George Orwell said that “to see what’s in front of one’s nose requires a constant struggle”. Searching for the best revenue forecasting models can be a challenging task. With so many different models out there, it’s hard to decide which to choose.
Random forests, linear models, growth estimations, and weighted moving averages are just some of the models used by data scientists to forecast future revenue. They can be complex to understand without the proper statistical skills.
I know our urge is often to adopt a complex, state-of-the-art model, but let’s face it — that’s not always the smartest idea. It’s better to go with a simple model and increase its complexity as you go.
An indexes model forecasts which users are more likely to convert to sales. An index won’t forecast your return on investment (ROI), but it will enable you to evaluate your assumptions and make smart decisions, and it can later be used as the foundation for more complex forecasting models — so you build your model in stages over time, rather than jumping into a complex model right away.
Using an index and a “stages” approach benefits you in several ways:
- It gets you and your staff/team/stakeholders used to forecast models.
- It provides a foundation for more complex models to come.
- It ensures that stakeholders will trust the revenue forecast, and it gives you time to address their concerns.
- It gives you the ability to reassess the model periodically, using control points you create to make sure the model still fits your needs.
The takeaway: Don’t jump in too deep before you learn how to swim in the shallow waters.