No Filter: The Inside Story of Instagram

by Sarah Frier

What Porch understood, which everyone else eventually would, was that Instagram’s power lay not in what was posted there, but in how those posts made people feel.

Instagram in one sentence

In early 2015, the singer-actress Miley Cyrus, with 22 million followers, was one of those top accounts. That year, she threatened to quit the app, concerned about seeing so much hate and vitriol for LGBT+ youth, especially in photo comments. Instagram found a way to turn her dissatisfaction into an opportunity to land a positive message. Charles Porch, Instagram’s head of partnerships, and Nicky Jackson Colaço, the head of public policy, flew south to visit Cyrus at her mansion in Malibu. They sat around her dining room table, surrounded by art she said she’d purchased off Instagram, and pitched a different plan. She could use the @instagram account as a venue to promote her new Happy Hippie Foundation, which was dedicated to protecting young people who were homeless or vulnerable because of their sexuality or gender identity. Cyrus and @instagram could jointly share thoughtful portraits of people like Leo Sheng, @ileosheng, a trans man, to increase visibility for the people Cyrus was hoping to support. Cyrus loved the idea and decided to keep using the app, even though Instagram lacked a broad solution to bullying.

Zuckerberg would eventually force Instagram to open the floodgates and let in ads from any random business buying on the Facebook website. Before they did, for the next few months, the app’s engineers raced against the clock to build a system that would save Instagram from death by pixelated digital billboards.

“Do it for the ’gram” start to catch on. The people who were trying to build businesses off their Instagram photography needed to stand out, so they would venture to picturesque overlooks and beaches, which saw an increase in foot traffic. On the one hand, this quest brought people outside more, and to new locales; on the other hand, it damaged the environment the photos were meant to appreciate with litter and overuse. National Geographic wrote about how Instagram was changing travel: visits to Trolltunga, a photogenic cliff in Norway, increased from 500 a year in 2009 to 40,000 a year in 2014. “What photos of this iconic vista don’t reveal is the long line of hikers weaving around the rocky terrain each morning, all waiting for their chance to capture their version of the Instagram-famous shot,” the magazine wrote.

Do it for the gram

Systrom had been bothered by the fact that Instagram’s headquarters weren’t obviously Instagram’s. Facebook plasters its walls with motivational posters, printed on-site, with phrases like “Done Is Better than Perfect” and “Move Fast and Break Things,” which represent the antithesis of celebrating craft. The previous year, in 2014, Systrom had ripped down some of them in Instagram’s micro kitchen, in a rare display of emotion. Then he spent millions on a renovation of the space, especially his own conference room, which he called South Park, after the company’s early offices. He adorned it with modern green chairs, a wallpaper patterned with blown-up details of employees’ thumbprints, and an acrylic table displaying his first-ever Instagram photo, of his fiancée’s sandaled foot plus a dog at a taco stand in Mexico.

Develin’s team also found that users weren’t posting more than one picture a day. It was considered rude, and even spammy, to take over your followers’ feeds by oversharing, to the point that people who did so started using a self-aware hashtag, #doubleinsta.

The Instagram team was going to try the opposite: launching Stories, at least a simple version of it, to all 500 million of its users at once. They called it a “YOLO launch,” after the acronym for “you only live once.” It was an extremely risky strategy by Facebook standards, but Systrom couldn’t be convinced otherwise. He thought it was such a big change that everyone needed to be able to access it, or else it would be starved of the oxygen it needed to work.

On the day of the launch of Stories in August 2016, the whole team arrived around 5 a.m. at Facebook’s headquarters, which were otherwise empty that early. In the Sharks at Work conference room, they stood around with breakfast burritos, which had been catered because none of the cafeterias were open yet. Supporters showed up until it was standing room only around Nathan Sharp’s computer. “FIVE, FOUR, THREE, TWO, ONE,” the team counted down, and Sharp pushed a button to send Stories live to the world at 6 a.m. PST. Everyone watched as the numbers climbed. A couple employees snuck some celebratory bourbon into their coffee when Systrom wasn’t looking. The office now had a glass case full of expensive bottles.

Soon after Stories launched, Instagram took a symbolic step out of its parent company’s shadow. The employees moved off campus, out of Hacker Square and into a multistory glass building about a five-minute shuttle bus ride from Facebook’s like button sign.

Systrom was feeling invincible. Two weeks after the Stories launch, he recovered from the anxiety with a vacation. He was still on his cycling kick, trying harder and harder rides on different types of bikes he purchased from King. So on the trip, he challenged himself to summit Mont Ventoux, one of the most difficult hills in the Tour de France. “I have never worked as hard as I did on this climb, but I survived!” he posted on Instagram, posing triumphantly with his bike and a bottle of Dom Pérignon. His caption told the world his time was 1:59:21, only double the overall record of one hour. With that, he had finally earned the right to order his ultimate fancy bike: a Baum. The Australian maker of the bespoke bike would need a couple months to craft it out of custom-butted titanium to make it as lightweight as possible and tune it for Systrom’s specific riding style. It would also feature specific red and blue stripes, an homage to Martini car racing, that would take thirty hours to paint. Nate King was tickled, because most of the people buying a Baum from his shop did it for the prestige; he knew that Systrom would actually ride it.

Instagram had its own billions in revenue, its own world-changing app, its own product vision and strategy, and its own offices. Its leaders had learned how to make difficult decisions by recognizing their blind spots and removing the high bar for posting. Employees allowed themselves to feel, for a few victorious months, like they might one day be as important as Facebook. A Facebook 2.0, making decisions more thoughtfully, in a way that made users happier, borrowing some lessons and rejecting others, modeling the future of social media. They might, if they kept going in this direction, make it to 1 billion users. But soon, Facebook would be in crisis. And Zuckerberg wasn’t about to let Instagram forget whom they were working for.

Earlier that year, as Instagram was building Stories, the online technology news site Gizmodo had written about a team of Facebook contractors who curated news into a “trending topics” module on the right side of the news feed. It was the only human-led editorial component of the social network. The blog cited anonymous Facebook contractors who said they routinely served up content from publishers like the New York Times and the Washington Post, but eschewed right-wing Fox News and Breitbart. Gizmodo also reported that employees were openly asking Facebook management whether they had a responsibility to prevent a Trump presidency. The reporter implied it was scary that Facebook’s employees realized their company was powerful enough to do this, if it wanted to. Facebook, in its response to the firestorm from the leaks, invited 16 of the most TV-friendly conservative political commentators, including Tucker Carlson, Dana Perino, and Glenn Beck, to its headquarters to learn about how the news feed was programmed. They reassured their guests that Facebook had no editorial bent. Later, they cut humans out of the process for picking trending topics, so that what trended on Facebook was determined by algorithm only.

It was all of these things Instagram avoided—hyperlinks, news, virality, edge stories—that cheapened Facebook’s relationship with its users. Facebook was indeed biased, not against conservatives, but in favor of showing people whatever would encourage them to spend more time on the social network. The company was also in favor of avoiding scandal, appearing neutral, and giving the public what they wanted. But as Facebook became a destination for political conversations, the human curation in “Trending Topics” wasn’t the actual problem. It was how human nature was manipulated by Facebook’s algorithm, and how Facebook looked away, that got the company in trouble.

What got Facebook into trouble

Stories that claimed the pope had endorsed Trump, or that Clinton had sold weapons to the Islamic State, were juiced by Facebook’s algorithms and promoted to millions of Facebook users. In the three months prior to the election, the top stories with false information reached more people on Facebook than the top stories from legitimate news outlets. Some of them came from makeshift websites designed to look real, with names like The Political Insider and Denver Guardian. On Facebook, the subterfuge worked. All links were presented in identical fonts on the news feed, awarding a scrappy conspiracy theorist the same credibility as a fact-checked report from ABC News. One such site even had the web address ABCnews.com.co, even though it was not affiliated with the network. The most shareable content on Facebook was what made people emotional, especially if it triggered fear, shock, or joy. News organizations had been designing more clickable headlines ever since the social network became key to their distribution. But those news organizations were getting beaten by these new players, who had come up with an easier, more lucrative way to go viral—by making up stories that played on Americans’ hopes and fears, and therefore winning via the Facebook algorithm. The fake news on the websites would be interspersed with stories that weren’t entirely fake, but were hyper-slanted with context or loaded language designed to reaffirm readers’ paranoia or political loyalty. People shared ...

a data scientist posted a study internally on the difference between Trump’s campaign and Clinton’s. That was when employees realized there was another, maybe even bigger way their company had helped ensure the election outcome. In their attempt to be impartial, Facebook had given much more advertising strategy help to Trump. In the internal paper, the employee explained that Trump had outspent Clinton between June and November, paying Facebook $44 million compared to her $28 million. And, with Facebook’s guidance, his campaign had operated like a tech company, rapidly testing ads using Facebook’s software until they found the perfect messaging for various audiences. Trump’s campaign had a total of 5.9 million different versions of his ads, compared to Clinton’s 66,000, in a way that “better leveraged Facebook’s ability to optimize for outcomes,” the employee said. Most of Trump’s ads asked people to perform an action, like donating or signing up for a list, making it easier for a computer to measure success or failure. Those ads also helped him collect email addresses. Emails were crucial, because Facebook had a tool called Lookalike Audience. When Trump or any advertiser presented a set of emails, Facebook’s software could find more people who thought similarly to the members of the set, based on their behavior and interests. Clinton’s ads, on the other hand, weren’t about getting email addresses. They tended to promote her brand and philosophy. Her return on investment ...

Facebook ads helped Trump

Systrom decided Instagram should lean into its feel-good image, giving people even more tools to block out what they didn’t want to see. By December 2016, Instagram was letting users turn off comments for posts entirely if they wanted. Systrom’s willingness was in stark contrast to the attempts by Facebook and Twitter to err on the side of leaving content up, in an attempt to promote environments they said were neutral and open, but that in practice were rarely policed. The same ideas, of letting users turn off comments or block them according to keyword, had been suggested several times at Facebook over the years. But it had never stuck. If there were fewer comments, there were fewer push notifications, and fewer reasons for users to come back to the site. Even on Instagram’s team, the former Facebook employees promised Systrom that they would find a way to build out the tool so it was difficult to find, and applicable only to one post at a time. That way, it wouldn’t be used as often. Thanks but no thanks, Systrom said. He explained that he wasn’t worried about losing engagement, that the team was thinking too short-term. Over the long term, if the tool was easy to find and well publicized, people would have more affinity for Instagram, and the product would better weather storms of bad publicity, like the kind Facebook was starting to receive. Systrom wanted to think even bigger than comments. He started talking to Jackson Colaço about a “kindness” initiative. How could...

By April 2017, Zuckerberg’s pronouncements about global community had started to feel more like preemptive strikes in a public relations battle. That month, Facebook put out a cryptic research paper explaining that it had found instances of “information operations” conducted by “malicious actors” on its social network. Basically, some entities (they didn’t say who) were creating fake identities on Facebook, befriending real people and spreading misinformation, trying to skew public opinion. As Obama had warned, the fake-news problem wasn’t just about a few shady entrepreneurs—it was about foreign actors that had weaponized the social network’s algorithm. This revelation played into political suspicion around whether Russia had assisted in Trump’s win, and whether all of those fake stories—about the pope endorsing Trump, or about Clinton working with the Islamic State—had gotten to be so popular on social media as part of an elaborate propaganda campaign. Was Facebook a part of Russia’s plan? Was Russia helping Trump?

Everyone with a blue checkmark, after realizing that their comments would be prominently displayed, had an incentive to comment more. The comment ranking helped brands, influencers, and Hollywood types fight their deprioritization by the main algorithm. Instagram commenting became marketing, or, in the vernacular of Silicon Valley engineers, “growth hacking.” The “hacking” didn’t end there. The most strategic Insta-famous weren’t just commenting on their friends’ posts, but on accounts that might make them seem more well-connected and relevant than they actually were. One influencer with a verified account, Sia Cooper, @diaryofafitmommyofficial, told Vogue she gained 80,000 followers in just a few weeks by lovingly commenting on Kardashian-Jenner posts, though she didn’t actually know the family: “I choose to comment on the highest followed accounts because this means my comment is more likely to be seen by many more users,” positioned at the top of the stack. Today, she has more than 1 million followers, and has inspired others to use the same strategy.

An Instagram user’s path to success was obvious, based on benchmarking against others. All you had to do was create the right kind of content: visually stimulating, with a reflective but optimistic caption, inspiring some level of admiration. En masse, those activities spilled over into real life and real business decisions. The version of Instagram that the founders had set out to create, one that would foster art and creativity and provide visual windows into the lives of others, was slowly being warped by the metrics Instagram prioritized, turning the app into a game that one could win.

Key to Instagram success

Every day, Camille Demyttenaere and her husband, Jean Hocke, choreograph experiences entirely for the purpose of posting them on Instagram. Once, through the open door of a dark teal train curving through a jungle in Sri Lanka, Demyttenaere lunged out the side of the train into a passionate kiss with Hocke. She leaned forward, both arms fully extended behind her while hanging on to the side of the train with both hands, on top of him, with her knee up near his biceps, as he leaned out and back, his left arm dangling, holding on with only one hand, hovering over treetops. “ONE OF OUR WILDEST KISSES,” the travel influencers, who go by @backpackdiariez, said in the May 2019 caption. Commenters reacted immediately. “Are you really ready to die for a pic????????” one said. The media picked up the story internationally, writing about travel culture and the dangerous lengths people were willing to go for the ’gram. Several of them cited a study that logged 259 deaths during attempted selfies between 2011 and 2017, mostly by people in their early twenties taking unnecessary risks. Ironically, the outrage was the best reaction the Belgian couple could have hoped for. Now, with their profile posted in more news sites, they increased their exposure and gained about 100,000 followers with a tripling of views for their Instagram stories. Their inbox overflowed with offers from tourism boards and hotels, which had discovered their profiles in the international coverage of the incident.

The Instagram effect has made it harder to sell expensive tangible products, like cars and clothing. Nine major retailers in the U.S. filed for bankruptcy in 2017, and many more closed their stores. Besides the rise of Amazon, analysts cited the experiences-not-things trend for affecting retailers’ bottom lines.

Instagram’s new comment filter would take all the worst comments and just make them not exist. Few users would notice the new default setting on their apps. It would just make Instagram seem more pleasant than it actually was.

Facebook’s culture for responding to crisis was fully reactive: the company addressed problems only once they resulted in major blowups that politicians and the media were paying attention to. With the crisis over Russian influence, Facebook was in a frenzy, while Instagram was insulated. At the time, Instagram only sent a couple communications and policy employees part-time to Facebook’s internal war room, so they could help figure out what had happened and answer government questions. But the rest of the team was updating the product as usual, improving Instagram Stories and the new algorithm.

Facebook executives talked among themselves about how ungrateful the WhatsApp founders had been. The consensus was that the team had always been high maintenance, asking for slightly bigger desks, longer bathroom doors that reached all the way to the floor, and conference rooms that were off-limits to other Facebook employees. If they wanted to leave in a huff at the slightest suggestion of making the investment worth it, after Zuckerberg had made them both billionaires, then good riddance. “I find attacking the people and company that made you a billionaire, and went to an unprecedented extent to shield and accommodate you for years, low-class,” David Marcus, the Facebook executive in charge of a new cryptocurrency initiative, later wrote publicly. “It’s actually a whole new standard of low-class.”

On Friday, March 17, 2018, the New York Times and the Observer simultaneously broke the news that years earlier, Facebook had allowed the developer of a personality quiz app to obtain data on tens of millions of users, which that developer then shared with a firm called Cambridge Analytica. Cambridge Analytica retained the data and used it to help build its political consultancy. The company aggregated information from several sources to build personality profiles on people who might be receptive to ads that would help conservatives win elections. Donald Trump’s campaign was a client. The story hit all of Facebook’s weak spots: shoddy data practices. Negligence. Lack of transparency with users. And a role in Trump’s win. It stoked distrust among politicians the world over.

On a Tuesday in June, Instagram finally reached the milestone they’d been working toward: 1 billion users. This was the pinnacle they’d realized it was possible to reach after launching Stories. It was also the same metric Facebook hit the week Instagram had joined the company in 2012. Now Facebook and Instagram were peers in shaping the world through their products at a massive scale.

Zuckerberg’s “family of apps” plan. Cox told Zuckerberg he needed to let the products build independently and not become too similar. “They’ll compete a bit with each other, but if we have more unique brands, we’ll be able to reach different kinds of users.” He and Systrom had spoken extensively about using Harvard professor Clayton Christensen’s “jobs to be done” theory of product development, which states that consumers “hire” a product to do a certain task, and that its builders should be thinking about that clear purpose when they build. Facebook was for text, news, and links, for example, and Instagram was for posting visual moments and following interests. But that wasn’t how Zuckerberg thought of it. “We should think of this globally,” Zuckerberg said. “We’re trying to build a global community—not a bunch of smaller communities.” If you added up all the unique people using at least one of the apps in the family, you got a community of 2.5 billion, which was bigger than Facebook. “I just think it’s going to be hard to do,” Cox protested. “These are pretty different teams, and their user bases are differentiated a lot already.” “Aren’t we taking an operational risk?” Systrom added. “If I have to worry about Instagram, but also have to worry about Messenger and Facebook, I’m not sure what that looks like in practice.” “I think it’s a risk we should take,” Zuckerberg declared.

The debate that precipitated the split

Zuckerberg directed Javier Olivan, Facebook’s head of growth, to draw up a list of all the ways Instagram was supported by the Facebook app. And then he ordered the supporting tools turned off. Systrom again felt punished for Instagram’s success. Instagram was also no longer allowed to run free promotions within the Facebook news feed—the ones that told people to download the app because their Facebook friends were already there. That had always brought a steady stream of new users to Instagram. Another of the new changes would actually mislead Facebook users in an attempt to prevent them from leaving for Instagram. In the past, every time an Instagram user posted with the option to share on Facebook, the photo on Facebook said it came from Instagram, with a link back to the app. Instagram’s analysis showed that between 6 and 8 percent of all original content on Facebook was cross-posted from Instagram.

After all the hours Systrom had spent in leadership coaching over the years, all the books he’d read about how to be a better CEO, and all his personal improvement quests, he was faced with an unexpected personal discovery: he wasn’t the boss. He started telling his close confidants that if Zuckerberg wanted to run Instagram like a department of Facebook, maybe it was time to let him. Maybe there wasn’t room for another CEO.

In the months after the Instagram founders left, their app was rebranded as “Instagram from Facebook.” The group in charge of Instagram’s direct messaging was transferred to report to the Facebook Messenger team. In late 2019, Zuckerberg made a cameo appearance at an Instagram-branded conference and took a selfie with the crowd. Internally at Facebook, he was talking about using Instagram to take on TikTok, the Chinese app that had replaced Snapchat as the top threat to Facebook’s dominance.

Mosseri’s answer to the important question was perfect by Facebook standards: “Technology isn’t good or bad—it just is,” he wrote. “Social media is a great amplifier. We need to do all we can responsibly to magnify the good and address the bad.” But nothing “just is,” especially Instagram. Instagram isn’t designed to be a neutral technology, like electricity or computer code. It’s an intentionally crafted experience, with an impact on its users that is not inevitable, but is the product of a series of choices by its makers about how to shape behavior. Instagram trained its users on likes and follows, but that wasn’t enough to create the emotional attachment users have to the product today. They also thought about their users as individuals, through the careful curation of an editorial strategy, and partnerships with top accounts. Instagram’s team is expert at amplifying “the good.”

Systrom and Krieger sold Instagram to Facebook because they wanted it to be bigger, more relevant, with more longevity. “You should be able to take a chance and build something of value for the world that should be able to grow and be worth a lot, and use that to give back socially,” Systrom explained to New York magazine. “We tried really hard to do that, to be a force for good.”

If journalism is the first rough draft of history, books build on that important work to propose a second draft. I’m grateful to all the reporters who asked questions of Instagram throughout the years, and to the people who continue to cover its impact on our society and culture, and its place within Facebook. Reporters have been cited in the endnotes if their work found a second life in these pages.

Well said

SARAH FRIER is a technology reporter for Bloomberg News. Her award-winning features and breaking stories have earned her a reputation as an expert on how Facebook, Instagram, Snapchat, and Twitter make business decisions that affect their futures and our society. Frier is a frequent contributor to Bloomberg Businessweek and Bloomberg Television. She lives in San Francisco, California.