Snap Decisions

Snapchat soared because it followed intuition over data. That may also be its downfall.


Photo illustration by Slate. Photo by Lucas Jackson/Reuters.

Evan Spiegel, Snap’s 27-year-old CEO, isn’t in the habit of explaining himself to the public. Until recently, he rarely had to: Snapchat’s wild popularity, especially with teens, spoke for him.

The eight months since Snap went public, however, have seen the company’s best ideas widely copied, its stock tumble, and influential users defect to rival platforms. All of which has forced Spiegel’s hand. On earnings calls and in media interviews, the former enfant terrible has begun trying to tell his company’s story and articulate its vision to an audience that has always struggled to understand it: adults.

That story and vision are interesting not only for what they say about social media, but for what they tell us about the relationship between data, intuition, and creativity in a 21st-century technology company. They tell us that you don’t have to obsessively optimize and crunch numbers like Facebook or Google to build an app that people love. But you just might have to do it if you want to compete with them effectively as a public company.

On the surface, Snap might look like your stereotypical Silicon Valley get-rich-quick story. Three brash dudes in a dorm room hacked together a seemingly frivolous app; it caught fire with the kids; six years later, it’s a $20 billion company rewriting the rules of media, publishing, privacy, and online interaction.

But listening to Spiegel talk about Snap suggests another way of looking at it—one that

sets it apart from its Silicon Valley rivals and holds it in opposition to prevailing winds in technology and industry. He justifies the company’s decision-making by appealing to theory, intuition, and anecdotal observation rather than empirical data. He portrays its sometimes confounding interface as the product of conscious design decisions that prioritize creativity and self-expression. And he explicitly rejects the notion that machine-learning algorithms can replace human editorial judgment.

In a time of ubiquitous rankings, ratings, and likes—when Google ranks search results; Facebook ranks your friends’ posts; Uber and Airbnb rate drivers, riders, hosts, and guests; Yelp rates businesses; and Amazon rates products—Snap’s success speaks to a backlash against the quantification of everything. In the most generous interpretation, it represents a triumph of human intuition, creativity, and whimsy over spreadsheets and algorithms. Less charitably, it might be seen as a victory for flying by the seat of one’s pants.

The good news for Snapchat is that teens love it more than ever. A Piper Jaffray survey released Saturday found that 47 percent of teens named Snapchat as their favorite app. That’s more than Instagram, Facebook, and Twitter combined. Yet one of Facebook’s Snapchat clones, Instagram Stories, needed less than a year to surpass Snapchat in active users. And a recent marketing study of 12 top “influencers” active on both platforms found that they were posting more to Instagram and less to Snapchat.

The question now, with Facebook’s social media empire targeting Snap for destruction, is whether that approach will prove sustainable—or whether Snapchat’s embrace of intuition will disappear as quickly as one of its messages.

* * *

In 2011, there were a lot of companies trying to be the next Facebook—to dethrone the dominant social network the way it had dethroned Myspace a few years earlier. (Google, Twitter, and a relaunched Myspace were among them.) There was also no shortage of startups billing themselves as the anti-Facebook: privacy-focused alternatives such as Diaspora, Path, and These companies saw what had worked for Mark Zuckerberg, and they all wanted to build something fundamentally similar, albeit with some sort of twist: an emphasis on interest groups, a 140-character limit, a mobile-first strategy, an open-source architecture.

As it turned out, the company that posed the greatest threat to Facebook was one whose product looked almost nothing like it. That’s because Snapchat wasn’t, at root, a canny business idea or an idealistic exercise in social entrepreneurship. It was a toy.

The rulebook for social networks at the time said that users were content creators, that they wanted to curate their own images for a broad audience, that they were motivated by likes and shares, and that everything they did was data that could be mined for the benefit of advertisers. There was ample evidence that this was a viable and lucrative model, and there was plenty of demand. All the smart money was on it.

But rather than copy what had worked for others, Stanford frat brothers Spiegel, Bobby Murphy, and Reggie Brown built the app they wanted to see in the world—one that was more likely to appeal to a hard-partying, time-wasting college kid than a number-crunching software engineer, a data-driven venture capitalist, or, let’s face it, a mature adult. It was not a social network but a camera app, a tool for snapping and sending funny or flirtatious pictures to your friends (and sexts to your romantic partners).

To foster intimacy, it focused on one-to-one communication, and privacy was the default. To encourage free expression, it dispensed with like counters or other public “vanity metrics,” as the company calls them. To cover users’ tracks, it automatically deleted photos once they’d been viewed. The app opened not to a content feed but directly to your phone’s camera, beckoning you to go ahead and snap a selfie without worrying about what other people were saying first. The name itself—“Snapchat”—was a breezy mashup of two onomatopoetic colloquialisms, each connoting casual spontaneity.

It quickly became clear that Spiegel and his co-founders’ hunch was right. They weren’t the only young people craving a way to share pictures with friends without simultaneously offering them up to their parents, their exes, distant acquaintances, random strangers, future employers, police, and the Silicon Valley surveillance complex. The app took off, propelled as much by word of mouth as by the scaremongering headlines from mainstream media sources that portrayed it as a honeypot for hackers and sexual predators. Even some venture capitalists, always on the alert for the next big thing, admit now that they initially considered Snapchat “trivial.” Without a ranking algorithm to help it predict user interests and behavior, the app lacked an obvious path to a lucrative targeted advertising business.

Google, Facebook, and many other Silicon Valley tech firms, in contrast, have ranking algorithms at their core. Google was started by academic computer science researchers seeking a more efficient way to rank websites by relevance to a given search term. Facebook’s news feed is built on machine-learning software that ranks thousands of possible posts for every user based on a complex model that predicts how they’ll interact with each one. Netflix, Salesforce, Yelp, and LinkedIn all lean heavily on proprietary data and prediction models.

Yet Snap has mostly resisted automated ranking, even as it has grown. Private messages and stories appear in your feed in the same order they were sent. Snapchat’s “Discover” tab relies on publishers to curate their pages manually. It’s a model that hearkens to an earlier era in media, when editors rather than computer programs decided what people would read.

That isn’t because Snap lacks the resources or programming talent to build sophisticated prediction models, Spiegel says. It’s part of the company’s philosophy.

In a rare public interview at Vanity Fair’s New Establishment Summit on Oct. 2, Spiegel told Walter Isaacson that he views metric-based ranking systems as inimical to the goal of encouraging users to express themselves freely. “On other services, where you open straight into media created by other people, other media that’s popular, you start with this layer of judgment. It’s almost paralyzing to think that anything you create is going to be judged by millions of people.” Likewise, he finds Facebook-style ranking of news stories to be hostile to high-quality content. “When we met with publishers, we felt like they were really struggling to express themselves on social media. The clickbait-y headlines, all the articles aggregated, you never really have an editorial point of view.” With the Snapchat’s “publisher stories” format, Spiegel said, “we restored the editorial perspective.”

Creativity, quality, and self-expression are concepts that Spiegel references almost constantly. He sometimes thinks and talks more like an artist than a tech CEO.

That doesn’t mean Snapchat eschews data altogether. But it seems to employ a different process than many of its Silicon Valley peers. Facebook’s news feed ranking team spends most of its time analyzing the data its users generate, looking for trends that suggest a need to tweak the algorithm. But Snapchat seems more often to start from a hypothesis, then seek data that could help to confirm or modify it.

For instance, Spiegel told Isaacson that when he started the company, 1 out of every 10 messages that people were sending on their smartphones was an image. That supported his theory that people might like a messaging service that was built around the camera, rather than around text. It also suggested that people were increasingly using smartphone pictures as a form of ephemeral, interpersonal communication, rather than to document their lives. But it’s not the sort of data point you’d even know to look for unless you first set out to rethink mobile image sharing.

Likewise, the company famously pioneered the vertical video format for ads in 2015, bucking the industry wisdom that videos had to be horizontal. Since almost no one was doing vertical videos, Snapchat couldn’t draw on extensive empirical data to prove that they would work. Instead, Spiegel told Adweek, the decision was based on the company’s own observations of how users interacted with its app. “People just don’t rotate their phones,” he said. Only after Snapchat began using vertical videos did the data exist to vindicate the move. (Other companies have since followed.)

This counterintuitive approach doesn’t always work. In September 2016, Spiegel used an exclusive Wall Street Journal story to announce that the company was renaming itself Snap and launching its first hardware product, a pair of sunglasses that could record 10 seconds of circular video. Despite the spectacular failure of Google Glass the year before, Spectacles won effusive praise from many in the tech press. But they have yet to catch on with a wide audience. Spiegel said the company has sold about 150,000 so far, a figure he compared favorably with sales of Apple’s original iPod. But Snap’s earnings data suggest Spectacles sales are slowing, not accelerating.

Last week, Snap announced its latest attempt to flip conventional wisdom around. In an interview with Backchannel’s Jessi Hempel, Spiegel announced Context Cards, which let users swipe up on snaps to do things like learn more about a business, reserve a table at a restaurant, or hail a ride. That sounds straightforward enough. Yet here too, Spiegel has a chin-scratching big-picture theory to underpin the feature. Whereas most forms of online search and discovery are text-based, he sees a future in which images and videos serve as the entry points for content.

That, anyway, is his hunch.

* * *

Snap’s approach to innovation, which prioritizes surprise and delight over efficiency, comes across as refreshing in an era when our biggest and most powerful tech companies seem to spend most of their time copying each other’s products. Clearly it worked wonders for the company until recently. And Spiegel seems convinced it will continue to work even now that Snap is a publicly traded company under intense pressure from investors. “One of the things that happens when you’re an innovator is there’s actually no benefit to being really, really fast,” he told Hempel. “You’re the one creating the new stuff, so there’s no one who’s racing you. It’s actually very important that you are slow and deliberate.”

It’s another way that Snap differs from Facebook, whose famous unofficial motto used to be, “Move fast and break things.” (Mark Zuckerberg sees continuous trial and error, combined with constant testing and feedback, as the quickest path to improvement.) But lately Facebook’s approach seems to be winning out. The company has copied Snapchat’s hit Stories feature not once but four times: on Instagram, WhatsApp, Messenger, and finally Facebook itself. Even Google is reportedly building a Stories competitor. Facebook’s “fast follow” strategy has it raking in more money than ever, and its stock has been soaring. Meanwhile, Snapchat’s first two quarters as a public company have both proved disappointing in terms of user growth and revenue.

The ever-confident Spiegel tries to brush aside concerns from investors and the media, but it seems clear they’re getting to him a bit. On the company’s second earnings call, Spiegel got snippy with an investor who questioned the company’s lackluster user growth, saying that Snap doesn’t rely on “growth-hacking” techniques like certain competitors. (He declined to name them.) In the Isaacson interview, he struck a more diplomatic tone when asked if he was worried Facebook might crush him. “We’re going to continue to build a culture within our company that celebrates creativity,” he said. “We fundamentally believe that by empowering self-expression for everyone, we’re going to grow our business.”

But when Isaacson asked him why Snapchat is still struggling to connect with an “older audience,” Spiegel interjected. “I’m not sure that’s totally accurate,” he said. “Half of the people who sign up for our product today are over 25.” Laughter broke out from the mostly well-over-25 audience.

There are some signs that Snap is opening itself to some forms of automation. But it isn’t clear yet how they’ll mesh with its analog vibe. For instance, the company launched a self-serve ad platform in May, stepping away from its previous approach of working closely with individual advertisers. (This handholding was necessary, the company felt, to ensure that its full-screen video ads wouldn’t scare users away.) Within months, Snap was taking flak for an avalanche of “bad ads” that were seen as cheapening the product.

Teens have long been viewed as a bellwether for a social platform’s future success. But Snapchat so far seems to be an exception—its growth is slowing even as teens are using it more than ever. The same elements of surprise, delight, and human curation that inspire devotion among the faithful may be limiting Snap’s appeal to a wider audience. It’s admirable in some ways that Spiegel refuses to simplify his company’s products to appeal to the least common denominator. But ultimately, in a business ruled by scale, network effects, and growth metrics, it’s that least common denominator that decides which company dominates and which one goes poof.