How We'd Run the Next Planning Cycle

Imagine a CEO hands us the keys to her next annual planning cycle. Not advisory, not a workshop — the actual cycle, for a quarter. She wants the document that comes out the other end to be the document the board reads, the executives plan against, and the company runs on. What would we do, week by week, for the next eight weeks?

This is the most concrete piece in our writing, and on purpose. We are often asked, in early conversations, what an engagement actually looks like. The honest answer is: it looks like the calendar below. The detail varies by company; the shape almost never does.

Week 0: the audit

Before we touch the new plan, we audit the old one. This is the step companies most want to skip and the step that produces the highest leverage in the whole cycle. We pull last year's plan, line by line, against last year's actuals. Not just the headline numbers — every meaningful sub-line. Revenue by segment. Headcount ramp by function. Capex against schedule. Major launches against their original ship dates. The five biggest strategic bets against what was promised at the start of the year.

We score each line three ways: was the forecast inside the original range, what was the absolute error, and what was the systematic direction of the error. Most companies discover, in this audit, that they have a signed miss — they consistently overshoot revenue forecasts by a similar percentage, or consistently undershoot launch dates by a similar number of months. The signed error is the most important number in the document, because it tells you which way to adjust this year's forecasts before you've even started. We have written elsewhere about why annual plans miss by twenty percent; the audit is what turns that pattern from anecdote into a number you can multiply by.

The audit also surfaces the decisions that worked because of luck and the ones that didn't despite good framing. We run a decision audit on the prior twelve months — twenty largest decisions, scored on the quality of the framing at the time, independent of how each turned out. The output is one page. It is uncomfortable to read. It is, in our experience, the single most useful artifact a CEO has when she sits down to plan the next twelve months.

By the end of week zero, the company has, on paper: its calibration error from last year, its biggest decision-quality patterns, and a short list of the assumptions that held versus the assumptions that broke. The new plan starts on this foundation, not on a blank page.

Weeks 1–2: reframing strategy as a set of bets

Most strategy documents read as if the company is choosing a destination. We reframe them as bets — what we have written about as strategy is a series of bets. The destination is the same. The language is different. The difference matters.

For each strategic priority in the existing plan, we ask the team to produce three things. What is the bet, expressed as a specific claim about the world ("we believe Segment X will grow at 30% over the next two years, and we are the third-best-positioned company to capture it")? What is our confidence in the bet, expressed as a number? What is the size of the chip we are putting on it, expressed as people and dollars?

Three things tend to happen in these conversations.

The first is that "strategic priorities" get sorted. Some of them are revealed, when forced into the bet form, to be operational hygiene rather than strategy ("we will improve customer success" is not a bet). Some are revealed to be vague enough that the team cannot agree, when pressed, on what the actual claim is ("we will expand internationally" — into where, with what product, against what competition?). The sorting produces a shorter list of real bets, usually four to seven.

The second is that the confidence numbers come out lower than the deck language implied. The team that had been writing about "accelerating in Segment X" turns out, when asked to put a number on it, to be at 55% confidence rather than the 85% the strategy slides suggested. The gap between the deck's confidence and the team's actual confidence is the decision tax made visible. It is also, more usefully, the place where this year's plan gets sharper than last year's.

The third is that the chip sizes get questioned. A bet at 55% confidence with a $40M chip on it looks different from a bet at 80% confidence with the same chip. Sometimes the chip needs to come down. Sometimes the confidence needs to come up, with additional pre-work, before the chip is fully committed. Sometimes the bet needs to be staged — a smaller chip, a kill criterion, a follow-on commitment contingent on what the first chip reveals.

Weeks 3–4: calibrated forecasting for the line items

With the bets framed, we turn to the line-by-line forecasts. Every meaningful number in next year's plan gets a range, not a point, and the range is calibrated against last year's signed error from the audit.

The form is simple. For each line — segment revenue, headcount, capex, major launches — the responsible owner produces an 80% range. Then we adjust the range by the historical track record. If the team's revenue forecasts have systematically come in 15% high, the central tendency of the range gets pulled down by that amount before the range is set. If launch dates have historically slipped by an average of three months, the launch dates in the plan get three months added. These are not optional adjustments. They are corrections for known calibration error, and they are what separates a real forecast from a marketing artifact.

The conversations during these two weeks are some of the most useful in the whole cycle. Owners discover that their stretch numbers and their honest numbers have been quietly merged for years. They discover that the "base case" they have been presenting is, on the team's own honest assessment, a 30th-percentile outcome. They discover that the timeline they have been promising is the one they would hit only if everything they have planned actually happened on schedule, which historically it never has.

The single most expensive forecasting error is the one that has been in the plan, in the same form, for three years running, and which nobody has ever scored.

By the end of week four, the plan has a different shape. The same lines are there. The numbers are mostly in different places — generally more honest, occasionally more aggressive (some teams find they had been sandbagging), and with explicit ranges and stated assumptions for each. The CFO can now do something she could not do before, which is roll the line-item ranges up into a portfolio range for the whole plan, and see honestly what the company is committing to.

Weeks 5–6: premortems on the biggest bets

With the bets framed and the forecasts calibrated, we run premortems on the three biggest bets in the plan. The premortem is free, and at this stage of the cycle it is also load-bearing.

Each premortem takes about an hour, run with the people who will execute the bet. The frame: it is two years from now, this bet has failed badly, what happened? Anonymous written answers first, then read aloud, then ranked by named consequence. By the end of the hour, each big bet has a risk list ranked by specificity, a named owner per risk, and — most importantly — a written kill criterion.

The kill criterion is the artifact we care about most. For each risk, what would the team see, by when, that would tell them the bet is going wrong badly enough to stop or to restructure? Not "we'll keep an eye on it." A specific signal, a specific date, a specific person responsible for checking. Most companies have never written these. The companies that have, almost universally, report that the kill criteria changed the way the year was managed — not because they had to use most of them, but because the act of writing them turned vague monitoring into scheduled decision points.

We also use the premortem output to revise the chip sizes one more time. A bet whose premortem surfaced four specific failure modes, three of which the team had not seriously considered, often deserves a smaller initial chip and a clearer gate before the rest of the commitment. A bet whose premortem produced mostly low-consequence risks deserves the original chip and a faster execution cadence. The plan gets re-balanced against the premortem findings before it moves to the board.

Weeks 7–8: the board-ready document

The last two weeks are spent producing the document the board will read. It is shorter than what the company has produced in previous years. The structure is what matters.

It opens with what the company is betting on — the four to seven framed bets, in plain language, each with a specific claim about the world. For each bet, the document states the company's confidence, the size of the chip, the named owner, and the kill criterion. The line-item forecasts appear as ranges, not points, with the calibration adjustments from the audit explicitly shown. The biggest assumptions in the plan are listed in a single page at the front: the three or four things that, if wrong, would invalidate large pieces of the plan, with explicit triggers for revisiting.

The document also contains, on its last page, what we have started calling the update schedule. For each major bet, the dates on which the team commits to revisit its confidence, and the kinds of evidence that would move it. The schedule is what turns the plan from a wedding vow into a living forecast. It is what the board will use, six months in, to ask the right questions rather than re-litigate the original commitments.

The CEO presents this document, by herself, in about forty-five minutes. There is no deck of supporting slides. There is the document, the conversation, and the schedule of when the conversation will resume. Boards trained on previous years' planning artifacts find the new format jarring for the first cycle. By the second cycle, they will not accept the old format — because the new one tells them, for the first time, what the company is actually betting on and what would change the bet.

What the difference looks like

The plan you end up with reads differently from the one the company has been producing. It is shorter. It is more honest about what is and is not known. It commits the company to specific bets, sized to specific confidences, with specific kill criteria and specific update schedules. It will be wrong in some places — every plan is — but it will be wrong in ways the company can see, name, and act on. The previous plan was wrong in ways nobody discovered until the year was over.

We have run this cycle, in various forms, with companies between $50M and several billion in revenue. The shape holds. The eight weeks compress and expand depending on the company's existing maturity, but the steps are the same and the output is the same: a document the board can read in forty-five minutes, a set of bets the executive team is committed to in writing, and a schedule of when the company will check itself.

The difference between that plan and the one the company has been producing is the difference we get paid for. If your next planning cycle is the same shape as the last one — same deck template, same line items, same single-number forecasts, same vague strategic priorities — and the last one missed in the same direction as the one before it, that is a conversation worth having. The cycle starts whenever you are ready. The first thing we do is the audit.