Managing complexity and different needs in forecasts

This article was originally published on LinkedIn in 2025

Just finished a forecasting workshop with J+D Forecasting and Ipsen... one thing that's still ruminating as we head to the weekend is that there’s this tendency when building forecasting tools to keep adding — more logic, more detail, more exceptions. I guess it starts with good intentions: trying to meet what seems like endless stakeholder needs and the need cover every possible scenario.

But the result is usually the same: a complex model that’s hard to use, hard to explain, and hard to maintain. I’m just as guilty of this as anyone. I’ve worked on models that started simple and ended up far more complicated than they needed to be. It’s easy to fall into the trap of trying to solve every problem in one place.

I suppose I need to have this printed out every time we build a forecast:

- A model can’t do everything.

- It can’t please everyone all the time.

- And trying to make it do so usually makes it worse.

- Simplicity improves usability and performance.

- Complexity often hides issues rather than solving them.

- Stakeholder expectations need to be managed — not just met.

Easier said than done though...