?>

Revenue forecasting models improved enormously

Revenue forecasting is nothing new. However, compared to ten years ago, a huge development has taken place. There are many more data sources available now, most of this data is more current and accurate, and forecasting models are much more sophisticated.

Mattijn Bezemer, co-founder of The Big Data Company, specializes in sales forecasting for ten years. The Big Data Company works for various (international) retailers to estimate the turnover of new retail locations. To do so, they prefer to use the rich and complete data set Locatus has to offer.

Identifying local opportunities

“Despite the major challenges in the retail landscape, there are local opportunities. The trick is to identify opportunities based on the presence of customer profiles, the strength of the type of shopping area and the positioning among competing offerings,” says Bezemer.

Together with Locatus, we have developed the Retail Revenue Calculator. This supplies retailers with a tool that would normally only be available for multinationals.

More and better data

In addition to continuous fieldwork, Locatus uses scraping and digital desk research to continuously update its dataset. This has greatly improved the quality of the data, which forms the raw material for modelling. New data sources include daytime population (where people stay during the day), detailed data on consumer spending via Whooz, future building plans and travel time modules for pedestrians and cyclists. In addition, retailers have more and more data at their disposal themselves. Think of customer reviews, visitor counters, customer origin data or purchase data from loyalty systems.

Better models

The innovations in the field of model development are rapid. In recent years, many new methods and techniques have been developed to unlock, enrich and model data. The set of tools available to the econometrician, data scientist and ICT specialist has become much more extensive. You have to be careful not to drown in a sea of data and possibilities. There is much more to it than that.

“Ten years ago, the development of a model was often assigned to one person; the possibilities were limited. Nowadays, there is a development team ready to work, cleverly combining various disciplines and techniques. This makes it possible to issue turnover forecasts with a higher degree of accuracy, often with deviations of less than 10%”, says Bezemer.

Faster analyses

Some prediction models have been around for decades, but were mainly suitable for scientific research until recently. This applies, for example, to the gravity models developed by The Big Data Company together with Locatus. This type of model distributes consumer spending in the market over the available supply. The basic assumption here is that consumers make a trade-off based on travel time and the attractiveness of the supply. “The models work particularly well if there is a lot of information available about consumers and their decision-making process. With better data, the models work better.

This type of model makes it possible to calculate multiple scenarios and to provide insight into the impact of a series of openings and closures in cohesion. 

Whereas ten years ago it took days to run such a complex model with millions of calculations, the Retail Revenue Calculator can now do this within minutes.

The Retail Revenue Calculator enables Locatus and the Big Data Company to offer retailers a tool that would normally only be available to multinationals. Bezemer: “With better sales forecasts, we’re not only improving the success rate of formulas, we’re also helping to make shopping areas in the Netherlands, Belgium and Luxembourg more future-proof.”

Astrid Custers

Astrid is Marketing Communication Manager at Locatus and thus comes into contact with lots of interesting information. As a result, she walks around the shopping streets with a certain degree of professional deformation – and writes about these observations…