Intergamma stored and analysed their data on-premise, which made their data hard to access and use. They wanted to become a data-driven enterprise and manage their operations in a more efficient way, at one place with real time access.
Making sense of all their data was another challenge since they were all unstructured. Intergamma wanted wanted to help customers find exactly what they were searching for when doing on-site search queries. They wanted to give personalised suggestions based on their past search and purchasing behaviour. They were looking for an artificial intelligence solution that would provide better insights.
Crystalloids started the process by migrating all the data, stored on different platforms, into one secured data lake; the Google Cloud Platform. Using Google Cloud Storage and Google BigQuery the data became easily accessible with business intelligence tools like InsightOS and Tableau. For the more advanced data analysts and scientists Cloud Datalab and Google BigQuery were used.
The data was mainly customer focused, so all the different information about one customer was brought together at one place and linked to one customer ID. The products they buy, the frequency of buying, the newsletters they receive, all the information was stored and linked. Also, the web behaviour of a unique client was tracked and used for making personal recommendations, both on-site as through other channels like the newsletter.
The next step was introducing machine learning to make sense of the unstructured data Intergamma was storing. For this, Crystalloids used TensorFlow, the open-source software library for machine intelligence to meet the need for systems capable of building and training neural networks.
The Google Analytics data of Intergamma provided all the information about on-site keyword search, product views, product carting, product checkout and product purchases which were used to train the neural network. In the end, machine learning made it possible to optimise the keyword search result order also called on-site search specific boosting.