If Twitter wants to be a media company, it needs to get a lot better at relevance

We’ve been arguing for some time now that Twitter is becoming a media entity in its own right, and some of the company’s moves around the Summer Olympics and other events have helped flesh out that theory. John Battelle of Federated Media argues much the same thing in a new blog post — that Twitter wants to become a media company, and that doing so means curating and even creating or “co-creating” content for its users. While this is undoubtedly true, Twitter is going to need to become a lot better at relevance and discovery if it really wants to be a new-media player.
In his post, Battelle describes how his thinking has been influenced by some of the recent offerings Twitter has come up with around broadcast events — such as the “Oscars Index,” (a partnership with Topsy) which tracks the sentiment around the Oscar-nominated movies and personalities leading up to the Academy Awards by analyzing tweets about them. Although he doesn’t mention it, Twitter also recently announced an even more ambitious effort to create a verified “Twitter TV Rating” for TV shows as part of a partnership with media-metrics company Nielsen.
Twitter Oscars

Having lots of data is great — relevance is better

The Federated Media founder points out quite rightly that one of the things that makes Twitter a potential gold mine — both for media companies and for advertisers — is the number of signals that the network can accumulate about users, their behavior and their interests. More than half a billion tweets a day is a lot of data, and somewhere in the midst of that are the keys to delivering better content, and better advertising (which is increasingly becoming the same thing, a topic we’ll be discussing at paidContent Live in New York on April 17). As Battelle puts it in his post:

“Twitter presents a massive search problem/opportunity. For example, Twitter’s gotten better and better at what’s called “entity extraction” – identifying a person, place, or thing, then associating behaviors and attributes around that thing… Real time entity extraction crossed with signals like those described above is the Holy Grail.”

This is fundamentally the same goal that both Google (s goog) and Facebook (s fb) are focused on as well: how do you show users only things that are relevant to them, and hide those that aren’t — in real time? Facebook has gotten criticism for the way it tweaks the news feed based on its algorithms, but the reality is that most users don’t want to see everything that streams through their networks. And Google started its Google+ social network, and built it into everything it does, in part because it needs more data signals about its users.
The problem for all of these companies is that doing this is really, really hard — every user’s stream consists of billions of data signals, and deciphering which are meaningful and which aren’t is a complicated business. To get a sense of how difficult it is, all you have to do is look at Twitter’s “Discover” tab, or the “Trends” listings, or look at the promoted tweets and promoted trends that show up in your stream (of course, Facebook is almost as bad with its sponsored pages, and it has orders of magnitude more data).

Relevance is the key to digital media

Twitter Icons
As Battelle notes, Twitter has gotten better at discovery, and the revamped version of its Discover tab is better than it used to be — and so are the suggestions Twitter sends to users for other accounts they should follow. But the Discover tab in particular is still light-years away from where it needs to be in order for it to be a compelling content-discovery mechanism for users, and the curated email newsletter Twitter sends out is even worse: it shows me Canadian news because I live in Toronto, even though it knows (or should) that I rarely ever tweet about that topic.
Simply put, relevance is the key attribute for any digital-media entity in the 21st century. Newspapers and other traditional sources of content are terrible at suggesting or curating relevant content for individual readers, but no one really expects them to be any good at it — they have zero experience in doing that. Twitter, however, has enough data that it arguably should be much better than it is. And it needs to get there quickly, before Google or Facebook (or god forbid, even Yahoo) get much better.
Coming up with visualizations about the Oscars or highlighting tweets about NASCAR may be useful for reeling in big media brands, but users are going to need a little more than that before they trust Twitter to curate content for them.
Post and thumbnail images courtesy of Shutterstock / noporn and John Battelle