Euronics is an association of companies that sell consumer electronics. They possess a large part of the consumer electronics market with over 11,000 independent electronical retailers in 37 countries, among which Germany and France. While this international presence is good for Euronics’ success, it also carries along some challenges. In total, the brand works with 19 PIM systems and 54 separate webstores for consumers. This could obviously lead to inconsistent and incomplete (product) information. Euronics decided that it could be efficient to combine all available product information from all different PIM systems. The only problem was that they didn’t exactly know how they should do it.
It appeared that some countries had more and better product data on certain products than other countries. Due to this insufficiency in quantity and quality, the product data in these other countries was able be enriched with available data from other countries. In order to do this and to get a better overview, product codes (e.g., EAN, UPS or GTIN) needed to be linked. However, product codes can differ per country. Data sets in different countries were not consistent and, in some countries, the quality of data was insufficient. So, the brand needed to think of something else to fix their problem. This is where Squadra Machine Learning Company offered to help finding a solution.
Squadra Machine Learning Company suggested to link separate products to one another by comparing their features. First of all, the data sets of three of Euronics’ organizations were collected and features within these sets were matched. After this, the features were mapped to create a clearer overview. Subsequently, it was checked whether or not the EAN-, UPS- or GTIN-code and model number could be matched, and it appeared that only the model number was able to be matched. After this, the available product features were matched (see the screenshot below). An algorithm was created based on brand name and product code. The next step was to replenish the incomplete data sets based on the data sets that were more complete. After this, the developed tool is able to show a similarity score between products. When a new data set is added, the similarity score offers you information in how similar sets are and there’s also the option to replenish the new data set.
After working with Squadra Machine Learning Company, Euronics has improved the consistency between all of their organizations. This provides customers with information of higher quality, and with the increased consistency and data quality the brand itself is also able to see the added value of the combined data. Moreover, it can be argued that this is the first step to more possibilities: due to the consistent data model, the brand may be able to get a clearer overview of where their products are sold. It would also be possible to ask suppliers for even more product data. With proper use all of this information, there are even more possibilities in automating business processes. For example, based on product features, Powertext.ai is able to generate unique and qualitative product descriptions. In this way, customers are directed to your website more easily. This leads to a better findability on Google which, together with the increased load of information that’s provided to your customers, may in turn lead to a better conversion rate.