Weather and Event-aware business intelligence for the optimization of campaigns and resources

The goal of this solution is to impact prediction of events and weather on sales and marketing campaign performance to suggest optimization of resources and possible marketing and price strategies on category and possibily also brand level.

KEY OBJECTIVES

Key objective is to enable better, faster, informed purchase & bussines decisions through (weekly) salesforce optimization, marketing activities / campaign management optimization (overall and by channel) and possibly also cattegory/supply chain management optimization.

EXPECTED IMPACT

Based on model results we expect:

  • Optimization of Salesforce on weekly level (i.e. extend weekends off-work for retail staff compensated with stronger sales force presence, dependend on the weather forecast);
  • Weekly optimization of promotional activities in general (enforcement of »must have« activities vs. »best performing« activities);
  • Weekly optimization of promotional activities online (how to monetize different behaviour of online campaigns / activities than offline baseline);
  • By predicting how events and weather affect the sales on category and brand level we also expect suggestions on possible marketing and price strategies;
  • Sales of know-how to suppliers based on internal performance track (extension of existing targets).

INDUSTRY IMPACTED

  • Brand Manufacturers/Distributors
  • Retailers in the segment of Consumer Electronics & Home Appliances

PILOT SERVICES

Within the Pilot integrated platform (based on MariaDB technology) was built, where data from BC partners are merged on daily basis (with 24 delay). Data are further enriched via API with weather data sets to predict how weather affects customer interest and sell-out. Database is built with different variables, where we run predictive models in real time and monitor the output of the prediction models on a daily basis.

PROGRESS REPORTING

Model is too much focused on weather (weather mostly becomes secondary factor in other categories). Predicition of sell-out on category and SKU level needs improvement and focus shift to a higher level (demand level channel / category).

In the next steps the existing models need to be improved by: inlcuding weather demand on more aggregated business features like traffic (on&offline), conversion rate and basket value; enrichment of predicitve models with internal marketing events from retailers, calendar events and seasonal differences; broadening predictions to new categories / products and product features.

RESEARCH & INNOVATION DOMAIN: Data integration, data aggregation, analytical and predictive modelling, user interface and visualization

TOOLS INVOLVED:  QMiner, SPSS, MariaDB, MS SQL, QlikSense

CONTACT PERSONPatricija Filipič Orel patricija.orel@bigbang.si – Lovro Verhovšek lovro.verhovsek@bigbang.si