Case study

Buily Product Ads

Buily Product Ads

The Buily Product Ads service is a contextualized advertising system dedicated to eCommerce solutions. Our Adserver downloads information about the shop’s inventory and product assortment and takes into account thematic similarities of ads and web pages to place contextually appropriate ads at partner sites.

Specific features of the service

For many marketers the main issue with managing online campaigns for thousands of products is the need to manually update the ads every time a price or stock availability changes. In order to automate the process, the internet shops integrate their offer with the Buily Product Ads system. This integration is made possible through the use of regularly loaded and updated XML files.

For the sake of efficient and effective cost management, the advertiser is presented with the ability to group the products according to their brand or category. The advertiser is also able to set a CPC rate, which may be reflected as a percentage of the price for each product in each group. To assure a perfect match between ads and websites, the Buily Prouct Ads system carefully analyses the content of a given website. It then utilizes the Vickrey algorithm, selects only the matching ads and initiates the bidding process.

Functional and graphic design

Since Buily Product Ads introduces a few innovative solutions, it was incumbent to design an interface that is entirely intuitive and easy to use.

Buily Product Ads screen

Axure RP software was used to design a fully clickable mock up of both the front end and back end system of the Buily Product Ads. To get the chance to test out the usability of the system, we conducted user experience tests. All the information gathered in those stages proved useful for our graphic designers, programmers and web masters. Finally it resulted in the creation of a fully functional and user friendly system.


It is crucial for every AdServer to have an efficient and reliable system. This can only be achieved by replicating both the front end as well as the database servers. Being aware of the fact that a sudden rise in system load or rapid increase of users can cause system overload or a crash, we applied a highly scalable architecture and caching tools, such as Varnish and Memcached to prevent such situations from happening. Contextual ad matching is accomplished by a careful analysis of the webpage in terms of the relevant keywords associated with the given webpage. The next step is locating the relevant keywords in the names and descriptions of numbers of products stored on the Lucene Solr server. Finally, a selection stage takes place. From a group of matching products, following a Vickrey algorithm and taking the CPC rates into account, appropriate to the corresponding web page content, ads are chosen.

This is a brief process and it takes less than one second to load the requested website and to show the matching ad box. In less than a second the system analyses the content of a given webpage, locates and chooses the matching ads, checks against the set CPC rates, saves the stats and sends information about a matching product along with its photo. We have made this possible by designing a sophisticated and dedicated software architecture through the use of nonrelational MongoDB database.

Selected clients of Dotcom River