Allan Engelhardt and Wan Heah writes on the Barnett Waddingham blog about the challenges of keeping predictive models current and relevant during periods of rapid change.
The past is no longer a good prediction of the future, simply because the present is so different from the past. This will remain true until we have reached some form of steady state in a new normal situation. We don’t yet know what that will look like or how long this transition will take. In some areas we may never return to “normal” as the pace and scale of change increases.
Therefore, all the models that you have built using primarily your customer relationship management (CRM), sales, claims and web data will be out of date. This means that your predictions and your forecasts are likely to be wrong. You may well find that your planning is so far from reality that you may be better off not planning at all! This is a challenge we have given some thought to below.
Key considerations covered in the article include:
Customer sentiment may now be one of your fastest and most important sources of model data. In a time of change, emotions trump habits. It is time to make the most of those customer experience programmes. Use the insights as a driver in your renewals and churn models.
- Are you keeping your models up to date?
- Is your data flowing well?
- Are you involving the right people?
- Do you think … “business first, models second?”
- What about emotional data?
- What external data should you be using?
- Why is it so important to embrace uncertainty?
Before, data was never good enough. Not accurate enough. Not complete enough. Now you have to make decisions with the data that is in front of you. In truth, you always did.