The QBR runs ninety minutes. There are twenty-two initiatives on the tracker. Eighteen are green, three are yellow, one is red. The slides advance in roughly the rhythm of a heartbeat. For each green initiative the owner says, in some variation, "tracking ahead, no concerns." The room nods. For the yellows there is a sentence about timing. For the red there is a longer slide and a wince. Ninety minutes later, everyone agrees the quarter was strong.
Nobody, in the entire ninety minutes, has asked whether the green boxes were earned.
This is the meeting most companies call a strategy review. It is not one. It is a results report with a strategy review's calendar slot. The difference is not academic. A results report tells you what happened. A strategy review tells you what you learned — which is a different and much harder document to produce.
The four questions a real review answers
For each initiative of any size, a real review answers four questions, in order. None are exotic. The discipline is in asking them with a straight face.
What did we know going in. What did we expect. What did we get. And — the only one that matters — where in the gap between expectation and outcome is signal versus noise.
Most QBRs answer only the third. The owner reports what got delivered, in dollars or users or shipped features, and the room treats the number as the entire story. The first two questions are skipped because they are uncomfortable: stating, on the record, what was expected invites the comparison the room has organized itself to avoid. The fourth is skipped because it requires actual thinking, and the meeting is already running long.
The cost of skipping them is not in the meeting. It is in the next decision the same team makes, with the same un-examined model of how the world works, against a new set of initiatives that will produce a new set of green boxes that nobody will examine either.
Signal versus noise is the whole game
A green box can mean several different things. It can mean the team's model of the market was right and their execution was crisp. It can mean their model was wrong but the market moved their way anyway. It can mean their model was right and the market moved against them, but the team executed hard enough to compensate. It can mean the bar was set low enough that anyone could have cleared it. These are four entirely different states of the world, and they each have entirely different implications for what the team should do next.
The results report treats them as the same. It says: green. The strategy review pulls them apart.
A QBR that does not separate signal from noise is a QBR that promotes the lucky and demotes the unlucky, on rotation, until the bench is selected for variance rather than skill.
We have watched this happen at close range. In one engagement — a mid-cap industrial business with a disciplined-looking planning function — we ran a retrospective on the prior eight quarters of initiatives. The greens, examined honestly, split roughly in thirds: a third were well-framed bets that worked, a third were well-framed bets that worked partly because of conditions the team had no control over, and a third were poorly framed bets that worked because the market was forgiving. The reds split the same way in reverse. By the end of the exercise the leadership team could not, with any confidence, say which of their own people were good at their jobs. The QBR had been telling them for two years, and the QBR had been wrong.
Why companies skip this
We have asked, more than once, why the four questions don't get asked. The honest answers are revealing.
The first answer is that asking them is uncomfortable for the initiative owner. Stating, in writing, what you expected exposes the gap between forecast and outcome in a way the green box does not. The owner who said the deal would close at $14M and closed it at $9M has a different conversation than the owner whose slide just says "closed."
The second answer is that asking them is uncomfortable for the CEO. The four questions, asked consistently, reveal that some of the executives in the room have been getting credit for luck. That is a politically expensive thing to surface, and most CEOs do not have the appetite to surface it routinely. So they don't.
The third answer is structural. The QBR template, in most companies, has no field for expected. It has fields for plan, actual, and variance to plan — and "plan" is a number that was negotiated into the document months ago, not a calibrated expectation by the person actually running the work. The two are subtly different. The plan is what the company committed to externally. The expectation is what the operator actually believed would happen. Most reviews compare actual to plan and skip the expectation entirely, which is exactly the comparison that would have surfaced the signal.
What the review should produce
A real strategy review produces three artifacts the results report does not.
It produces a recalibration of the team's model. If the team consistently expected X and got 1.3X, that is information. Either the team is sandbagging, or the market is structurally better than they thought, or the measurement is wrong. Each of those has a different fix. The review is the place where the fix gets named.
It produces a roster of bets re-classified by quality, not outcome. The good-decision-bad-outcome bet gets praised on the record. The bad-decision-good-outcome bet gets praised more quietly, and the lesson — that the team got away with one — gets written down. This is the discipline that compounds, slowly, into a leadership bench that is actually better than its peers, rather than just luckier on a rolling basis.
It produces a short list of things to stop doing. The under-examined feature of QBRs is that they are biased toward continuation: every initiative on the tracker has an owner, the owner has a team, the team has a roadmap, and nobody is incentivized to walk in and say "we should not be doing this." The strategy review is the meeting where that sentence is supposed to be said. If your last four QBRs have produced no such sentence, the meeting is not doing its job.
What this looks like operationally
The lightest version of the fix is a column. Add, to the tracker, a field for expectation at start of period, owned by the operator running the work and locked at the start of the quarter. At the review, compare actual to expectation, not just actual to plan. Ask, for each material variance, whether the gap is the operator's model being wrong, the market being different than assumed, or noise. Write the answer down. Read the prior quarter's answers at the start of each new review.
This is, mechanically, a small change. The first quarter it gets done badly. The second quarter, slightly less badly. By the fourth quarter the conversation has changed shape: the meeting is shorter, the operators are sharper, and the executives who were quietly getting credit for variance are now visible. Some of those executives will leave. The ones who stay will be the ones whose decisions are worth backing.
We help leadership teams redesign their reviews to produce strategy, not theatre. If your last QBR produced no surprises — if every initiative landed within a comfortable band of where it was supposed to — that is not evidence of a well-run company. It is evidence of a review that has been trained not to look.
Writes about decision quality at Bayeseon. Reach the team at hello@bayeseon.com.