DailyMe: Seeing How You Are Being Seen
Once you register for the site and start tooling around articles, your recent behaviors are outline and categorized by news entities in the new Newstogram feature. The DailyMe system extracts from the articles news categories (economy, construction, politics, etc.), people, companies, organizations and more to present a set of bar charts illustrating your relative interests. The user then can click on any of the granular sub-segments and choose to track that topic in a My News box in the upper right of every screen.
"We let you view how the system views you," says founder and CEO Eduardo Hauser. The model lets you use your own recent behaviors to act as a navigation tool and to shape your own profile. In the Newstogram graph, the entities reflect inferred interests according to your recent travels; users can tell the system to "Track It" and elevate the segment or topics into a declared interest.
The user can click on any of these My News categories and get a dynamically created page around that segment. Ultimately, all behaviors and interest also morph in the front page of the site. Beneath a set of features all users see, the rest of the page is occupied by "Headlines for You," which reflect your interests based on behaviors.
The interesting part about DailyMe is that the underlying technology is being offered as a set of tools that third -party publishers can productize in a number of ways.
The publisher gets a real time view on the specific topics, people, companies, in their content that are most popular at a given moment and trending. "For partners, we are advocating as a first step installing the code and seeing if the data that comes out of that code is of any value," says Hauser. "Then we can show you derivative products from that data."
The Newstogram profiler can be used at another site or the data simply can be used to create various related news, recommended news and dynamically created topic focus pages for an individual. While others offer similar services, Hauser says that the DailyMe approach is tuned specifically for news and information sites, as opposed to e-commerce. As well, he highlights this unique combination of entity extraction from articles layered in with behavioral tracking and relevance ranking.
As a tool for ad targeting, DailyMe offers partners an API that can be woven into calls to the ad server. The publisher can pass along to its ad server either the contextual information about that page or the behavioral information about that user. The publisher pays a 2-cent fee per API call. Otherwise, a partner can elect to let DailyMe run targeted ads into recommendation or Newstogram modules that run on site and split the revenue.
It remains to be seen how much more effectively the DailyMe engine works to target ads or content than existing approaches. News content can be difficult to target ads into aside for broadly defined categories. Issues of sensitivity abound, and the content of a news article can be so far-reaching, the possibility of false or inappropriate ad targeting arises. It isn't clear to me whether advertisers will be able to leverage at scale the kind of granularity of news segmentation that an engine like this makes possible. Several large scale news publishers are testing the technology, Hauser says.
From a consumer's perspective, the most interesting piece of this effort is the Newstogram tool, although the naming convention (a play on "histogram") is too clever by halves. Nevertheless, it is one of the best executed instances of surfacing for the user the ways in which a behavioral tracking system is seeing his activities. Perhaps he or she won't want to tinker with his profile. Perhaps we don't want this level of control over our online profile. What is important is that it is there and visible.
If users have this kind of access to how the machine is viewing them for content and ad targeting purposes, it peels one more layer of mystery and suspicion away. What is interesting about marrying the ad and content targeting approach is that the benefit: In a unified system of content and ad targeting, marketers and technologist no longer have to make the strained argument that BT improves their online experience solely by reducing irrelevant ads. I have to wonder if users would be more amenable to ad targeting if they see the same technology clearly benefiting their overall content experience.
Transparency alone is not the right answer to the quandary over privacy and targeting. The users must not only feel in control but be able to see a real benefit from the technology. DailyMe's approach is among the more promising examples of being both transparent and communicative.