Two weeks ago, my post on the back-to-office debate referenced an episode of The Curiosity Shop by Brené Brown and Adam Grant. Somewhere in that conversation they reached for the iceberg model, which I touched on briefly in the post but did not have room to do justice. It is a tool I use regularly with clients. So I thought it might be a good idea to share my thoughts about it this week.
The iceberg has four layers. Only one is above the waterline, and that one is the event. Below the surface, in order of depth, are patterns, then structures, then mental models. The deeper the layer, the harder it is to see and to change. The lower you go, the higher the impact when you manage to shift something. The model itself is not ours. It comes out of systems thinking, where it has been around for decades, and I use it because it works.
Events are the visible part. A missed deadline, a meeting that went badly, a colleague’s email arriving late at night. The question is “what just happened?” and the response is “react.” Most reactive management lives here, which is also the hamster wheel that exhausts research team leads. The only way out is downward, into the water, where you have to do real work to find what produced the event in the first place. You get wet. But the reward is a permanent solution rather than another reaction three weeks later.
One important caveat. If something happens exactly once, treat it as a one-time event and react. Do not overanalyze it, search for root causes, or redesign your process because the projector died during one presentation. Not every event is the tip of an iceberg.
Patterns are events that have happened before, and more than once. The question becomes “how often, and under what conditions?” The response is “anticipate,” which is better than reacting but still leaves the underlying machinery untouched. You can become skilled at predicting trouble without ever stopping it.
Layer three is structures. The routines, the agreements, the meeting formats, the incentives, the metrics, the handing of work from one stage to the next. Everything in your team that has been built, and that you can therefore rebuild. The question is “what is producing the repeating behavior?” and the response is “design.” This is where real change starts, because structures are something you can both touch and redesign. It is also the level where I personally find the best effort-to-outcome ratio in leadership work.
The fourth layer, at the bottom, is mental models. These are the beliefs underneath everything else. What does this team think leadership is for? What does a good scientist look like, and what does asking for help signal about the person doing the asking? The response is “transform.” Of any layer, this one has the highest impact and the slowest movement. Beliefs do shift, eventually, but on their own schedule and never on yours.
Take a research team that has missed a deadline. Maybe one in your own building, maybe one you have only seen from the outside. Look back over the last six months and find that four of the last seven deadlines were also missed. That is a pattern, and patterns earn their own question.
Drop to structures and ask what might be producing it. Maybe there is no real performance management in the group, so the pattern has had nothing to interrupt it. Or the big picture is not shared, so people do not know which deadline matters more than which, or whether some could have been moved. Communication might be thin, and team members might not help each other because no one knows what anyone else is actually working on. Any one of these could explain the pattern. Several together is even more likely.
Drop one more level, to mental models. Now the question is what beliefs might be running under the structures. Asking for help could be read on this team as a sign of weakness. Quality might be held untouchable, even when it is the quality itself that costs every deadline. Deadlines themselves might be seen as bureaucratic intrusions on real work. And when nobody seems to be checking in, the unspoken conclusion may be that the work cannot be important enough to track. On real teams, the answer is almost always some combination of these, plus things you would only see from inside.
Diagnosis is not the point. What matters is what happens if you stay at the surface. Telling the team to “hit the next deadline” addresses none of the structures, none of the beliefs, and none of the patterns that produced the missed deadline in the first place. So the next one gets missed in roughly the same way, and three weeks later you are having the same conversation again.
Mental model work is essential, but it is slow. Beliefs do not change because someone tells them to change. They change when the structures around them shift, and people start to live inside a different system. Structural change, by contrast, you can design and put in place in weeks. My preference is always to start there.
I am working with a client right now who leads a research group where he wants every member to bring their own plans toward a shared goal, and the weekly agenda is supposed to emerge from the team rather than from the project owner. But under stress (and there is always stress) his project owners default to a different mode. They tend to be more directive, assigning tasks to project team members directly. The mental model running them in those moments is something close to “if I do not take charge, this falls apart,” or “leadership means deciding.” Changing that belief at the root would be the deepest possible fix.
Let me ask you something. What do you default to when you are under stress? Probably something in the same family. And could you change that pattern starting tomorrow morning, on purpose, by deciding to? I doubt it. I cannot either.
This is where structures earn their keep. A two-minute check at the end of every planning meeting that asks how tasks were attributed. A simple count of what share of the week’s plan came from team members versus from the project owner. Asking the team to bring written proposals before the planning meeting starts. Rotating who chairs the planning meeting. Putting a clearer prioritization process earlier in the project, so the planning meeting itself does not have to absorb every accumulated decision. None of these require the project owner to “be different.” They change the structure itself, making the old reflex harder to slip into and the new reflex easier to land in. That is most of the work.
What happens after that is slower and less dramatic than people expect. In the first weeks, the project owner may feel the new structures slow him down and take longer than just deciding. After a while, someone says something useful in a planning meeting that he would not have said himself. The team produces a plan that is stronger than what he would have written alone. At some point in there, without anyone announcing it, the mental model starts to move. This is the order that has worked for me: structure first, then behavior, then belief.
Start below the waterline, but start at structures.
It is easy to stay at level one. Every problem is treated as an event. The fight about office days from last week’s post is a level-one debate. So is “how do we make the next deadline?” and “why did this person quit?” and a long list of others. The cost of staying at the surface is that you keep paying for the same problem with different surface fixes, and the problem keeps coming back, a little worse each time.
When you find yourself reacting to the same problem for the second time, that is the signal to look down. Patterns first, structures next, beliefs last, in that order if you are following my preference. For research team leads, this is one of the highest-leverage shifts you can make in how you spend your attention. You do not need a consultant to do it. A notebook is enough, an hour without interruption helps, and the willingness to ask the next question after “what just happened?” will do more for your team in six months than any amount of advice from people like me.
– Robert
