Hürriyet, established in 1948, is the leading daily newspaper of Turkey. With its news portal, content verticals (finance, lifestyle), and classified verticals (real estate, cars), Hürriyet receives more than 20 million unique visitors and two billion pageviews monthly.

Considering the size of its audience and huge traffic numbers, by the end of year 2012, executives at Hürriyet decided it was time to start a Big Data project with several goals in mind: to know, to engage, and to provide better experience, content, and services to its customers. 

Although the project is just one year old, we have already seen solid outcomes, which are encouraging us to take further steps.

Multi-channel CRM system and rich data: Hürriyet has more than 30 diverse data sources, consisting of demographical and behavioural user data from vertical Internet portals such as hurriyet.com.tr (news), Bigpara.com (finance), Mahmure.com (women and lifestyle), and consumer data acquired from print newspaper promotion programs or campaigns. 

Recently, all data sources have been consolidated into a centralised CRM system to manage customer information files, multi-channel communications, offer and campaign executions, communication history management according to permission marketing rules, and contact-point optimisation in in-bound and out-bound call center operations and complaint management.

Segmentation and content/offer/ad interest category scorecard implementations will be the next steps, and at that point, Hürriyet will get the most use out of Big Data. 

Real-time event and action management: Vertical portals have large quantities of  anonymous user traffic. This is an opportunity to understand the behavioural usage patterns and get in real-time personalised contact with the users, depending on contextual and behavioural scenarios. 

Hürriyet has implemented a Complex Event Processing (CEP) infrastructure to analyse and act on the data stream in real time as follows:

• Preventing churn: CEP is monitoring Web traffic to identify the most-exited and most-visited pages. When a user visits a high-score exit page, CEP positions a widget that offers the most-read pages (categorically) that haven’t yet been consumed by the user.
  
  
• Increasing pageviews: CEP is tracking users’ content consumption and records, and which contents are commonly consumed together. If a user consumes specific content for example (let’s call it “Content A”), then CEP offers the contents that are previously consumed with Content A by other users.
  
  Different Content Recommendation (DCR, like collaborative filtering) algorithms are working for different business requirements.
  
  
• Increasing number of registered users and engagement: CEP is tracking the keywords/tags of the contents in which a specific user is interested (tracking is performed for each and every user). If a user consumes contents that have the three same keywords, CEP asks the user if he wants to receive alerts about contents including this keyword.
  
  So each time a new content with the keyword comes up in hurriyet.com.tr, the user would get an e-mail with a link to the content (which requires registration).
  
  
• Transferring Web traffic within our portals: CEP tracks users’ behaviour on hurriyet.com.tr. Based on the users’ contextual and behavioural content consumption, conversion offers for other portals are fine-tuned.
  
  For example, for those users who consume the economy content the most on hurriyet.com.tr, CEP recommends articles selected from financial portal Bigpara.com. CEP also tracks the location of the users and manages to place location-based classified ads.
  
  
• Segment management: CEP is used to track users’ interest categories and segment them according to analytical behaviour models.
  
  For example, a user becomes matched to a segment such as “high interest in health” if there is contextual engagement to health articles. Whenever that user comes to any of the Hürriyet sites, CEP identifies that the user is from that segment and makes a relevant offer.
  
  
• Insights to Web editors: CEP provides real-time traffic data to editors, such as currently most-tweeted content, currently most-shared content, and most-liked content on Facebook.

Loyalty and customer experience monetisation: Having segmented rich consumer data and the means to act on it, Hürriyet now aims to execute loyalty programmes to motivate existing users to engage more and reward them.

Strong relations with the advertisers allow us to monetise the consumer data with strong partnerships via direct CRM campaigns. Hürriyet is now capable of managing behaviour-based banner ads, direct mailings, SMS texting, and personalised offers, etc.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .