Applied Research

Development of Wonder.pics

AdWiser.net Inc.

Challenge/Opportunity:

One of the hottest topics for any online business is creating efficient conversion points: landing pages, display ad, product's pages, promo-widgets, video previews, articles leads etc. And the most important part of them is a right picture. However in most cases such pictures just an arbitrary choice of a designer or an editor. In best cases there are A/B/n tests. Though even this method can increase conversion for dozens per cent, it faces significant difficulties in the selection of the most affecting component - a picture. The problem is that to choose a right picture it is necessary to test many versions of it, but math of A/B/n testing approach needs disproportional increase of views for any new test in a row, what makes it useless in real cases.

Wonder.pics uses an innovative algorithm which is developed especially to choose a statistically significant winner from a large number of pictures-candidates without disproportional increase of views. This dramatically increases chances to find high-efficient picture and significantly uplift conversion or other related goals.

Solution/Collaboration:

The main technical challenge of the project is in developing innovative algorithm, which can solve the current problem with statistically significant result. To do this we are going to use good elaborated scientific approach which hasn’t been used for such kind of commercial problems. On the current stage key elements of technology are selected, some elements of user interface are prototyped, structure of the data base is developed. In collaboration with the college business logic will be implemented and user interface will be elaborated. As a result of this project Adwiser.net Inc. will have ready technical part of the product and can start its business development.


Funder: OCE-VIP
Researcher:

Pedram Faghihi

Area: Information Technology

Projects List

Back to Top