Common Org Design Mistakes Founders Make
Org design errors are unusually repeatable. Founders in completely different businesses make the same ten mistakes, in roughly the same order, for the same understandable reasons — and each one has a recognisable symptom and a known fix. What makes them expensive is not that they are subtle; it is that the symptom always appears to be about a person, so the fix gets applied to the wrong thing. This page is a self-audit.
What’s inside
Ten mistakes, with symptoms and fixes
The meta-mistake
Where Blomma fits
Ten mistakes, with symptoms and fixes
One. Designing a chart that only works with you in the middle.
Symptom:your two-week absence stalls three things; most escalation paths end at your desk.
Why:every individual decision to retain something was efficient at the time.
Fix:map the escalation paths, then assign or write a rule for every conflict that resolves only with you.
Two. Giving someone an outcome without the decisions to affect it.
Symptom:a capable senior hire is underperforming, and it is being attributed to them.
Why:you handed over the function and kept the calls that drive it — pricing, roadmap, headcount.
Fix:for each senior role, list the five to eight decisions it owns and trace where each actually sits today. Move them or change the outcome.
Three. Building roles around people instead of decisions.
Symptom:a senior role nobody can describe, which becomes unfillable when the person leaves.
Why:keeping good people is a real priority and assembling a role around them is the fastest way to do it.
Fix:define outcome, decisions, boundaries, and measure for each role. Person-shaped is fine early — write down that you have done it and what the role would look like rebuilt.
Four. Splitting teams by discipline when the work is cross-disciplinary.
Symptom:every piece of work needs two teams to agree on priority.
Why:discipline splits look tidy and match how people describe their skills.
Fix:run the traffic test — list the ten most common pieces of work and count how many cross the boundary. Redraw where traffic is lightest.
Five. Adding layers to solve a decision-rights problem.
Symptom:you added a management layer and decisions got slower, not faster.
Why:slowness looks like insufficient management capacity, and adding a manager is more actionable than writing a document.
Fix:write the decision rights first. Most companies that think they need a layer need one page instead.
Six. Reorganising to solve an individual problem.
Symptom:a boundary in your chart exists to keep two people apart, or to give one person a bigger remit.
Why:the structural change is easier than the conversation.
Fix:have the conversation. Structural workarounds for relationship problems become permanent inefficiencies that outlast both people.
Seven. Ignoring weighted span.
Symptom:managers who look fine on the chart are underwater; development conversations have quietly stopped.
Why:the chart counts reports and does not weight for new hires, dissimilar work, interdependence, or a manager also carrying delivery.
Fix:audit spans with those four weightings. Expect to find two or three managers running far wider than the chart implies.
Eight. Announcing a restructure before the details are settled.
Symptom:a week of questions you cannot answer, and a story forming that the change was not thought through.
Why:the pressure to say something once the circle widens.
Fix:settle terms, successors, and individual messages first. Then announce everything in one day, individuals before groups.
Nine. Presenting a change as having no downside.
Symptom:people stop believing your explanation, and the reorg gets read as politics.
Why:naming losses feels like weakening your case.
Fix:write the one-sentence trade-off — this makes X faster and Y slower, and we are accepting that because Z. Say it out loud. If you cannot complete the sentence, you have not designed anything.
Ten. Restructuring too often.
Symptom:a reorg every six months; people no longer invest in the current arrangement.
Why:each individual change was defensible, and reorgs feel like progress.
Fix:annual org planning with quarterly checks, so changes are adjustments rather than corrections. Roughly once a year is the sustainable ceiling.
The meta-mistake
Underneath most of the ten is one pattern: treating a structural problem as a people problem, because the symptom always arrives attached to a person.
A slow team, an executive who does not take initiative, a function that keeps missing, a manager who seems to be micromanaging — every one of those has an available person-explanation, and the person-explanation is more actionable this week than a structural fix that takes a quarter. So it gets used, and the structural cause survives.
The reliable tell is repetition. If you have replaced someone and the identical problem reappeared with their successor, the cause was structural and the replacement was an expensive way to find out. Two occupants of the same seat with the same outcome is close to conclusive.
The question that catches it:would the best possible person in this seat succeed, given the authority this seat actually holds?If not, you have a design problem, and no hire will fix it.
One honest addendum. Founders have a second incentive that keeps this pattern alive: the structural fixes on this list mostly require giving something up — a decision you enjoy making, a veto you have retained, a relationship you personally hold. The person-explanation asks nothing of you. That asymmetry is worth noticing in yourself, because it explains why these mistakes persist in companies whose founders can describe them accurately.
Where Blomma fits
Read the ten again and notice how many trace back to the founder rather than to the chart: retaining decisions, keeping the veto, avoiding the conversation and reorganising instead, presenting a change without its cost. Those are the expensive ones, and they are the hardest to see from your seat, because the people best placed to tell you report to you and have every reason not to.
Blomma is an always-on AI career coach with no stake in your structure. Use it to run the ten as an audit and find the two you are actually making — most founders have two or three, and one is usually mistake one or two. Use it to run the attribution question on the situation you are currently reading as a people problem, particularly if you are about to replace someone. Use it to write the one-sentence trade-off before your next structural change. And use it in the moment that determines whether any structural fix holds: the fortnight after you hand a decision over, when the outcome is worse than yours would have been and every instinct says take it back.
There is a second application that runs through this whole cluster. Almost every structural change creates someone doing a job they have not done before — a first-time manager after a split, a manager of managers after a layer is added, a leader absorbing a merged team. Your capacity to develop those people is thinnest exactly when the company is changing shape fastest. Coaching that does not depend on a manager’s calendar having room is how a company grows its bench rather than just its headcount..
None of these mistakes reflects poor judgment. They are what happens when a structure built for eight people keeps running a company of ninety, and nobody has revisited the design. Naming which two you are making is most of the work.