Quantifying 'growth friction' when projecting target goals

As part of my DS work I spend some fraction of my time helping the team make growth projections, either for setting growth targets or when forecasting actual data.

There is obviously a range of ways to go about doing this but the one thing I don't have a good solution for at the moment is being able to fold in or at least quantify how much harder it is to grow in a market this year as opposed to historical years. It is easy to anchor growth on historical rates but in markets at which you've got a solid penetration already and are perhaps heading toward saturation it is likely a harder task to maintain a historical level of growth than it was to achieve that level the first time around.

I've thought about a few approaches that basically involve taking the derivative of a growth curve to look at the rate of change of growth, but it is hard to know if that is really probing the truth of how much harder it is to grow, or if its just that we're doing a worse job and growing more slowly!

Does anyone have a good approach to either measuring some sort of 'growth friction' for a current year vs historical years, or perhaps some other forecasting technique that allows information about this to be folded in somehow?

Topic forecast

Category Data Science


So, your system really has two independent variables:

  • how much effort your company is putting into growing
  • how much your company is actually growing

I agree that the derivative of the growth curve is a good start, but this is an obvious issue:

it is hard to know if that is really probing the truth of how much harder it is to grow, or if its just that we're doing a worse job and growing more slowly!

so, really, I think you need a way to retrospectively analyze how much your company has put into growth. Let's call it growth_x. Whatever current metric your using to calculate growth is growth_y.

I think it's reasonable to assume that the derivative of growth_x should be proportional to the derivative of growth_y in a stable market, and any deviation in proportionality is the result of changes in outside factors.


One way to frame the problem is as a dynamic system, describe and predict the interactions over time between multiple components of a phenomenon.

From your framing, it is important to have predictive ability and interpretability. Using autoregressive moving average (ARMA) model might be an option.

Cohort analysis divides the domain into groups that share common characteristics or experiences within a defined time-span and model patterns across the life-cycle.

About

Geeks Mental is a community that publishes articles and tutorials about Web, Android, Data Science, new techniques and Linux security.