“What you want actually is small data, you want insights.”
"Mesure the right data, and learn the right things."
“Engagement doesn’t mean performance."
“There is virtually no correlation with how many people engage with an article and tweet it with who the article is read by.”
“If you see an article that creates a lot of anger, it is a good thing to do.”
“You can build models to say ‘this is the actual article you want to see.’”
“Data products are more or less those free things.”
“In news industry, recommending articles is a big deal.”
“To use tweets as a measure for recommendation is not a good thing.”
Lutz Finger is a director at LinkedIn. He is an authority on social media as well as text analytics. He’s also a co-founder and former CEO of Fisheye Analytics, a media dat-mining company whose products support governments and various NGOs. Finger is highly regarded technology executive who built a sales center for Dell Europe as well as an incubator for mobile applications at Ericson. He is a popular public speaker on business analytics and serves as an advisor and board member at several data-centric corporations in Europe and in the United States.
In a pre-World Congress survey, INMA community members said “data analytics” is the third most important issue at their media companies. Making data actionable was a recurring topic during the final day of INMA’s 2014 World Congress.
Lutz Finger, author of Ask, Measure, Learn and co-founder of Fisheye Analytics, conducted an experiment with the conference’s audience. By using LinkedIn, Finger gained knowledge about the attendees from their profiles.
“Now I have some data,” he said.
Finger’s data showed statistics he gathered from the conference audience, such as earliest adopter of LinkedIn, most connected person, and most popular person.
“It means you get the most page views and you are most looked at,” Finger said about INMA president Yasmin Namini. The collected data showed she was the most popular profile in the audience.
Finger showed the importance of data in the form of a graphic: a series of connected dots, representing the LinkedIn profile connections in the room: “The bigger the dot, the more connection it has.”
The data was able to reveal characteristics about the group as a whole.
“This is very knit community, very tight community,” he said. “Most of the people are connected.”
The data went further as to show the No. 1 skill people possess in the room, which was graphic design. “Media and Entertainment” was the industry that the majority of people were affiliated with.
With a heap of data, it is helpful to move toward a smaller part of it. With focus on a small set of data, a news media company can gain insights about content consumption habits.
Finger has a plan to decipher the data: “Ask, Measure, Learn is the framework.” This structure allows an interpreter to understand data fully, leading to a better strategy for engaging with consumers.
“The ‘ask the right question’ is the hardest part,” he said.
The measurement portion can be difficult because of the large amount of data that has been gathered, but also can be due to a misunderstanding what good data measurement is.
With more than 50% of online traffic being attributed to bots, Finger said media companies cannot always rely on views and clicks to judge engagement. These numbers do not always compare to the actual reader.
“There is virtually no correlation with how many people engage with an article and tweet it with who the article is read by,” he said.
Being able to understand the consumption patterns of individual readers aids future engagement strategies: “You can build models to say ‘this is the actual article you want to see.’”
Different types of data products to encourage future engagement include benchmark, prediction, and recommendations and filters. Benchmark is the comparison of one item to another, as in articles, and is a simple concept to grasp.
“It’s very easy. We learned this in school,” Finger said.
Prediction is a data product in which predictions are made about what a person will want to purchase or read based on their past behaviour. Target takes part in this with its loyalty card programme, but the data product is not widely used.
“You don’t see a lot of examples on prediction,” he said.
Recommendations and filters show suggestions based on past behaviours also, and articles are chosen and shown based on recommendations. In the news business, however, people may be wondering about these data products and their usage.
“How is this relevant to me? I’m not selling products,” he said.
Explaining their purpose in this business, Finger said such data can be helpful when thinking about ad placements and to understand the habits of the audience, further driving better engagement.
This strategy can be utilised by following five steps:
- Centralise data.
- Build a content profile for each reader to help track how they act.
- Create a knowledge profile for potential journalists.
- Hire a data team.
- Invision data products.
“When you have done the first three steps, you do have Big Data,” Finger said.