We currently forecast connections at the network level for all operators worldwide. From that dataset, we then derive forecasts for other KPIs. A full list is shown below.
At the operator level we set a “target” value for the number of connections in 5 years’ time. This is based on trend analysis, competition in the market and analyst judgement. It is not a ceiling, but rather a point that we believe the operator will have reached by that time, based on its current performance in the context of the market where it is active. We sanity-check these with the market penetration they imply.
For operators where there is a reasonable amount of reported historic data, we then use an interpolation function to find a best-fit polynomial curve for the historic data and the target value.
For operators where some of the recent historic data is missing, this same function enables us to provide estimated values for the missing points.
For operators where very little historic data is available we use a different approach. If their numbers are reported as a group aggregate (as with e.g. Orange Caribbean), we estimate a split of the total across the properties based on population and wealth of the different countries.
If their numbers are not reported at an aggregate level, we use growth profiles from similar operators in other countries to estimate the change in penetration, then add a factor for the degree of competition in the market.
We update the forecasts in 2 ways:
- Ad-hoc
When new data is released for an operator, our forecasts update automatically with the forecast growth rates being applied from the new actual data point. If the new data point is very different from the forecast data point, we re-forecast that operator from the start.
- Quarterly
Every 3 months we review the entire set of forecasts to make sure that the automatic adjustment process has not generated results that make no sense. We re-forecast any that need to be re-forecast.