Optimising spend on ad buying


Optimising spend on ad buying

"To work with a company like Halfspace has been incredible. The speed, agility, knowledge and expert hands-on development of well scoped machine learning models has been a true eye opener on what is achievable to our organisation", CEO


Lightning fast low latency bidding

Reducing real-time bidding time significantly


The client, which included stakeholders from Spain, Brazil and Argentina, wanted the right tools in order to optimise their spending on ad buying. Specifically, they had banner ads but no indication of which impressions would have high or low click-through-rate (CTR). The client required advanced models which could give insights about CTR of different ads for several campaigns in order to minimise costs by improving bidding strategies and the overall effectiveness of ads.

Our role

Halfspace has been leading an agile project with regular stakeholder meetings to ensure alignment and has outlined a roadmap consisting of six key phases and clearly defined deliverables for each step. Delivered an end-to-end AI and analytics solution which has allowed for immediate actionable insights about banner ads.

Halfspace is in the process of identifying areas of improvement, and further extending the CTR models, implementing more campaign-specific CTR models, introducing bid price prediction models and recommender systems based on user segmentation.


The client was able to get valuable insights from the initial model to implement new bidding strategies. Such bidding strategies are currently being tested and analysed. Valuable and intuitive visualisation tools for monitoring changes in data allow the client and us to know when to update and retrain models.