Discovering valuable intelligence in raw shopping data using Attribution+
A well-established loyalty program has many advantages, among them is the ability to gather insights on shoppers in order to build meaningful relationships with them. But how much data do we really have?
When shoppers register to a loyalty program, they might be asked questions about age, household size or maybe even income and monthly grocery spend. But over time, this data might be incomplete, inaccurate, outdated or simply missing. We might also track shoppers purchasing history at our stores. But how much more valuable would it be to also see the missing opportunities in categories and brands that shoppers buy elsewhere?
To fill these blind spots, GfK and SO1 have developed “Attribution+” — a platform that enables retailers to enrich loyalty program data with information from GfK Consumer Panels and beyond.
Fill in missing socio-demographics and loyalty in store
Imagine you want to run a promotion campaign on a newly introduced premium wine brand targeted at young families earning 50k+ with monthly overall grocery budget of at least 1k. This is the type of data that you hardly ask shoppers directly.
With Attribution+, you don’t need to ask anything. The tool uses advanced machine learning algorithms to derive attributes automatically by observing their shopping behavior and comparing it with available GfK panel data.
This way, loyalty program data can be quickly enriched by new attributes such as age, household size, available income and loyalty in store (how much shoppers spend with you vs. elsewhere).
Discover shopper purchases with your competitors
Now imagine knowing the amount of money your shoppers spend at their top 3 retailers to buy the rest of their requirements. You could specifically design and target promotions to avoid churn, win shoppers back and win over these competitors.
Attribution+ will also provide insights on what categories shoppers buy at competitors, so you are able to increase sales in your underrepresented categories via targeted promotions, product placement or CRM communication (i.e. weekly leaflets, app or email).
How Attribution+ works
The system relies on 2 fundamental elements:
GfK Consumer Panel data – consists of robust shopping histories from thousands of households in different countries over several years
SO1 Engine – a cutting-edge artificial intelligence that is able to learn on basket data and analyze the market at scale.
The tool is using deep machine learning to process GfK panel data and find behavioral patterns (models) within predefined segments. Each model is an ensemble of 400 decision trees. Once the AI learns these models, it is able to identify new shoppers based just on their behavior (co-occurrence of products in their shopping baskets across time) and assign them missing attributes. With this detailed intelligence on socio-demographics, preferences and buying potential, retailers can run highly targeted promotions to increase basket size, raise sales of underperforming categories, discover new brand opportunities, and beat local competitors.
First results show that based on just five shopping trips, Attribution+ is able to correctly predict attributes in ~68 % of the cases. In general, the more historical purchasing data per shopper there is, the more accurate the predictions will be.
Future expansion of Attribution+ will include attributes such as price sensitivity, consumer styles, nutrition habits, and brand type preference.
Attribution+ is a software delivered via email. To experience the value and convenience of Attribution+ for yourself , we are offering a free trial version for 6 months, with full function for 2 core shopper attributes – age and household size. Further attributes of choice can be purchased. Attribution+ runs on premise, meaning no need to provide your data to any third parties. The tool is simple to use, well documented, offers easy integration, and can be used literally out of the box.
It will first be available in Germany, the Netherlands and Poland. In the next few months, we will add all remaining countries that run GfK Consumer Panel such as Austria, Belgium, Czech Republic, Denmark, Hungary, Italy, Romania, and Sweden.
Want to see Attribution+ in action?