The end of a quarter isn't one meeting. It is a cascade of retrospectives from the bottom up

In many companies the end of a quarter looks the same. Someone notices that the objectives need closing, the teams add a couple of sentences of conclusions, and straight after that the writing of objectives for the next quarter begins. That isn't a review, it is a formality squeezed in between one round of planning and the next. And it is the moment when the company loses the most, because that is exactly when it has the full set of information about itself.

Working with objectives is working in a cycle. If that cycle is a quarter, its end is the only agreed moment when you can stop and check how far off we were in the assumptions we made at the start of the year. It isn't about assessing people or ticking off what percentage each person delivered. It is about answering the question of whether what we were doing was right, and adapting the objectives if the answer came out differently from what we expected.

For us, a review like this isn't a single meeting. It is a cascade of retrospectives that runs from the bottom up and ends with the board. Its point is a feedback loop: conclusions from execution travel back up and are confronted with the annual goals, the strategy, the market environment and the budget. Only after that confrontation does anyone sit down to write objectives for the next quarter.

This text is about that cascade. What happens at each level, in what order it runs, and what stops working when one of the levels drops out. One assumption is needed up front: we write objectives at company level, not at the level of individual silos, and only then do we put together working groups for those objectives. The reverse order, departments first and then their own objectives, gives you at the end of the quarter as many reviews as there are departments, and not a single review of the company.

The cascade starts where the work happened

The first level is the retrospectives of the working groups. The rule for how many is simple: as many retrospectives as there are objectives for the quarter, one for each group that worked on a given objective. This isn't the quarter review yet, it is only the input for it. Without that input, every higher level talks about impressions, because it has nothing else to talk about.

A team retrospective starts with numbers, not conclusions. Someone enters the current value of each key result at the end of the quarter, and the spreadsheet calculates from it the percentage achieved for each result and for the objective as a whole. This is the dullest part and the easiest to skip, but it sets up the whole conversation: facts first, then conclusions. There is one condition: all the teams keep this data in the same place, because otherwise someone will be collecting it instead of reading it.

Then comes the question the whole retrospective is held for. Was the hypothesis for the quarter validated positively, so we carry on according to plan, or negatively, so we need to replan. To be able to answer that, the hypothesis has to be written down earlier, when the objective is written, in the form: we believe that if we do this, we will manage to achieve that. Linked to the annual goal and the strategic one. An objective without a written hypothesis can be assessed on percentages and nothing more, and the percentage achieved doesn't tell you whether our thinking was right.

Next come three things: what the team learned, which actions really brought it closer to the objective, and what went wrong. This is where teams escape, and very consistently, into a conversation about how working together went and whether we had enough meetings. That is an important conversation, just not this one. The facilitator has to bring it back to the substance of the objective: if the team was building a proof of concept for a product, and testing in the middle of the quarter showed that the client wants completely different functionality, that is exactly what needs probing. Every session leaves documentation, written or photographed, and it goes into one shared repository, most often a board in Miro or Mural.

What to do about it this quarter

The end-of-quarter cascade only works if somebody was looking at the objectives during the quarter. That is a separate rhythm, and we gave it a separate episode in this series: a short update every week, at a fixed time, built into meetings the team already has, instead of adding separate reporting alongside them. Its value is not reporting progress. It is catching that two teams are doing exactly the same thing, or that the same effect can be reached by a shorter route.

At the end of the quarter the order is fixed. First the team retrospectives, then, if you have several levels of objectives, a retrospective of leaders or of the division, which looks for conclusions common to the whole group of objectives. The board comes last. Only after that comes the session to write objectives for the next quarter. The materials for the board retrospective go to everyone before the meeting, so that nobody has to read them out loud, and the meeting itself is an hour, an hour and a half at most. That isn't a boast about efficiency, it is a consequence: the whole cost sits in the preparation.

The year then closes almost by itself. If every quarter has been reviewed properly, at the annual review you have a full set of ready conclusions from all the quarters, and the only work is looking for repetitions in them. For a long time I did this by hand; today I sometimes run those conclusions through a language model so that it points out the recurring elements. The difference is that a company either knows its year, or tries to remember at the end what happened in it.

Failures of this rhythm always look similar. Objectives not updated with concrete values, so the conversation runs on impressions. A review not written down anywhere or scattered across several documents, so after the quarter nobody knows what is where. A meeting everyone turns up to without data and starts looking for it in the room, which is a retrospective of preparation instead of a retrospective of work. And leaving the meeting without a list of actions, which makes it a pleasant chat nobody can be held to account for. The most serious failure, though, is a quiet one: skipping the team level and going straight to the board retrospective. The board then discusses its own impressions of other people's work, and the people who actually worked through the quarter have nowhere to say what they learned and what they would change in the assumptions.

What travels back up, and why

The board level is the only one where conclusions from below meet assumptions from above. A prepared set goes into that meeting: the conclusions of all the teams, the state of the annual goals with lagging metrics, a review of the strategy together with the assumptions about the market and the competition, the financial data and the budget, and the state of the objectives for the OKR rollout itself. Throughout the quarter we were going down, from vision and strategy through annual goals to quarterly ones. Now we go the other way and check whether what happened in operations changes anything in the assumptions above. If it does, the strategic bets need rethinking, not another quarter tacked on to them.

Analysis of the financial data is compulsory at this meeting, and this is where companies stumble. They work on objectives, tick off key results and don't check the budget along the way, as if strategy could be delivered in isolation from money. A separate item, and the easiest to forget, is the objectives for the OKR rollout: not a business result, but a way of working. For example, that in the first quarter around ten leaders work more on objectives than on tasks. If nobody checks this, the rollout quietly stalls at the level of meetings about objectives.

We run the meeting itself in the start, stop, continue format: what we start doing, what we stop, what we carry on with. At the end we write down the changes planned for the next quarter and a list of actions with names and deadlines. A retrospective that ends in a good mood and nothing more is a lost hour for everyone in the room. From that list come three separate things to plan: changes to the annual goals, changes to the strategy, and a session to write objectives for the next quarter. These are three different pieces of work, and that is why they have three separate dates instead of being stuck on to the retrospective itself.

I know that, laid out like this, it sounds heavy. We break it down into its parts not because every company should have exactly this many levels, but because a review is where it is easiest to skip something, and a skipped level only shows a quarter later. After several dozen implementations I also know that the hardest part here is neither gathering the data nor the agenda. The hardest part is watching two things at once: that the conversation doesn't slide into a story about how working together went, and that it ends in a commitment. A company that does the same thing in the same order all year doesn't need to remember anything at the end of the year.