Part III — Cultivating Organizational Flow · Chapter 51 · 5 min read · First Public Draft

Learning

Closing the Loop

The shortest, cheapest stage in the whole circulation, and the one almost everybody skips.

Reduces frictionCreates valueLearning cycle time

We Have Solved This Before names three reasons the loop stays open: the outcome arrives somewhere other than the decision, the learning is owned by a document rather than a person, and admitting the lesson costs somebody something. This chapter is about those three, in that order, because they are not equally hard and organizations reliably attack them in the wrong sequence.

The wrong sequence is to start with the third. Announce psychological safety, run a blameless post-mortem, and hope the honesty arrives. It occasionally works and it usually produces a room where everyone is being very careful about being blameless.

The first two are structural, and doing them first makes the third considerably cheaper.

Send the outcome back to whoever chose

The most common reason nothing is learned is that the people who found out are not the people who decided.

The team who built it know exactly how it landed — they hear from users, they see the support tickets, they watch the adoption curve flatten. The people who chose to fund it have moved on to next year’s portfolio and will encounter this decision again only as a line in a review deck.

So the practice is unglamorous: when an outcome arrives, route it to whoever made the call, by name, and do it whether or not the news is good.

That is it. No forum, no process. The thing that makes it work is the by name part, because a report addressed to a distribution list is addressed to nobody, and everyone on it assumes somebody else is the intended reader.

I have watched this change behaviour faster than anything else in this part of the paper, and the mechanism is not shame. It is simply that people who find out how their decisions turned out start making different decisions, and people who never find out cannot.

Give the lesson an owner, not a document

A retrospective produces actions, the actions go into a list, and the list has no owner — which is how a genuine insight becomes an unread row in a spreadsheet within a fortnight.

The fix is to stop producing lists. One lesson, one name, one date. If nobody will take it, it was not a lesson; it was a complaint, and it is better to say so in the room than to bury it in a document where it will look like progress for six months.

Which means most retrospectives should produce fewer outputs than they currently do. A session that generates eleven actions has generated none. A session that generates one thing somebody will actually do has changed the next quarter.

And write down what you expected, not only what you decided. This is the same discipline that Measuring Value asks for, one layer down. A decision record with no expectation in it cannot be checked later, because there is nothing to check it against. We think this will do X by March costs one sentence and converts a future argument about interpretation into a factual question.

Then make the third one survivable

Once outcomes reach deciders and lessons have owners, you will eventually hit the case where the honest conclusion is that a senior person got something wrong.

This is where psychological safety stops being a pleasant idea and becomes an operating requirement — and not in the comfortable sense. It is the specific ability to say what happened without it costing the person saying it, and it is produced almost entirely by what visibly happens the first two or three times somebody does.

Which puts most of the weight on one thing: whether the senior person can say it about themselves first. Not as a performance of humility at an all-hands. In the ordinary meeting, about the ordinary decision, in the same tone they would use about anyone else’s.

I have seen that done well, and the effect is immediate and slightly startling — the room recalibrates in about ten seconds. I have also failed to do it, more than once, and the cost is not that people call you out. It is that they quietly stop bringing you the difficult half of what they know, and you do not find out for a year.

Let the loop close on things that went well, too

Organizations examine failures and celebrate successes, which means half the available learning is being applauded rather than examined.

The question after something worked is the same question: what would we do again on purpose, and what happened by luck? Teams are usually honest about this if asked, and almost never asked, because the celebration is the point and nobody wants to interrogate a good quarter.

What it costs

Very little, which is the strange part, and I think it is why the loop stays open. Nothing here needs budget, headcount or a programme. It needs somebody to send a message they were not going to send, and somebody senior to say a sentence that is mildly uncomfortable.

The reason it does not happen is not cost. It is that no single instance seems worth the effort — and the return arrives a year later, in decisions that were never made, which nobody will ever attribute to this.

An organization that closes the loop learns faster, and the part you notice first is that it stops paying for the same lesson twice.

This cultivates: We Have Solved This Before and It Breaks in the Joints, in Part II.