Introduction
Chapter 1: Introduction
It was a Friday morning in summer 2014. I was working at Facebook, and it was time for my first “Zuck review.” That day and time slot was always reserved for something product-related for Mark Zuckerberg, the founder and CEO of Facebook, to review. I was presenting to Mark and his lieutenants, who my team and I fondly referred to as “Mark and friends,” with the ultimate strategy for driving news content discovery and consumption on Facebook.
To this day, I don’t think any of us could really see and discern the full global impact of the decisions we would make that day. But I had a hunch.
I made a comment during the presentation that got a harsh reaction out of Mark…
Me: “Mark, if we are half as successful with this strategy as we might be, at some point we have a responsibility.”
Mark: “Why? We’re just providing information to whoever wants it.”
Me: “Well, if we do this even half as well as we think we can, we might create the world’s largest echo chambers.”
Mark: “Why?”
Me: “Well because we’re inferring likes to news sources, ranking them in newsfeed, considering social graphs, and commonality between them all.”
Mark: “Why?”
Sheryl [Sandberg], COO of Facebook: “Mark, let him finish.” (And then to me, with her soft politician’s grip on my forearm, “I’m so glad you’re here.”)
Sheryl had long been regarded as “the adult” at Facebook. She was just a year older than I, and understood that since I was an “industry” hire (which meant that I was recruited from industry as opposed to the more common homegrown Facebook kids straight out of college), I would be thinking of this opportunity differently than a product manager that had grown up inside the walls of Facebook.
Typical Facebook product managers in those days were known for not fully considering the draft on impact of decisions made on a platform with billions of people on it.
And so we proceeded.
After grounding everyone with tons of data on the current distribution of news consumption in the world, I began to make the pitch that Facebook was well-positioned to distribute the right news to the right people at the right time. It started with how news was naturally being discovered and consumed on Facebook organically, even before this particular news strategy was conceived.
In those days, a news publisher like The New York Times would publish roughly 40% of their articles on Facebook on a given day. The hope was that a Facebook user that had “liked” the Times’ page on Facebook would receive posts in their own feed made by the Times’ page. These posts, usually article headlines, linked back to the full article on the Times’ website when clicked. This is known in the business as “referral traffic.”
From the Times’ perspective, referral traffic meant that Facebook was sending readers directly to their website. Once users landed there, the Times could serve them ads and keep 100% of the revenue without needing to share it with anyone.
Even better, those users were now navigating the Times’ “owned and operated” space, where analytics tools tracked everything. The Times could collect all the data they needed to count those visitors as part of their monthly active user (MAU) base, analyze their user behavior and journey, and leverage that data to impress advertisers. The more MAU and the richer the data, the more likely an advertiser would want to buy.
What became an era of go-go times for publishers leveraging Facebook to drive referral traffic to its online properties wasn’t without its issues, though. For all the marketing dollars the Times spent inside Facebook to get “likes” or “followers,” its posts weren’t reaching all of their audience on Facebook. The newsfeed ranking team at Facebook would frequently change ranking priority from one type of content to another, leaving publishers constantly frustrated and creating a rollercoaster experience for users.
One day the feed was dominated by cat and food pics, another day it was news articles, and another day it was ISIS videos, etc. In those days, people were eligible to see an average of 50 Facebook posts per day. Each time a user visited Facebook on a given day, the algorithm would re-calculate which 50 posts to show. So if Mark decided to prioritize “public content” (an early term for anything news, sports, celebrity, etc), a user would see posts from the public content publishers they had at one point liked or followed. But if Mark decided to prioritize “friend content” on another day or quarter, a user would see more posts from their friends.
This constant reshuffling of prioritized content made it interesting for those working on the various types of content areas. Each team was always competing with one another at the mercy of Mark’s evolving strategy.
Another problem for publishers like the Times was the lack of tools to both target and measure its audience on Facebook. A Facebook “Page” was an account for a business, and the team that built and operated Pages didn’t get very granular on its Page Management features. Consider a “Page” for a church vs a “Page” for a news publisher vs a “Page” for a pro sports team. They’re all uniquely different, but they share the same “surface” by which to have a presence on Facebook.
The other problem we recognized was on the user experience side: the average load time it took for a user to reach a news article after clicking a link on Facebook was about 10 seconds. The longer it took to load a referred website, the more likely a user would abandon the session. Most of the time, users browsing Facebook are there to browse Facebook, not bounce to some external website.
We came up with several key questions to address as part of this new strategy for news on Facebook:
1) How can we incentivize publishers to publish all their articles on Facebook?
2) How can we match the right new stories (Isis, Trump, pop culture, etc) and formats (article, video, photo, etc) to the right user at the right time?
3) How can we make the news more discoverable, consumable, and shareable in the newsfeed?
4) How can we provide publishers validation that Facebook was a better place for them to continue publishing their content?
5) How can we compensate news publishers for sharing their news on Facebook?
At the top of any Facebook strategy or initiative at that time was “how is this good for Facebook” and “how is this good for users?” It wasn’t so much “how is this good for legacy ecosystems or business models?” That’s where I came in. After having been a publisher at Vevo, where we published music videos on Facebook, I understood the power of the platform as well as the pain points publishers and users faced.
In the review with Mark and friends, my team was essentially pitching the notion of convincing news publishers like the Times to let us deliver their news directly within Facebook rather than linking out of Facebook to their websites.
The plan would start with getting an RSS feed from the publisher, allowing us to load every article instantly rather than taking the 10 seconds to load remotely and risk the user abandoning the session. We’d also provide smart targeting tools to help publishers find and mine their audience, and like audiences, on Facebook.
We’d create a new, more immersive format for news consumption on Facebook with our “Instant Articles” feature. This new format would stand out from everything else in the Newsfeed while driving more exploration, sharing, and conversations about news. On the monetization side, we’d even let the Times keep all the ad revenue. News publisher revenue wasn’t on the magnitude of Facebook revenue, so the idea to let publishers keep it all would be a good gesture of faith to entice them. What’s more, is we’d do the right integrations to ensure that publishers could get all the analytics they normally got from their own websites.
Mark asked another question, “Why would we do all this when we could just scrape all the news from publisher websites?”
Sheryl happened to know Mark Thompson, the then-CEO of the Times personally, so she stepped in once again: “Mark! These are businesses with quality journalism that make our world go around. We’re never going to have that level of journalism ourselves, so we need to support these businesses so that they can continue to create high quality content. If we upend their businesses, the quality content will just go away.”
By the time the meeting was over, everyone was generally aligned on next steps. I’d have to follow up with Andrew Bosworth—aka “Boz”—to make sure he was good with the ad decisioning piece. Although he was originally a teacher’s assistant in Mark’s AI course at Harvard, he now ran the ad tech side of Facebook. Doug Purdy, my engineering counterpart at the time, knew Boz already, so he brokered the meeting and we walked him through the plan together.
We also had to take the pitch on the road to some news publishers to ensure the whole thing resonated with them before we invested in building any of it. It would be important to keep this tour under the radar for fear of it leaking. Mark made mention of, “If this thing leaks it’s not gonna be good.”
So we hit the road to meet with a dozen prominent news publishers. I say “we” because this wasn’t all me. It was my entire product management team’s effort. Holly Ormseth would run Publisher Tools, Michael Reckhow would run Instant Articles, David Chen would run Trending, Andrew Song would run newsfeed ranking, Mitu Singh would run Facebook browser performance, and Allison Swope would run Facebook Mentions, the app made just for public figures and celebrities. (This would be the app that we’d eventually launch Facebook Live on as a test, before taking the feature into the Facebook app.)
Every publisher we met had the same beef with us. They’d spent all this time and money building their audience on Facebook, only for us to make major policy changes that would severely impact their efforts. They were always waiting for the other [Facebook] shoe to drop. I tried to assure them that this time would be different, that we were super committed to making this good for people, Facebook, and publishers. After all, we were letting them keep all their ad revenue as part of this program. In the end, most publishers we met agreed to be beta partners in the program.
A couple months later, a journalist at the Times ran a story on our ambitions to partner more closely with publishers. Mark was irritated because he suspected our ambitions might leak, putting us on the defensive. In an email he sent around, he wrote “See…”
The reality is that publishers love to write about the news (themselves) and love to write about Facebook. The silver lining to all this was that thousands of other publishers reached out to our Partnerships team led by Dan Rose, asking to get early access to the program. The market signal was strong, and gave us confidence to resource this well and move quickly to build it all out.
In a matter of months, we began implementations with our beta partners. And within the year, we had our first 100 publishers on this new program we called “Instant Articles.” That list would grow over time to thousands.
In parallel, the Facebook Mentions app was getting pretty good traction with public figures. They were adopting it and posting content on Facebook, but we decided the app needed a more “killer app” kind of feature. After much discovery and debate, we developed a hypothesis that public figures, if given the ability, would stream themselves live from the Facebook Mentions app on their phone. And when they did, we’d notify all of the users that had liked or followed them so they would tune in.
We knew the audio fidelity would not be great from a phone, but that didn’t matter—the appeal was more about the real-time nature of it, the authenticity of it, and that it could only be seen on Facebook. This idea struck a nice chord with Mark. He was becoming increasingly competitive with YouTube and their continuing uptick in share media consumption. Live streaming would be the one feature YouTube didn’t have at the ready.
We did some early trials of live streaming with a few well-known public figures to test the waters. Ricky Gervais did a livestream of himself telling jokes from his bathtub. The Rock did a livestream of a workout right from his home gym. And Chris Martin from Coldplay did a livestream of himself playing “Fix You” on a piano from his home. Public figures were embracing the medium, and their fans/likes/followers were watching. We were definitely onto something.
We decided to bootstrap this new feature even more by building out a media pop-up facility in Facebook’s New York office that public figures and their teams could use to film and stream.
In the spring of 2015, I visited Facebook’s New York office for the week, and was working from a standing desk about 20 feet away from our media pop up area when Donald Trump stopped by to do a rapid-fire Q&A live on Facebook.
His team, along with Facebook officials, aligned on a set of about five rapid-fire questions for him to answer. An iPad would display the question, and Donald would answer before swiping to the next one. The crew explained to Donald how this was all going to work. But he kept asking for clarifications on the process, even after everyone thought he understood how it was going to work.
“Who’s asking the questions, you or me?” Donald asked.
“You’re going to read the question, then answer it, and then swipe for the next question,” said someone on the Facebook crew.
“Well who’s answering the questions then?” continued Donald.
“You are, Mr. Trump.”
He finally got it, and proceeded with the rapid-fire questions. After the session wrapped up, the moderator concluded with, “Is there anything in particular you want your fans to know about you?”
Donald’s answer alluded to something that would eventually change the trajectory of society and politics forever: “I’ve got big news coming. [I] will be making an announcement early Summer which is on approach. And I mean BIG NEWS.”
That following June, he announced his candidacy for presidency of the United States. He would go on to push the boundaries of social media like never before, using it as a tool to gain voters and secure his ultimate election victory in 2016.
That thing I mentioned during the “Mark and friends” review of our news strategy in 2014—the part about us possibly creating the world’s largest echo chambers—was coming to fruition in real time. The ultimate role of a product manager is to drive hypotheses about what to build for whom and over what period of time, in order to drive impact with outcomes. This was not the impact nor outcome I’d hoped for.
There has perhaps never been as great of a divide in American history as there is now. This work at Facebook certainly wasn’t single-handedly responsible, but it definitely played a contributing role.
While this tale is a lesson in judgment, power, influence, and projecting unintended consequences, it’s only one of many lessons this book aims to provide readers. You can only encounter some of these experiences and gain these kinds of insights by actually going out and building something.
Takeaways
Think Past the Metrics – What you build today might shape the world tomorrow—good, bad, or ugly. Always ask: “What happens if this works too well?”
Juggle the Stakeholder Circus – Product management is part diplomacy, part therapy. Balancing Mark’s platform obsession, Sheryl’s publisher hugs, and user needs? That’s the job.
Disrupt, but Don’t Destroy – Innovation is about shaking things up, but not so hard that you break what works for others. Build trust and leave room for win-win scenarios.
Solve Like a Pro – Big problems need big thinking. Nail the problem, map the mess, and layer in smart solutions—from tech to business models.
Power = Responsibility – Platforms can make or break entire industries. Use that power wisely—or brace yourself for unintended consequences (hello, echo chambers).
Lead Like a Swiss Army Knife – Bring the right people together, mix their skills, and get out of the way. Cross-functional leadership isn’t optional—it’s table stakes.
Brace for Blowback – No product is immune to surprises. If you’re building at scale, always consider what else might happen once your genius idea hits the real world.
