Forecasting in the time of coronavirus

We now know a lot more about the economic and social impact of coronavirus than we did in the spring, when we predicted a 'new normal' (a phrase that is now much over-used...). But with such huge dislocations occuring in virtually every sector, there is understandable scepticism in the accuracy of any demand forecast. In this article we explain why we still believe Skarp can significantly improve the accuracy of your forecasts.

A reminder

To start with the blindingly obvious, Covid-19 has turned the world upside down. As well as its tragic effect on so many lives, the global lockdown has caused a dislocation in many areas - from shutting down the entire hospitality industry to causing overnight shifts in consumer spending patterns.

Making a hard job harder

For those of us in the business of trying to predict the future, this makes a hard job much, much harder. As we posted about in June, many AI and machine-learning models have been knocked sideways by coronavirus, especially those using historical data to forecast the future.

There is a school of thought that says now is the time for the much-hyped data science algorithms to be put on furlough, and for people (remember them?) to take over.

We think this is taking it too far - yes the world has changed, but those changes are mostly quantifiable and a well-designed predictive analytics engine should always outperform someone with an Excel spreadsheet. It simply isn’t possible to take account of all the factors that affect demand and how they interact with each other without using statistical and data science techniques.

The case for more data

We do agree, though, that predictive algorithms need adapting for the post-coronavirus world. In particular, any forecast that only uses internal data is going to have a hard time keeping up with a major external event like Covid-19.

At Skarp we have always used external data to improve our forecast accuracy and have spent the last few months adding new data streams that help our algorithms make sense of this strange new world. For example, our models now incorporate:

  • Infection numbers
  • Traffic / public transport flows
  • A quantified measure of lockdown severity (increasingly at a local level)
  • Online search volumes
  • Social listening
  • Economic indicators

This doesn’t turn Skarp into an oracle, but it definitely improves forecast accuracy - and is available to all our current and future clients at no additional cost. It’s a neat demonstration of the benefits of using a fully-managed forecasting service over trying to run it in-house.


If you’d like to find how we could help you improve your forecasting accuracy, do drop us a line here.

If you'd like to learn more, these articles might be of interest:

Thanks for reading.

What we do

Skarp uses machine learning-powered predictive analytics to generate accurate, automated demand forecasts - and an explanation of what is actually driving performance.

By removing uncertainty and quantifying the impact of factors affecting performance, Skarp can reduce costs and improve customer satisfaction.

We offer a fully-managed service, designed for organisations with limited in-house data science resources.

Find out more

Get started for free

There is no setup fee or minimum contract term with Skarp, and we offer all new clients a proof of concept free of charge. We believe the accuracy of our forecasts will speak for itself.

Book your demo