Find out how a Smart Data Solution from ERBACH can provide daily sales and visitor forecasts for a furniture store chain based on historical data.

Our Smart Data & AI solution enables time and target group-optimized marketing measures that lead to amazingly precise predictions of visitors, customers and sales. In this way, daily visitor and customer numbers, expected sales and the necessary personnel can be planned for individual locations. It is based on historical data from the furniture store chain, which has been supplemented with local weather data as well as public holidays and school vacations.

50-70%

Cost savings for marketing and personnel

Smart Solution

The furniture store chain was able to provide us with complete historical data for the last 3.5 years for the project. This included daily visitor numbers as well as sales of the individual stores according to product groups and information on marketing campaigns carried out with discount amounts and duration.

Erbach Smart Solutions has enriched this information with local weather data as well as vacation and public holidays appropriate to the selected locations of the furniture stores and transferred it to a central Excel spreadsheet. The data collected was around 40 MB in size with approximately 200,000 entries grouped into 25 unique features for analysis and prediction.

The first challenge was to show that Smart Data & AI add value for the furniture store chain. For this purpose, the effect of marketing campaigns on visitor numbers, daily turnover and sold product groups was shown using various analysis methods. In addition, it was examined whether and how Smart Data & AI predictions on visitor numbers and sales are possible. Finally, by combining analytics and predictions, we were able to perform an assessment of daily success. Based on this, targeted marketing measures can now be started.

The starting point of the project was a Smart Data Potential Analysis, which we carried out together with the SDSC BW. Various analytics and machine learning methods were examined and finally a base model was developed to create forecasts using knowledge of visitor and sales figures and marketing measures. The algorithms were developed using statistical information from the time series (e.g. mean and autocorrelation) and finally trained, tested and optimized with historical data.

 Result & Outlook

Merely by merging the various data sources in a central database, remarkable analysis results have already been achieved. This made it possible to show which marketing measures had a particularly positive effect on the sales of certain product groups.

With the optimized and trained Machine Learning Model, daily forecasts of sales and visitor numbers of the individual furniture stores are now possible with a high degree of accuracy (up to 97%). From this, a statement on the respective daily success can be derived in real time for each individual furniture store. Cockpit charts, analyzes and forecasts for individual furniture stores and product groups can be configured and displayed for different user groups in our cloud-based Smart Data Center.

In the next step, the system can be supplemented with additional data sources and charts, e.g. on competitor prices or to optimize supply chains.