4 Reasons Personalization is Hard to Do
Publishers are using personalization to drive a variety of editorial, sales, and audience development goals. But challenges still remain to make personalization as effective as possible and drive the results publishers want. Leaders from Hearst Magazines, Haymarket Media, and ALM cited a number of issues that their organizations needed to tackle in order to launch personalization initiatives. The executives spoke during a keynote panel at FUSE Forum: The Pursuit of Personalization in June. Following are four challenges they encountered as they launched personalization campaigns and how their teams surmounted them.
1. Building the “Golden Record”
Panelist Gene Bishop, VP of technology at ALM said building a “golden record” of the audience is critical to personalization. While the B2B publisher had massive amounts of data about its audience, pulling that together onto a single database and then extracting real insights from that database proved challenging. “We want to build what we call this ‘golden record’ of who the customer is and then use that in applications like Exact Target and Journey Builder,” said Bishop. “You can watch what your customers are doing and even develop new products as a result.”
To do this, ALM invested in a NoSQL database that brought together all of its data from a variety of platforms. The two-year process combined ALM’s data onto one platform, but it remained difficult to analyze this data and pull meaningful audience insights from the database. Bishop said ALM had focused so much on building the software that it had invested little time in considering the operations side and the people who needed to use this application. As a result, ALM is considering working with an outside provider to analyze its audience data and create a master record so that its teams can more easily glean insights about individuals audience members.
Michael Dugan, advisor to Hearst Magazines, agreed that analyzing and understanding the data is the real challenge. “You all probably have the data you need. It’s just [a question of] are you looking at it? Are you assembling it? Are you running it through the right platforms?”
2. Personalization Changes the Sales Dynamic
Joshua Storch, senior director of email and data operations, said Haymarket Media is planning to use personalization to sell more valuable, targeted ads. “I think we’re ready to pivot the conversation so that a client doesn’t do run-of-site or an entire section or block of our audience,” said Storch. “We want to get our salespeople out there and convince these clients that you don’t need to pay X dollars to reach an audience of 100,000. You pay 2X dollars to reach an audience the quarter of the size, but we’ll get you more revenue because you have the right people looking at your ads.”
However, in order to have this conversation, salespeople need to be trained in a new way of selling, said Storch. Currently Haymarket is training its sales staff on how to use data insights to sell personalized ads to clients. Bishop agreed that data fluency among sales staff is crucial. “It’s going to be a dynamic shift in the way that salespeople have to sell,” he said.
3. Identifying the Right Signals for Personalization
Dugan said an effective personalization algorithm, that can recommend the right ads, stories, or subscription offers, needs to capture the right signals for audience interest. At Hearst, Dugan found geographic location was a surprisingly useful signal. “We found that regionally, people tend to consume a lot of the same content,” said Dugan. “We generated a little fly out unit that would kind of steer the visitor to the next article or piece of content. We drove clicks through that unit 25% higher than the human program.”
But Hearst soon realized that just because readers clicked on a story recommendation, didn't mean they stayed. They found that exit rates were high on recommended pieces. So Dugan and his team adjusted the algorithm to take into account other signals that indicate deep reader engagement like scroll, scroll speed, and stray mouse clicks. Dugan recommended publishers use these types of signals to identify user interest, not just clicks.
4. A Technology Solution Is Only Valuable If People Use It
All three panelists agreed that getting employee buy-in across editorial, sales, and marketing is necessary for the success of their personalization efforts. And getting colleagues to see the hidden value of personalization is an important task for technology gatekeepers. “The hurdles are actually the people,” said Storch. “You have to show them that it’s not just about hitting revenue numbers today. It’s about the future of your database -- it’s about reducing opt-outs. You may not see revenue immediately, but as we transform the way we communicate with our audience, as we give them messages that they want, we’ll see that revenue come back.”
Dugan added that showing the numbers, however logical they might be, won’t always earn buy-in for a new tool. Technology leaders must navigate the people and culture of an organization as well. “Don’t approach it from strictly a data-driven solution,” he said. “There is always going to be a human element…Be prepared to compromise in some areas that either don’t feel right to certain parts of the organization or don’t reflect the brand in ways that they want. Be careful to navigate that or people will quickly cool on your efforts. Consensus is vital.”