What I learned during 5 years in Facebook / Meta

Photo by Laurenz Kleinheider on Unsplash

I hit 5-year mark in Facebook / Meta in Feb 2024 (internally referred as MV5). As celebration, I reflected on what I learned, picked the top 5, and wrote this article. Now that I just left Meta and started my 2nd startup in May 2024, I realized this article still helps illustrate what motivated me this time, and I only needed to redact a small portion for Facebook / Meta internal information. So here you are, enjoy.

Learning #1. User acquisition is also a user conversion problem

When I sold Leap.ai to Facebook in 2019, my goal was clear: to learn about user growth. Back then, I was in the mind of seeking the answer to the question "when I build a new product, how do I acquire new users?"

I've first hand observed / led many 0->1 efforts in Facebook/Meta since then, and many of them grew fast. In my mind though, I was still thinking, "these don't count, since all of these are converting existing Facebook users to the new product". In other words, we were mostly doing user conversion problem (e.g., how to convert a Facebook user to a Facebook Marketplace user, and how to convert a Facebook Marketplace user to a Facebook Marketplace Video user). That doesn't fully answer the question I sought to answer, which is how to address user acquisition problem.

Then one day, it dawned on me. At the root of it, user acquisition problem is also a user conversion problem; it's just the user conversion happens from a broader sense.

Take SEO as an example. I can think of it as "acquiring a user from Google search channel". In that sense, it's a user acquisition problem.

But I can also think of it as "converting a Google search user to a Google search for my product user", thus it's also a user conversion problem. It's the same funnel conversion analysis and problem solving, and it's just that the first stage of this funnel is beyond my product.

That moment was a lightning moment for me.

Learning #2. 0->1 product development has 2 stages: pre-data and post-data

I've done 0->1 efforts almost my entire career. I'm a risk taking person; I don't mind failures; I enjoy learnings.

One thing I kept searching though, is what's a simple secret for 0->1 efforts that I can have high confidence to say it's mostly true.

From the title of this learning point, you can already guess the simple secret now I believe in. But before talking about that, let me first talk about the non-true ones.

• False: Quality determines 0->1 product success.

  • This was my belief when I was at Google. When Google Search started, it was just way better than everyone else, so it won nearly instantly. When Gmail started, the thread model, free storage, and spam detection, way better than everyone else, so it won nearly instantly. There were so many examples of high quality product clearly wins out.

  • I believed in it so much, and when I did my first startup, I followed that religiously. Nope, that didn't work. Product / technology can be great, but if no ones knows about / uses it, it's just not going anywhere.

• False: data analysis determines 0->1 product success.

  • This is the trap I see many Meta teams fall into. Many "thorough strategy analysis" and "alignment reviews” before anything gets started. And most of them fall apart afterwards.

  • To all these heavy-data-analysis-before-starting people, I ask a simple question: what data did Steve Jobs have to indicate touch screen smart phone will be success when he started development of the first iPhone?

What I believe now: 0->1 product development has 2 stages: pre-data and post-data.

• Pre-data: understand user psychology, and develop your intuition, and have a conviction to start it.

  • If you have some data in this stage, especially if you have data in a similar way to indicate the likelihood of success, use it as side bonus.

  • But the main thing to rely on here is understanding user psychology. The product, if done right, will address a user pain. That's the real starting point.

• Post-data: respect data after it's started.

  • Our pre-data intuition is never 100% correct. I'd be lucky if it's 50% correct.

  • How do I tell which part is correct, is to look at the data once the product started. Data will tell you in amazing details what resonated with users, and what didn't.

  • Iterate, based on the data. Amplify the resonated part, change the unresonated part.

0->1 success comes from the combination of a) pre-data conviction being somewhat in the right direction, plus b) fast iterations based on post-data adjustments. More so on b) than a).

#3. Successful leaders of 0->1 need 2 seemingly-opposite attributes

It's a direct deduction from the previous point.

• a) Unwavering belief that they will succeed. They demonstrate complete arrogance on the question of "whether".

• b) Extreme flexibility on how they achieve success. They demonstrate complete humility on the question of "how".

People without a) don't start a 0->1 effort, or easily give up when the first sign of non-success shows up.

People without b) becomes stubborn during execution stage, and lead product to eventual failure.

These 2 aspects (arrogance on one, and humility on the other) are almost exact opposite of the spectrum, so it's very rare to identify leaders with such combination.

Recently I mentioned this to one of my early stage VC friends, and his reaction was "that's so true; everyone has been doing this without clearly articulating it out. I'm gonna write it down and use it intentionally!"

#4. A product needs to be in the right org/company for it to succeed

Except Marketplace, every single product I have worked on in Facebook / Meta has been shut down. They were all loved / heavily used by users, but every single one of them got shut down due to an unfortunate reason, in my opinion. One common thing was, they were all moving a metric that the parent org didn't care about.

• <3 examples and their detailed explanations, redacted here>

None of these products were shut down because they didn't serve their users well. They got shut down because they didn't serve their orgs well.

Build the right product, in the right org / company.

#5. Success can come from any culture

I used to believe success only comes from the right culture. I no longer hold that view. To be more explicit, I no longer believe culture has "right" or "wrong". It only has "matching" or "not matching".

• Apple has a very secret culture. Teams don't talk to each other; competing teams / projects are welcomed. I would hate such environment. But Apple is extremely successful (because of its amazing product design).

• Amazon has a writing-heavy culture. No slides allowed; silent read at the beginning of each review meeting. I would be tortured in such environment. But Amazon is extremely successful (because of its amazing customer experience).

• Google has an eng dominant culture. My early growth benefited a great deal from there. At Google, often times I made launch decisions (instead of PMs). Most companies who copied Google didn't go anywhere, so it tells this model probably isn't a good one. But Google is extremely successful (because of its amazing product quality).

• Meta has an everyone-is-scared-of-performance-review culture. Everyone thinks how this will show up in their performance review before doing work. I tore my hair out every time I hear it (thankfully I've been keeping a near-shaved haircut style for 20 years :) ).  But Meta is extremely successful (because of its amazing user engagement).

Any culture, going to the extreme, and hitting the right chords, can lead to successful business. You just need to decide how much you resonate with it, and can thrive in this culture. There are always parts that you don't agree with; learn to cope with that part, if it's overall net positive.