(Bloomberg Opinion) -- A few weeks ago, Twitter gave users an early holiday present in the form of "Topics," a feature that lets them subscribe to feeds on a given subject. As the company explained it, the feature reduces the need for users to track down accounts to follow; rather, “the conversation will come to you.” Although Topics sound like it might be a boon for users, it’s a barely veiled attempt to improve the platform’s ad targeting. And Topics shares the deficiencies of the rest of Twitter’s model, so it probably won’t work out that well.
As I've written before, Twitter faces a fundamental problem in leveraging its user data. Because the platform encourages brief and superficial interactions, it’s hard for the company to evaluate content quality, much less learn about specific users’ preferences and interests. That’s bad for Twitter's top line, which depends on ad sales. If the company’s doesn’t know what you like, it has trouble inferring which forms of advertising you’re most likely to engage with -- and that means that the opportunity to advertise on the platform isn’t worth that much.
That’s where Twitter is hoping that Topics will help: Before Topics, Twitter couldn’t tell an advertiser with certainty whether you like a given subject. Now, if you subscribe to the relevant Topic, then the platform can at least claim it knows you’re interested.
Twitter is quite transparent about this, explaining that when you follow a Topic, “You’ll see Tweets, people, events and ads” about it (emphasis added). And many of the Topics that are available for now are tied to specific ad categories.
I appreciate and almost admire the company’s audacity in explicitly asking users to indicate which ads to show. But at the same time, the need to do this exposes the true conundrum of Twitter’s current model. We all knew that Twitter’s ad matching was bad, but who knew it was so bad that the company would have to beg users to help target themselves?(1)
And there’s a deeper problem with Topics: Twitter’s fundamental inability to evaluate content makes it hard to curate feeds, and as a result the available Topics – at least so far – seem pretty unfiltered and low quality. The "Star Wars" Topic, for example, is full of "Rise of Skywalker" spoilers and rants. The “Video Games” Topic shows a bunch of random people’s Pokémon Go screenshots, plus a large number of holiday posts from a single game producer -- presumably the one that has bought the most ads (and may have even paid for the privilege of being featured above its competitors).
As a result, users don't have much incentive to seek out Topics directly. Making the problem worse, Twitter’s inability to target content applies to Topics as well. (For example, the platform keeps asking me if I want to subscribe to the “Astrology” Topic. (3)Presumably that's because I periodically tweet about Libra – but when I do so, I'm talking about the blockchain technology, not the zodiac sign!)(6)
Twitter has also made a seemingly unforced implementation error: the list of Topics you subscribe to is public, just like the list of profiles you follow. That's counterproductive because you might have interests -- ad-relevant ones -- that you'd be happy to note in private but wouldn't necessarily want to broadcast in public. ("Bad '80s music," anyone?)
Twitter had more than $3 billion in revenue last year, most of it from ads, so it’s hardly a basket case. But at the same time, its ad business has been struggling, with some commentators predicting a bleak outlook for 2020. What's more, almost two-thirds of Wall Street research analysts have "hold" ratings on Twitter’s shares -- not exactly a vote of confidence in the company's prospects.
I’d like to see Twitter find ways to move beyond an ad-based business model. But if the company really wants to make selling ads work in the long run, it has to encourage more meaningful user experiences from which it can infer more valuable data. That probably means doing a better job of curating content and encouraging more private interactions on the platform. Topics, at least in its current incarnation, doesn't seem like the solution.
(1) At some level, this is really stunning. I would have thought, for example, that users' on-platform search behavior would be enough to reveal the same amount of information -- anything a user searches for more than a couple times is probably a true interest. But maybe average Twitter users search sparingly, preferring instead just to scroll through endless feeds?
(2) For what it's worth, knowing who subscribes to "Astrology" might indeed be high-value for advertisers.
(3) I am an economic adviser to Calibra Inc., a subsidiary of Facebook Inc. that is one of the founding members of the Libra Association.
To contact the author of this story: Scott Duke Kominers at firstname.lastname@example.org
To contact the editor responsible for this story: James Greiff at email@example.com
This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.
Scott Duke Kominers is the MBA Class of 1960 Associate Professor of Business Administration at Harvard Business School, and a faculty affiliate of the Harvard Department of Economics. Previously, he was a junior fellow at the Harvard Society of Fellows and the inaugural research scholar at the Becker Friedman Institute for Research in Economics at the University of Chicago.
For more articles like this, please visit us at bloomberg.com/opinion
©2019 Bloomberg L.P.