One of Pecan’s customers is a mobile ad-tech company, which helps customers all over the globe to accurately reach their target audiences, more specifically mobile users. The company drives mobile engagement to acquire high-value users and maintains a programmatic ad exchange.
The Company’s business advantage is that they place the correct advertisement in front of the right set of eyes, at minimum cost. To optimize the advertising budgets of its customers, the company implements a segmentation process which is based on a combination of 'hard rule' criteria. For example, Jessica will be marked as a sports fan if she uses a sports-related app at least 5 times a week. This criterion, though sensible and accurate most times, is limited in nature. There might be many more people and audiences who don't open sports apps, or simply haven’t downloaded one but are not any less enthusiastic about sports.
Understanding that in order to succeed in business, one must adopt new technology, the company understood that AI was an opportunity to gain deeper insights to broaden the relevant target audiences with more accurate predictions – both on a macro, and micro, level.
The team did not have the time and expertise, namely trained data researches, to embark on a data science project that entails training and maintaining multiple deep-learning predictive analytics models. They also did not have time to waste, they needed to understand where issues may arise, now.
The Company quickly understood that Pecan and its Predictive Analytics Model could address this challenge. The Pecan platform was first used to train a model to build a predictive supervised learning model that uses the hard rule and all its labels but annuls the critical feature defining the segment (i.e. the use of sports applications in our example).
The Pecan platform was first used to train a model to study all the people in the database that had been using sports apps more than 3 times a week, and identify other variables that were shared by this group of people. Next, Pecan created a new database with all the people with the relevant variables but did not use sports apps more than 3 times a week.
Afterward, Pecan built a model that predicted how likely the people in the new database are sports fans. These look-a-likes broadened the cluster of relevant sports fans which an ad campaign can address effectively.
Using Pecan, the customer was able to expand its high-value target audience by over 200%, increasing its clients’ performance marketing campaigns’ effectiveness by 12%-23%.
Pecan’s platform allowed the company to build the AI models faster, 5 days versus 4 months, and at a significantly lower cost as a result of Pecan’s automated AI tools.
The Pecan based solution broadens dramatically – in order of magnitude of hundreds percent– the size of the cluster of any target audience which is relevant to any campaign. Advertisers can now optimize their ad budgets, increase ROI and revenues and increase loyalty to Pecan's customer.