How to Reflect and Course-Correct as a Leader
Most leaders who try to build a reflection practice start with journalling, produce three weeks of entries, notice that nothing has changed, and stop. The problem is not discipline. Unstructured reflection reliably returns whatever is most emotionally available — usually the last difficult interaction — and produces a sensation of having thought about things without extracting anything reusable. What works is structured, and it is structured differently at different time horizons. This page is the mechanics.
What’s inside
Why unstructured reflection fails
The three horizons
Course-correcting without thrashing
The decision record
The two failure modes
Where Blomma fits
Why unstructured reflection fails
Three reasons, and they are all about the absence of a question.
Recency dominates.Open-ended reflection gravitates to the most recent charged event, which is rarely the most informative one. You end up processing Thursday’s meeting for the fourth time and never noticing the pattern across the quarter.
Feeling substitutes for extraction.Thinking about your week produces a genuine sense of having reviewed it. But a retained lesson requires naming a pattern and a change, and that only happens if something forces it — usually a question you have to answer in writing.
Nothing closes the loop.A reflection that produces an insight and no follow-up is a nice moment. The value is in comparing what you decided to do with what you actually did, and that requires the previous entry to exist and be reread.
The fix in each case is the same: a specific question, answered in writing, revisited. That is considerably less appealing than open reflection and it is the thing that works.
The three horizons
Different time horizons answer different questions, and collapsing them is why most practices produce nothing.
Weekly — ten minutes, three questions.What did I say I would do and did I do it? What surprised me? What am I avoiding? The third question is the highest-value one on this page: your avoidance list, written weekly, is the most accurate development plan you will ever have, and it is the thing you will most want to skip.
Quarterly — an hour, four questions.What actually changed in the company this quarter? Which of my decisions look wrong in hindsight, and was the reasoning wrong or just the outcome? What did I plan to work on, and did I? What is the constraint now? This is where patterns become visible — a weekly review cannot see them and an annual one is too coarse.
Annual — half a day, four questions.What can I do now that I could not a year ago? What did I avoid all year? What decisions am I making without effort that used to be hard? What am I now the constraint on? This is the closest thing to the performance review you no longer receive.
One rule that makes the stack work: each review reads the previous one. A quarterly review that does not open the last quarter’s entry is starting from memory, and memory reconstructs your reasoning to match the outcome.
Course-correcting without thrashing
Reflection that produces constant change is worse than reflection that produces none, and this is the failure mode of leaders who take it seriously.
The pattern: a review surfaces something, you change your approach, next month’s review surfaces something else, you change again. Each change is well-motivated and none is held long enough to take. From your team’s side this reads as an operating model that changes monthly, and unpredictability from the top costs more than any single gap.
Two disciplines prevent it.
Distinguish signal from instance.One bad outcome is an instance. The same thing three times is a signal. Change your approach on signals; note instances and wait. Most of what a weekly review surfaces is an instance.
Change one thing, and hold it a quarter.Behavioural change requires repetition, and repetition requires that you are not simultaneously changing four other things. One deliberate change per quarter, held long enough to become normal, produces more actual difference than six attempted changes.
And when you do change something visible, say so to the people affected and say why. That single act converts what looks like inconsistency into something your team can follow — and it models the behaviour you want from everyone below you.
The decision record
The highest-return component and the one almost nobody maintains.
When you make a significant decision, write three lines: what you decided, what you expect to happen, and what would tell you it was wrong. Ten minutes at the time.
Three things follow. You get calibrated — reviewing a year of these tells you which classes of decision you are reliably good and bad at, which is genuinely impossible to learn from memory because hindsight rewrites your reasoning to match the result. You stop re-deciding, because when new information arrives you can check it against the recorded reasoning rather than reopening the whole question. And you get an honest record when someone later characterises what you were thinking, including yourself.
The critical part is the third line. “What would tell me I was wrong” is what converts a decision from an assertion into something testable, and it is the line that makes the quarterly review useful rather than an exercise in justification.
Review them quarterly. Most founders discover a specific asymmetry — typically too slow on a class of reversible decisions, too fast on a class of people decisions.
The two failure modes
Reflection as rumination.Going over the same difficult event repeatedly without extracting anything. This is not reflection and it makes things worse. The tell is that you have no new question — you are revisiting rather than examining. If a topic has come up three times with no new angle, the honest move is to write down the decision it implies and stop.
Reflection as self-assessment.Turning the practice into a running evaluation of whether you are good enough. Produces a great deal of discomfort and no learning, because the question “am I doing well” has no actionable answer. Keep the questions about behaviour and pattern rather than about worth.
Both failure modes have the same fix: a fixed set of questions. Questions constrain the practice to the useful shape and prevent it from becoming either a loop or a tribunal.
Where Blomma fits
Two things make this practice fail for almost everyone who tries it alone. Nobody asks you the questions, so it gets skipped in exactly the weeks it would matter most. And the most valuable question — what am I avoiding — is the one you will not answer honestly with no one else present.
Blomma is an always-on AI career coach with no stake in your development. Use it as the mechanism for the weekly ten minutes, the quarterly hour, and the annual review, which is the difference between a practice that happens and one you intend to start. Use it for the avoidance question specifically, because that is where the value concentrates and where self-administered reflection reliably goes soft. Use it to keep the decision record and to run the quarterly calibration pass over it. And use it to distinguish signal from instance before you change something — the discipline that separates course-correcting from thrashing.
The second application: this is the practice that most reliably separates managers who plateau from managers who keep improving, and almost none of them have it..
Experience does not compound on its own. Founders running flat out for five years frequently have one year of learning repeated five times, and the only difference is whether anything was reviewed against a question that was written down first.