Grow media company revenue with data, AI
The 8th Big Data & AI for Media Conference, organised and hosted by World Newsmedia Network in association with FIPP, takes place on 1-2 November at Microsoft’s headquarters in London.
Confirmed speakers include data and revenue executives from BBC, Bibblio, CNN, Condé Nast, Mittmedia, Expressen, United Robots, Financial Times, Neue Zürcher Zeitung, Pangaea Alliance, Civil and Styria Media.
How would you describe the skills makeup and scope of your data team?
Our data team underpins the company, since Bibblio’s value is derived entirely on data. The key team members are our head of product, Nic Young, and lead data scientist, Dr. Mahbub Gani, who both have extensive experience with semantic data and machine learning from previous roles and lectureships. His main role is leading our research and implementation of the recommender systems algorithms that power Bibblio’s SaaS service.
What are the most important components of your data and AI strategy?
As a company that’s built on using AI in conjunction with content and data from customers to deliver value for them, everything is essentially data and AI strategy to us. It’s probably fair to reframe this slightly and look at how we’re an important component of our clients’ overall data and AI strategy. We help media companies increase the value they get from their content and visitors by finding connections between content and user behaviour to power engaging recommendations and drive better economic outcomes. At the same time we allow them to simplify their own technology stack by delivering this as a service.
How is data used to drive revenue?
AI applied to the data we have about our clients’ content, audiences and site interactions is what allows us to create engaging recommendations that clients want to pay for, so without intelligently analysed data there wouldn’t be a Bibblio. We also generate data for media companies in the form of contextualised information about their visitors’ interactions with our recommendations.
If you were advising other data directors on how to build their operations, what key lessons would you impart?
Think outcome and business case first, not technology and algorithms. A large part of getting value from data is about making sure it’s the correct data, in a form that’s actually useful to you. Also, look for commoditization when possible and take the easiest route to results. You don’t need to build or re-invent natural language processing to benefit from it – and the simplest, well-proven, algorithms are often the most effective.
How does audience data factor into product development and improvements at Bibblio?
Audience data is everything to us: without it we can’t understand the reaction of clients’ visitors to our recommendations, so we’re always incorporating insights from audience data into how we evaluate and develop the product.
In your opinion, what does the future hold for the business use of data and artificial intelligence in media?
Nowadays it’s essential for media companies to understand their audiences; otherwise they can’t build the content, trust and relationships that create loyalty and revenue going forward. To YouTube, Netflix, Facebook and Spotify recommender systems are responsible for +80 per cent of all engagement time. AI is crucial to that task, because it’s by far the best scalable way to make better predictive models for each individual user; in fact that’s essentially all that these algorithms are doing. Even with regulatory roadblocks like GDPR and concerns around filter bubbles and fake news, data and AI are going to be one of the essential pillars on which media businesses rest in the future. We just have to learn to harness it well, not blindly run away or embrace it. Either can be disastrous in their own right.
Join us at the Data & AI Media Week events in autumn 2018. The 8th Week of events is in London and is themed Data- and AI-driven Revenue and Audience Development for Media. The 9th week of events is in Amsterdam and is themed Product Development for Media Companies, using AI and data.
More like this