There were three main take aways from this structure:

Takeaway #1: Break down a numerical problem mathematically as long as the math remains meaningful/insightful – then get more layers using qualitative “mini-MECE-structures”

As with most thing problem-solving related, this is not a rule written in stone.

There are a few numerical problems that are best structured with a qualitative structure. And you don’t always need to do the qualitative layers afterwards.

But usually the best way to break down a math problem initially is to break it down into an equation first, as you’ll be able to quantify how each driver contributed to the problem.

And usually the equation alone won’t be enough to bring you to the meaningful stuff.聽

In this case, for example, if we were only mathematical in our structuring we would have missed important elements that show real world business intuition, such as “maintenance”, “aircraft weight” and “mix of aircraft in the fleet”.

Takeaway #2: Stop each branch when it can reasonably聽explain the source of the problem

I have stopped some parts of my tree in Layer 2, other parts in Layer 3 and others in Layer 4.

How did I make this call?

A lot of people have asked me this in the past: how can I know that my Issue Tree is done? How many layers do I need?

The rule of thumb is to stop when your buckets can reasonably explain the problem.

For example, on Layer 2 you have a bucket which is “# of trips flown has risen”. This can reasonably explain why fuel costs might have risen. It’s pretty logical – if you fly more trips, your fuel costs will rise as well.

Now, one could ask “why has the # of trips flown risen” and if that’s the actual problem going on, I as a consultant would want to know that. But that’s getting granular, you don’t need to go that far unless the problem is proven to be there.

If I told my mom or someone on the street that an airline’s fuel costs have risen because the # of trips have risen, they’d accept the answer and probably not question it further (and they certainly would tell me I’m a weirdo for caring about an airline’s fuel costs).

Now, if I told my mom or a random guy on the street that fuel costs have risen because liters of fuel per km flown have risen they would: (1) think I’m really really weird, and (2) not take that answer as it is.

Even if I used more accessible language and said that this airline’s fuel efficiency was down, they’d still ask me “why is it down”? (That is, assuming my mom is actually interested about airlines).

If I had stopped that branch on the 2nd layer, I wouldn’t be telling the whole story.聽

And so I went a level deeper.

Now, on the 3rd layer if I say that fuel efficiency is down because we’re using less efficient types of aircraft, most people would be satisfied with that answer. I can stop the Issue Tree here.

But in the case we’re flying the same aircraft, most people would NOT be satisfied. They’d be like “Hey, you’re telling me you’re less fuel efficient even though we’re flying the same aircraft? How come?”

And so we dig a level deeper on that one. Maybe the aircrafts are flying with more weight. Or we’re doing less maintenance. Or we’re flying at lower altitude and facing denser atmosphere. Or our pilots are changing speed all the time.聽

Most people would take any of those as sufficient answer. Which means we don’t need to dig a level deeper.

Takeaway #3: You can still go deeper in the buckets you need

If the last take away gives you an idea on where to stop structuring the Issue Tree, this one gives you permission to dig deeper than that.

Say your interviewer tells you the problem is that this airlines is flying their planes heavier and asks why that might be. Well, weight was at the end of our tree, right? But we can still investigate the reasons behind that increased weight.

Here I would segment the things that add weight to airplanes into their categories: people, cargo, equipment, fuel itself (we may be flying with excess fuel and thus spending more fuel to carry fuel itself).

Or say that the interviewer tells you that fuel prices have gone up even though we’re buying the same product from the same supplier.聽

Why that might be happening?

Well, either this supplier’s cost has gone up (because crude oil is up in price, for example) or their margins are higher (because we’re not negotiating as well, for example). We could dig deeper into each one of these factors if need be.

The point here is that even though you need somewhere to stop your Issue Tree (otherwise you’d spend the whole day building 15 layers), you also need to be aware that you can go as deep as you need to in the specific parts of your structure that the problem really is.

You find where the problem really is by getting data, numerical or not, for each part of your structure.