Grupa Onet: Success of a 2018 Global BIGGIES Award winner “Data-Driven Process Revolution” was in the details

Big Data & AI for Media:
Define the problem to be solved or technology disruption to be addressed 

Bartek Zarebski:
There are three key problems which drove us to create the data-driven process of publishing.

• Previous drivers of the digital publishing business growth in Poland stopped working after 15 years, such as: organic growth of reach of the Internet audience, digitalization of the content niches, mastery in content distribution via home page of news service.
• Our traditional model of the business upscaling – the more page views equals the more revenues – failed.
  At that time, our publishers produced more page views but we made less in revenues, didn’t know why.
• There’s been a change of the clients’ criteria of advertising purchase from quantity of ad inventory to quality of ad inventory becouse of programmatic transparency.

In response to these changes, we have created several solutions by leveraging our data.
Our data-driven solutions are:

• We changed the KPIs of the entire value chain delivery
  (editorial and product) from page views (PV) to:
  • IF (inventory factor) – the number of ad queries to adserver, produced by 1PV or video view.
  • RpQ (revenue per ad query) – metric of effeciency of the produced inventory monetization
  • Engaging page views – the number of PV engaging user for more than 5 sec
• We established process of inventory management with the goal: to produce ad inventory, in amount and quality, required to deliver the assumed budget targets
• We deliver to the business predictive analytics in order to find and close the gap between the market potential and potential of ad inventory production, across entire portfolio of brands.

Big Data & AI for Media:
What is the time frame for project, and milestones that took place, including details on planning, testing, experiments, rollout, review and improvements.

Bartek Zarebski:
The project took 18 months of development, including the following milestones:

• Standarization of traffic measurement across entire portflio, including new metrics required by the model of predictive analytics
• Standarization of the inventory production (coherent ad stack across entire portfolio)
• Agregation of data of traffic measurement system, CMS, CRM, ad servers in one data warehouse
• Structurisation of agreagated data to the common for entire portfolio structure, reflecting profit centers, types of content, distribution, ad market
• Value chain mapping – based on data, description of business relations among: content>traffic>inventory>revenues
• Development of tools supporting insights for business, editorial and product in order to delivery new KPIs (above)

Of the 18 months of development, we took six months to implement the changes, including:

• Editorial teams training
• Plan of product development supporting new goals
• The budget adjustment according to real potential of ad inventory production
• Launch of the inventory mgmt process in entire portfolio at once
• Monthly revisions of gap between the budget assumptions and potential of the inventory production
• Risk management – how to close the gap

Big Data & AI for Media:
Please describe of the resources used for the project, including budget, technologies, team members, skill sets and time spent

Bartek Zarebski:
We used the following technologies for these projects:

• Hadoop, Tableau, private cloud
  (meantime decision of migration to public cloud has been made)
• Internal system of measurement of all required events and its metadata
• Domain systems as sources of data:

– DFP + homemade, native ads, adserver

– Homemade CMS

– Customized CRM Update10 

– SAP

All of these newly implemented technologies required us to cultivate new skills, including deep knowledge on how adservers transform traffic into ad inventory and for what purposes. We also needed to add several highly skilled people, including a big data architect, data flow architect, business data architect, analysts, business process engineer, software developers.

In paralel, about 15-25 people worked on the project in different configuration of teams and skills, depending on the project’s stage.

Big Data & AI for Media:
Please describe challenges with management, resources, audience, technologies, etc.

Bartek Zarebski:
We faced many challenges to overcome in order to affect these changes, including:

• Lack of common structure of data
• Lack of common definitions of business events among data sources
• Lack of common understanding of the definitions in different departaments (editorial, financial, sales etc)
• Lack of overall perception of entire value chain – silosing issue
• Top management lacked digital experience, overall perception of the complexity of the value chain, expecting immediate business effects

There is only way to solve all these challenges: the best in class project management, engaging all contributors and all stakeholders, supported by agile culture managing expectations

Big Data & AI for Media:
Please describe specific progress made with the project, such as increased revenue, engagement, audience growth, teamwork, subscription increase, etc.

Bartek Zarebski:
We have achieved many positive effects from these data-driven changes, including:

• Reduction of unnecessary page views by 30% (cost reduction)
• Despite the page views reduction, the number of total ad inventory increased by +20%
• Viewability of ad inventory increased by 25%-55% (depends on type of ad inventory)
• Sell-out increased by 15 %
• 1PV gain more revenues by +20% average

What is the most important, nowadays we overdeliver the budget assumptions.

Big Data & AI for Media:
Please identify long-term issues, such as plans for growth, change of technologies, projection for revenues or audience, impact on other projects, impact on company’s resources, management, strategy and/or resources.

Bartek Zarebski:
Our next challenges are growth of eCPM, Time spent on content by users, and transformation from manual to automized content distribution in personalized way. We believe that these factors will support our business KPIs the most.
We also aim to centralize our data management competencies BUT decentralize our insights detection and its business implementation.

Big Data & AI for Media:
Can you summarize your recommendations for other media companies hoping to duplicate your success and avoid challenges you may have faced?

Bartek Zarebski:
Such a project requires a cross -functional core team composed of members of different departments of the company.
Before you start, make sure that the goal and reasons are clear for all participants and top management.
Support of top management is neccessary.
This is a cultural change for the media company, not only KPIs.

Enter the 2018 BIGGIES EMEA competition, with 12 categories focused on Big Data- and AI-driven products and processes at media companies. Deadline is Sept. 7. Enter now!