Once upon a time there was a semantic web startup dubbed Knewco, touting a knowledge discovery/ contextual advertising system for health care and life sciences content providers and advertisers (see this story). After a long journey toiling in the land of targeted health sites, it realized these prospects didn’t provide the best opportunity for its micro-targeted ad strategy. So, with a restructured management team and some fresh capital in place, in the latter half of 2011 it began a new quest: To become the semantic platform for premium publishers to help deliver personalized content recommendations and ads to readers and get more revenue, traffic and stickiness in the process.
Renamed Personalized Media, and now headed by CEO Rajiv Salimath, principal back in the Knewco days, the company’s technology now is in an advanced testing stage with premium publishers who will be white-labeling the system. Salimath can’t disclose their names, but suffice it to say that you’ve seen some of them prominently mentioned in these virtual pages before. So, what does Personalized Media bring to the party that’s attractive to some of these sites that already have taken steps – sometimes big ones – to bring semantic intelligence to their web presence?
From There to Here
When The Semantic Web Blog spoke with Salimath in late 2009, he discussed the technology’s prowess at understanding concepts and inter-relating them to other ideas. In its current incarnation, Personalized Media’s semantic search algorithms make it possible to find and suggest to readers other content – text, video, even apps, from the source itself, its associate properties or elsewhere on the web, depending on publisher preferences – that’s relevant to any word or term they highlight, or to the page at large. The content appears in a bubble at the user’s click.
The idea is to let users pull content that relates to their interest, and to let users initiate sponsored media and ad activity from Personalized Media's network partners that’s also tied to concepts that interest them. The goal is to deliver the relevant content and apps to the user within the page, via the bubble, so that they don’t go wandering off to other sites in search of related information. The company says that most premium publishers lose 30 to 50 percent of their traffic to search engines.
“The Personalized Media product brings all the other pieces of content that might impress you – other articles, applications, videos – anything to help you understand the concepts,” Salimath says, without getting into cookie-tracking or other privacy-gotchas.
Where the Real Challenge Lies
Understanding concepts semantically is not the hard part, however, he says. In fact, it’s a nut already cracked by some of the premium publishers on its client list. Understanding the concepts and tying them to interests is the bigger challenge that many publishers haven’t solved on their own. There’s more to personalization than serving up more and more content about diabetes medications, for instance, to a reader who was perusing a story on the disease, or who may once have done so. At some point, that reader will want to move along, so it’s better to try and figure out what other interests someone who's explored a story about the disease in the health section might have – maybe, for example, low-carb recipes is one of them.
There’s not an obvious relationship there, but by mining lots of different semantic data stores and doing a lot of curation on top of that, you can discover that people who are looking at diabetes information would be interested in such recipes, Salimath says. “That kind of mapping is one thing that is unique and part of the secret sauce.”
The other capability Personalized Media brings to the table is having partnerships in place with providers whose content can be leveraged to relate to concepts and interests. “So we can bring text content or video applications and customize them all to be within the bubble for a nice user experience,” he says. “We have hundreds of different content APIs [for this]. It would take a [premium publisher] a lot of time to negotiate these content relationships themselves. It’s easier for them to just work with us, and we constantly keep adding new content sources and figuring out the right relationships between those sources.”
Spotlight on Ad Revenue
Personalized Media’s technology also can be used by publishers for maximizing the value of their premium content. For instance, house ads can be more informed. Intelligent house ad units can offer related articles and other content that could be interesting to someone reading, say, a piece about healthy desserts in the recipe section, to drive that person to higher-value pages – content about controlling cholesterol in the health section, perhaps, where the CPM tends to be higher. “Publishers struggle with different advertising models…, and they keep coming to the same point. There are some premium areas with premium content to sell high value ads, and they want to push users from lower-value to higher- value pages without the experience feeling jarring or artificial to them,” Salimath says.
The technology errs first in favor of prioritizing what is more relevant and interesting for the user more than for the value of the page. But assuming a lower-value and higher-value page have equally interesting content the latter gets preference.
As Salimath sees it, premium publishers want to get better at monetizing their own digital content (see this story about some the difficulties they’ve had trying). “Publishers have become very wary [working with third party networks] because they take those users and there’s a lot of data leakage,” he says. “The premium publishers we’ve talked to all say they want to control all the data about their users, and they want to figure out how to monetize that. We’re responding to that." With its platform, publishers can directly sell contextually-relevant ad inventory.
"This is such a big opportunity and such a big problem. I wish we had the foresight two years ago to do this and not chase all the smaller health sites,” Salimath says. Helping publishers solve the challenge could make for a happily-ever-after all around.