OKRs don't lack execution. They lack an answer to how you know the objective is the right one

Working with objectives tells you how to deliver an objective and how to know it has been achieved. It says nothing, however, about where to get the objective from in the first place. Evidence-Based Management answers this by measuring four value areas: the current value the company delivers today, time to market, the ability to innovate, and unrealised value. The fourth is the most important in this conversation, because it sets where the company should go, and at the same time it is almost always neglected. It is established by researching the market and asking users, never by internal discussion, and this is exactly the step at which organisations stumble in implementation. A person responsible for a product, asked what their users think, replies that they are not allowed to talk to them, or that another department handles that and doesn't share its data. The framework then forces a conversation nobody had before.

The strongest line of the episode, though, concerns how this measurement tends to be implemented. Around ninety per cent of implementations take only the metrics from the framework and skip the goals, and a separate ninety per cent skip the hypotheses and experiments, the whole part that turns measurement into a decision. Measurement alone then brings no value at all, because nothing follows from it. Add to that the second typical mistake: the company measures one area, the one closest to the department running the implementation. The business looks at current value, delivery looks at its own pace, and you can deliver a million new features without knowing what value they carry. In one of the companies described in the episode the objective was: deliver a great many new things. Measurement showed that the real cause of user dissatisfaction was old technology that had to be put right across the whole organisation, not more features.

There is also a difference in the very way of thinking about objectives, and it only came out during this conversation. Objectives are not created by a waterfall-style split of a strategic goal into smaller pieces. They are created by choice: what is our best hypothesis for this year, and then what is our best hypothesis for this quarter. A quarter is short enough to say honestly afterwards that the assumption makes no sense and go somewhere else. The same order gives rise to the rule of one experiment at a time: with five running in parallel you don't know which one helped and which one did harm, and this isn't a metaphor, because Evidence-Based Management came to business from medicine. I have to say, too, what our side of this pair lacks. There is no official guide to OKRs, there are good practices and books, so when someone asks us for the one correct way to write them, we have no document we can point to. Copying the model of one famous technology company ends exactly the way it ends: it works for them, not necessarily for you. Both frameworks are, moreover, uncomfortable for organisations, because they expose inefficiency and money being burned, and resistance during implementation often comes from the fact that someone already knows something is wrong and would rather keep things as they are.