcomes over years.
Most of us are the assumption.
Wirecard is the state-level version. From 2015 to 2020, FT reporter Dan McCrum and a queue of short sellers and auditors ran a continuous public pre-mortem on the company. The findings were correct. BaFin, the German financial regulator, responded not by investigating Wirecard but by opening a criminal investigation against the FT journalists who named the risk. That is what Speech Tax enforcement looks like from the top. For five years, the market watched what happened to the people who named the risk and drew the obvious conclusion. Then Wirecard collapsed exactly as predicted and everyone acted surprised.
But Pre-Mortems Do Work, Sometimes
Here is the objection that should worry anyone making the argument above.
Pre-mortems work. There are bank CROs who will tell you that their formal annual risk-surfacing exercise caught the counterparty exposure everyone had been quietly flagging in DMs, killed the position in April, and saved the book in October. The discipline does what Duke says it does. It raises specificity. It surfaces what status reports hide. The mechanism is real.
The cleanest operator-facing example is not Pixar or Bezos. It is CCAR. After 2008, the Federal Reserve’s Comprehensive Capital Analysis and Review, formalized through Dodd-Frank, required the largest US banks to run annual stress tests against severely adverse scenarios. Banks have to model their own failure modes, in writing, and submit the results. The process is not run by a founder. It is not run by an exceptional culture. It is run by bureaucrats.
It works.
JPMorgan, Goldman, Citi, BofA have all used the discipline, under protest or otherwise, to surface concentration risks and adjust capital before they hit. The bank that treats CCAR as a genuine pre-mortem and adjusts on the findings is safer than the bank that treats it as a form-filling exercise. Andrew Ross Sorkin’s Too Big to Fail documents in detail what happens when banks do not run this discipline - or run it and ignore the answer. The CCAR regime exists because the alternative is Bear Stearns in March 2008.
So the Speech Tax argument has a hole in it. Formal risk-surfacing works at the largest, most political, most silo-ridden organizations on earth, under conditions nobody would mistake for psychological safety.
Sit with that for a paragraph before reading the next one.
CCAR works because the Fed pays the Speech Tax for everyone in the room. The cost of not naming the risk - failing the test, capital restrictions, dividend blocks, the front page of the FT - exceeds the cost of naming it. An external authority prices honesty higher than internal silence does. That is the condition under which formal pre-mortems produce accurate output. Not culture. Not founder charisma. Price.
You cannot import Bezos. You might be able to engineer that price. Your board can require a written pre-mortem with named authors, attached to the funding decision, reviewed at quarterly intervals against actual outcomes, with consequences for teams whose forecasts keep missing in the same direction. That is steeper than scheduling a meeting. It is also the only move with evidence behind it outside founder-mode cultures.
And even with that price in place, the deepest layer of risk is not the layer the regulator scores. CCAR does not ask your CEO whether the Q3 date they committed on the earnings call is honest. Your board’s pre-mortem review will not ask your CPO whether the pet feature tests well. Those risks still fire.
The pre-mortem works when someone outside the org makes silence more expensive than speech. When that someone does not exist, the artifact produces theater. Both statements are true at once.
The Risks That Stay Off the List
Even with external accountability, the risks your team names in a pre-mortem cluster in one zone. Vendor risk. Integration complexity. Compliance timing. Hiring gap. Scope creep. These are real risks. They also share a property: they are safe to name. They live below the pay grade of the people in the room. Naming them implicates nobody’s career directly.
The risks that actually sink fintech projects usually live one layer up. The CEO has personally committed to a Q3 date on an investor call. The CPO has a pet feature that the customer research does not support. Two business units are in a quiet knife fight over where the P&L lands. The board has a strategic review running that nobody is supposed to talk about, which will reallocate the engineering budget in six weeks. There is a rumored acquisition that no one can confirm.
None of this goes in any pre-mortem. Even CCAR does not solve this. Naming it would require saying, in writing, in a meeting with a record, that the CEO’s commitment is the problem. That the CPO’s favored feature is dead on arrival. That the business unit head’s political game is more important to them than the launch. The career cost of naming these is not “difficult” or “negative.” It is terminal. Everybody in the room knows this, which is why everybody in the room stays in the safer zone.
The formal pre-mortem produces a clean list of risks that the organization is structurally willing to discuss. It produces nothing about the risks that will actually kill the project. Leadership reads the list, congratulates the team on thoroughness, and proceeds. The deeper risks fire on schedule, six months later, labeled in the post-mortem as “external factors” or “unforeseen market conditions.” Neither description is accurate. Everyone knew.
Naming What the Org Will Not Let You Name
The pattern shows up in the same shape everywhere, regardless of company size or geography.
Picture a regional payments rollout in a tier-two European bank. The official risk register has 14 items. All 14 live in the safe zone: vendor certification, sanctions screening timing, integration test coverage, hiring for the late-cycle support team. Three months in, the head of integration writes a paragraph in his personal notes app. “The CPO promised the board real-time settlement by Q4. Real-time settlement requires the new core. The new core does not ship until Q1 at the earliest. Either the date moves, or the feature moves, or someone announces a degraded version. Nobody is going to say this until October.”
He is right. Nobody says it until October. The program goes live with a degraded version nobody announced, and the board is surprised. The post-mortem six months later cites “scope ambiguity in customer commitment.” Everyone in the room when that sentence was written knew it was false. The integration head still has the note.
That integration head did the discipline the organization did not allow. He did not fix the Layer-1 risk - he could not. He named it to himself, with the date, so his future self could not claim surprise. That is the closest thing to honest risk management a middle-layer operator has. The scene repeats in every fintech program I have seen go sideways, always with the same post-mortem euphemism.
If you run pre-mortems, run them for the zone they can reach. Named risks. A pre-committed Kill Rule tied to a specific person (”if the sanctions-screening vendor is not certified by 15 September, Maria Chen cancels the Q4 launch”). Leadership naming three ways their own preferred path could fail before anyone else speaks. Without that last move the meeting is theater. With it, the meeting surfaces the zone the regulator also cares about.
Everything deeper, you write down by yourself. Dated. In your own words. Shared with one trusted person outside your reporting chain. Revisited monthly. Not to send. To remove the option of future surprise.
Run the formal pre-mortem honestly for what it can reach. Do the harder work separately, by yourself, in writing.
The 15-Minute Audit
Grab a notebook. Set a timer. This is not brainstorming. This is measurement.
Price the private pre-mortem (5 min)
Pick your current most important project. List every person who has, in the last thirty days, privately told you - DM, kitchen, car, drinks - that some specific part of this project will fail. Next to each name, write the concrete Speech Tax they would pay to say the same sentence in your next steering committee. Not “discomfort.” The actual currency your organization uses: the project they would be taken off, the promotion cycle they would miss, the cross-functional seat they would lose, the exit conversation that would start. If you cannot write a specific cost for any name on the list, you do not yet understand how your organization prices dissent, which means you are the executive the list is being kept from.
Write the dismissal phrases (5 min)
List the three risks on your project that you know are real, that you have never said in a meeting, and that you never will. For each, write down the exact phrase your CEO, CPO, or board chair would use to dismiss the person who named it. “Not a team player.” “Lacking commitment.” “Off the bus.” “Not seeing the bigger picture.” “Solution-focused, not problem-focused.” If you can write the dismissal phrase in the voice of a specific senior person at your company, that phrase is the Speech Tax invoice. If the phrase comes easily, the tax in your organization is high enough that people have long since stopped arguing. Keep the list. Do not send it.
Open the next meeting with your own three (5 min)
Write the three specific ways your flagship project could fail because of a decision you personally made or endorsed. Not external risks. Not the team’s risks. Your calls, attached to your name, written in the first person. Put them at the top of the agenda for your next team meeting. Read them aloud before any other item. Then require each direct report in the room to name one of their own before the meeting proceeds. No opt-outs, no “pass,” no polite agreement with yours. If you cannot bring yourself to go first in writing, or cannot hold the room until everyone else has, you have already answered the pre-mortem question. Your team was right to keep it in DMs.
Questions worth sitting with
Name the three people on your current project who have told you, privately, that some specific part of it will fail. What is the total career cost to them, priced in the currency your organization actually uses, if they say the same thing in the steering committee next week? Can you personally reduce that cost before the next review, or would reducing it require you to pay a tax you have never paid at this company?
Open your last three project status decks. Find the most recent amber or red item. Now name three people on the team who, if asked privately, would tell you the rating is wrong in either direction. If the answer is “I do not know who would tell me, because nobody ever has,” your team has already calculated the Speech Tax in your meeting and decided to pay it somewhere else.
Write down the last risk you named in writing, with your name attached, to someone who outranked you. If it was more than six months ago, consider whether you are still practicing the discipline you are asking your team to practice, or whether you have quietly become the executive your team has decided not to tell the truth to.
At the start of your next pre-mortem, will you read aloud three ways your own preferred path could fail, in writing, before anyone else speaks, and refuse to move on until every person in the room has added one of their own? If the answer is no, you already know what list you will produce and what will not be on it.