One of the biggest campaigns we took part in was one for a national chain of hardware stores. It was an enormous operation - it encompassed over 500 locations at the same time. Each of the stores could make an individual decision on the level of involvement in the campaign - depending on the needs, resources, location, etc. Impression Retail was selling… well, impressions - numbers of viewers of the ads for every store location. The key element of the project was its dynamics - we controlled the incoming data on an ongoing basis and, based on it, we adapted the tool to the recipients. This made it possible to hit the right groups with the right messages. We checked which methodologies were the most efficient and we invested in them, saving on ineffective activities.
The entire frontend was based on dynamic, personalized ad creation. This would not be possible without extended data collection, processing and analysis. Generated ads had dozens of millions of impressions monthly. Adding to this data collected from external providers, including, inter alia, information about devices located near the locations of individual stores - it’s not hard to imagine that a huge data waterfall was constantly flooding our software. We analyzed the received data on a monthly basis and updated the system based on them. Analyzing data and discovering insights, we made necessary changes that allowed us to determine a campaigns success and optimize tactics to achieve desired results.
The whole process required a harmonious team work, in this case a team of six. The scope of work included, first of all, a seamless combination of the frontend (dynamic, personalized ad creator) and python data analysis. Good collaboration means a smoother operation of the entire project, which makes it so much easier to maintain after the main phase is completed.Our team put a lot of emphasis on being available to the client - we had weekly stand-ups and used organization and management tools for tracking progress.