Twitter Needs to Fix Its DM Search

Thankfully, the company knows it

Photo: Chesnot/Getty Images

As a freelance journalist, I rely a lot on Twitter direct messages. I use them to interview sources, chat with acquaintances, network, and gossip. And my DM habits are nothing compared with some journalists I know, who essentially lean on the product as an addendum to email and Slack. Twitter DMs, for a lot of white-collar media workers, are a further diffusion of how we communicate with each other. We are surrounded by messages, coming and going from seemingly a thousand different locations.

But Twitter DMs are missing one critical feature: search. Or, more accurately, content search. In 2019, the company released a DM “search” that can only find users. I can search for messages from “Damon Beres,” but I can’t search for the content of those messages.

According to computer scientists and UX/UI designers I talked to, while public-facing search products like Google and DuckDuckGo have gotten smarter and more accurate in recent years, private search is a different animal. This may be part of the reason why Twitter hasn’t yet implemented a more comprehensive DM search: It’s a labor-intensive endeavor for the company when, clearly, it has plenty to focus on with its public-facing product.

Twitter told me it’s working to improve its DM features by May.

But I believe that for a communication company to truly become invaluable to its users, it needs to allow for full-featured private messaging, and that includes search. If I can’t easily pull up a month-old message from a friend in which she sends me her mailing address, for example, I’m not going to have important conversations there. And if I’m wary of holding important conversations via DMs, it’s never going to be anything more than a shallow and cursory way for me and many others to send articles and tweets back and forth forever.

Twitter told me it’s working to improve its DM features by May, and a member of their product team recently tweeted asking what kind of features people thought would make DM search better. Nearly 15 years after Twitter was born, it’s finally considering the weighty endeavor of improving DM search beyond its paltry update in 2019, when it finally allowed people to search for messages from specific users.

While the company works on tweaking DM search, it’s worth considering why this feature is so valuable in the first place.

Consider how search has become an essential element of email. In his 2012 study on email behavior, University of Sheffield data scientist Morgan Harvey, PhD, writes that previous research shows “that email messages tend to be searched for more often than any other kind of media, including visited web pages.” His study, which looked at email search behavior found that the majority of email searches involve references to people; it also found that searches tend to repeatedly reference specific people or topics but that the searcher isn’t always clicking on the same specific message with each search. In other words, maybe I will search “Damon” or “pitches” each week in my email, but I’m typically clicking on a different email with each search despite relying on the same term to start the search.

We’re not poking around our inboxes for fun or to learn something new. We’re in there because we need to get stuff done.

A 2017 study found that, not surprisingly, search is becoming more and more important the longer people use it and the more messages continue to pile up in the inbox. It found that, unlike with web search, 84% of email searches are for something very specific. For example, I often search my email for “UPS My Choice” or “HarperCollins,” but I’m not always looking for the same message from UPS My Choice. (I get a lot of them!) And while people often conduct web searches to explore and discover new information, according to the 2017 study, people typically conduct email searches with a specific email in mind. We’re not poking around our inboxes for fun or to learn something new. We’re in there because we need to get stuff done.

Of course, there aren’t any studies on Twitter direct message search specifically, and it’s worth noting that the way people use email and DMs is likely to be at least a little bit different. But I suspect that messaging platforms, such as Slack, Facebook Messenger, iMessage, and Twitter DMs, are increasingly encroaching upon the role email plays in our professional lives. I spend way more time messaging people via chat platforms now than I did five years ago and not just because I’m communicating online more (though that is certainly a factor). Messaging platforms, quite simply, are becoming more critical forms of communication. How often do you email your friend versus text her or DM her? What about a brief message with a colleague at work?

Still, as noted in a 2016 paper from the Association of Information Science and Technology, email is easier and more dependable than other messaging platforms in large part because of the search functionalities email has perfected while other platforms have not. But as more and more people migrate to other platforms, search will become even more critical in allowing users to take full advantage of these other messaging channels. If I can search my email or iMessage, I’m going to prefer to have conversations with friends and co-workers via email or iMessage than via Twitter or Facebook, where texts may as well be disappearing into the ether without the functional and privacy-related benefits of actually having them disappear into the ether, ala Signal.

But search is difficult and expensive to implement, which probably explains why it’s taking Twitter so long to get around to it. Search, in particular internal search, is “not easy to do,” says Kevin Park, an assistant professor of UX/UI design and development at the New York Institute of Technology. “It does cost a lot of resources to create an internal search function if you’re not relying on Google.”

While any number of public-facing search features on a website can rely on Google’s Programmable Search Engine to power search of all the content on that specific website (such as when you’re searching for a specific recipe on your favorite cooking blog), private message search generally can’t do that.

That means Twitter needs to build its own search feature from scratch — a difficult task when the platform has little data to feed its own algorithm. Think of it this way: Google powers its search by cataloging all the public-facing data it can find on the web. Every website, every sentence becomes privy to Google’s “spiders” that “crawl” the web, grabbing and storing all that information and using it to understand what kind of information is kept where and what it’s typically contextually related to. The more data it can get its creepy-crawlers on, the better it’ll be able to find and deliver search results when you want to look up “best brioche recipes.”

Google also uses the rest of a user’s web-browsing data — what sites the user frequents, their previous searches, what results they end up clicking on — to further perfect the results. This is why if I look up “bagel shop,” Google will show me plenty of results of bagel shops in Brooklyn. It has figured out that I live in Brooklyn, so that’s probably where I’m going to want a bagel.

But private messaging platforms don’t have this gigantic treasure trove of data to rely upon, making it considerably more difficult to implement search than for public-facing data.

“Google has billions of web pages and billions of previous clicks it can learn from, whereas if you’re trying to build a search index for a single user’s filesystem, there’s not nearly as much to go on,” says Harvey. Not to mention, he adds, “Twitter is not noted for having particularly good search in general.”

There also might be a demand issue. Twitter DM is limited in scope; therefore, people simply might not bother to invest too much energy in using it at all. Not everyone is a journalist, and not everyone needs to comb their DMs trying to find the relevant essay an acquaintance sent over three weeks back. But if we could, it might encourage people to rely on Twitter more for their communication needs (just look at how deeply ingrained Facebook Messenger has become!). And that investment would pay off for users and Twitter.

I’m a columnist for OneZero, where I write about the intersection of health & tech. Also seen at Elemental, The Atlantic, VICE, and Vox. Brooklyn, NY.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store