The burn plan the board actually wants from engineering.

What you have probably noticed

The CFO asks for an engineering cost breakdown. You send one. The meeting does not end. The board nods at the headline number and then asks the same question three ways: is this the right amount, what would we lose if it were smaller, what are we actually buying with it. You answer with headcount, tooling, and a cloud bill, and the room stays unsatisfied. The breakdown you handed over is organised by where the money goes. The question is about what it buys. Those are two different documents, and you brought the wrong one.

The tell is that the number itself is never the sticking point. Nobody in the room doubts that you spend what you say you spend. What they cannot get from the spreadsheet is whether the spend still matches the plan the company is actually running, or the plan it was running two years ago.

What the board is actually asking

Strip the politeness and the question is simple. Of everything you spend on engineering, how much keeps the current product alive and shipping, and how much is held against a future you projected and have not yet reached. The first kind buys delivery. The second kind buys optionality: capacity hired a year ahead of the demand that justified it, a platform sized for ten times the traffic, a rewrite begun for a growth curve that has since flattened, a second region nobody uses. Optionality is not waste. It is a bet. And a board that has just watched the growth story slow is asking, correctly, whether you are still paying to hold bets you no longer need. This is the same move as pricing technical debt per month: you cannot manage spend you have not named.

The burn plan that answers it

Sort the spend into three piles, not a dozen.

• Run. Keeping what exists alive: on-call, security, support, and the infrastructure the current product needs to serve today's customers. This should be flat or falling per unit of product. If run is growing faster than the business, that is its own finding and belongs in front of the board on its own line.
• Deliver. The spend that ships the roadmap the company has already committed to this year. This is the part that buys delivery, and it is the last place you cut. Cutting here does not save money. It moves the shortfall to a date further out and calls it a saving.
• Bet. Everything held against a future that has not arrived: capacity ahead of demand, platform work for a scale you have not reached, a rewrite that pays off only if a particular curve resumes. Each bet gets one line. The future it assumes, and what it costs per month to keep holding.

Now the board can read the plan in one pass. Run is the floor. Deliver is the engine. Bet is the portfolio, and it is the only pile where a cut does not slow this year's roadmap. The honest version names which bets have quietly expired: the curve they assumed did not resume, the customer they were built for did not sign, the scale turned out to be two years off instead of two quarters. Those are not failures to bury in a sub-line. They are the answer to the board's question.

What to do about it

1. Force every line into one of the three piles. No fourth pile, no "it is a bit of both". The argument about which pile a given spend belongs in is where the real information lives. Half the value of the exercise is watching a line that everyone assumed was delivery turn out to be a bet nobody had reviewed since it was made.
2. Price the cut, not just the spend. For each bet, write two numbers: what shrinking it saves per month, and what it costs to rebuild if the assumed future does arrive after all. A bet that saves EUR 40 000 a month and takes a quarter to restart is a different decision from one that is gone for good once you stop. The board can make that call. It cannot make it from a total.
3. Bring it before the board asks a fourth time. The plan is worth most when you table it, not when it is dragged out of you. A CTO who walks in with the spend already sorted, and the expired bets already named, is answering the question the board has been circling. That is the difference between being asked to defend a number and being trusted to own it.

I have sat in the version of this meeting where a platform team was the entire disagreement, and the disagreement had lasted three quarters. The team was good and the work was real, so nobody could say the uncomfortable thing: it had been stood up for a scale the company projected years earlier and had not reached. Putting it in the bet pile, with the assumed future written on the same line, let the board settle it in ten minutes. The call went the way the numbers pointed, and the argument that had poisoned three planning cycles was simply over. The same split is what the 3-week diagnosis produces, and it is the groundwork under any honest right-sizing decision.

When not to reach for the knife

The burn plan is a translation tool, not an instruction to cut. Some bets are worth holding through a slow quarter precisely because they are cheap to keep and expensive to restart. Capacity you will genuinely need in two quarters is not optionality you no longer want; it is delivery you have not scheduled yet, and cutting it just to flatter this quarter's number buys you a hiring scramble later at a worse price. Run is the one pile where a cut shows up as an incident three weeks on, not a saving on the slide. The plan exists so the board can decide with the real shape in front of them. If the honest answer is that all three piles are already sized right, the plan says that too, and saying it plainly is worth more than a cut made to look responsible. The overcorrection is to treat the exercise as a mandate to shrink and to cut delivery to make the headline move. If the real problem is that the organisation is simply too large for the stage, that is a restructuring question, not a burn plan, and it is answered differently.

Frequently asked

Written from patterns I've seen across engagements. If any of it sounds too specific to be general, that's why.