| Business need | To improve the effectiveness of web campaigns, developing statistical models to better classify cookies in terms of “clicking propensity” relative to different content domains (Sport, Movies, Music…). To optimize web campaigns ROI, enhancing a real-time process of cookie-based banner selection and visualization. |
| Smartstat’s action and results | Smartstat is developing inferential models for cookie clicking propensity estimation. Future developments will involve the analysis of temporal effects and the exploitation of supervised classification techniques. Developed models provide Admantx with a set of cookie scores, used for banner selection. The models are being tested with a case-control strategy and are leading to statistically and commercially significant improvements in the click-rate parameter. |
| Other features and by-products | The project is quite complex, due to the large amount of data to be processed (millions of records on a weekly basis), and requires big data ICT processes, provided by the expertise of Sinte. As a by-product of the project, Admantx’s processes of data generation and treatment are being revised and improved, both in terms of quality and efficiency. |
