What I Learned Leading User Acquisition for Instagram from 700M → 1.1B people

And how it applies to startups

I was the product leader for the Instagram User Acquisition team from 2017 to 2019. In that time, Instagram grew from 700M to 1.1B people and was the fastest growing app in the world.

Looking back, there was one thing my team did differently than others that helped to drive that outlier success: Asynchronous brainstorms.

Asynchronous brainstorms

Before we started asynchronous brainstorms, we did it the traditional way: get everyone in a big room, use post-it notes for ideas, put them on a whiteboard, let people build off them, then type it all up at the end. It worked decently well, but it had one problem: most of my team wasn’t born in the US so English wasn’t their first language. I noticed that folks who were shyer or felt uncomfortable speaking in large groups would share fewer ideas. I wanted to hear from them too.

I devised an asynchronous brainstorming format that asked each team member to spend time independently reflecting on how to reach our goals and then send them to me ahead of time. This did a few things:

1. It signaled that everyone’s ideas mattered and everyone was expected to contribute.

2. It held each person accountable for deep thinking to contribute to our goals.

3. It created a safer, more easeful way for introverts and non-public collaborators to contribute.

The asynchronous brainstorm format worked like this:

1. 1-2 team members put together a summary doc of key data insights and lessons from the last quarter and presented it to the team. This put everyone on a level playing field in terms of data to facilitate ideas.

2. I wrote a list of questions for folks to use for deeper reflection on what we should build next.

3. I sent out both the data doc and the questions doc with a deadline for people to submit their ideas.

4. I blocked an hour on everyone’s calendar to remind them to fill it out, though they could use whatever time they wanted for it.

5. I compiled the responses into one document anonymously. No one got “credit” for their good ideas or needed to fear embarrassment for their bad ones.

6. We had a “brainstorming” session where we all sat in a room together and read each others’ ideas on our computers in the shared document. People commented and added new ideas to build off them. We also discussed ideas live.

7. I asked team members to “emoji vote” for the ideas they thought were the most promising. This formed the basis of how I narrowed down the ideas towards the final roadmap.

8. We “opportunity sized” the ideas we thought had the most potential, using data from past experiment results to help estimate how big the impact could be from each idea.

9. We also added an engineering cost estimate (small, medium, large) to put the impact into context. Each of those sizes was tied to an approximate number of engineering days of work.

10. We prioritized the ideas based on impact relative to engineering cost, and drew the line of what we put on our roadmap based on how how much engineering capacity we had on our team (with some buffer).

11. We summed up the total opportunity sizes for the projects and that became our quarterly goal. And since we drew the line based on total engineering capacity, the whole team could look at that list and say “yes, we can and will do this.”

That winning mindset was ultimately what drove the winning outcome. We had better ideas because everyone around the table had thought deeply about what was possible, and then we brought it all together and signed up for it as a team.

The results spoke for themselves: after two years, we saw in our long term experiment holdouts that the products we shipped drove an incremental 50M monthly active people per year above organic growth.

Beyond the asynchronous brainstorm, there were additional ways we operated the User Acquisition team that drove success:

1. A single clear goal, with a metrics-based project list that added up to the goal

2. Metrics-based accountability with a blend of personal autonomy and collaboration

3. Weekly experiment reviews

A single, clear goal

It was very clear what the User Acquisition team was driving toward at Instagram: Monthly Active People (MAP). Each quarterly planning cycle, we set a number of incremental MAP we wanted to add according to experiment holdouts. This meant that the sum total of the experimental value of the projects we shipped would add up to our team goal.

The important thing wasn’t just the goal: it was the fact that we had a list of projects, each with an estimated MAP impact, that laddered to that goal. We didn’t just have a pie-in-the-sky ambitious goal, we had a believable roadmap for the journey of how to get there.

By the end of each quarter, some of the projects would have an outsized impact compared to what we estimated, and others would have fallen short. But each time, we’d find a way to exceed the goal because we believed it was possible to hit from the start.

Metrics-based accountability

Individuals on the team were rewarded in performance evaluations according to the metrics impact of the projects they worked on plus feedback from the team. This created an excellent blend of personal autonomy and collaboration.

On the personal autonomy side, individuals were motivated to find larger-impact projects to work on. On the collaboration side, several people could receive “credit” for impact achieved, so it made sense to work together, particularly across functional disciplines.

As the team’s product leader, I kept a detailed log of projects shipped, metrics impact, and who worked on them, and shared this information during the performance review process so my team would be rewarded for their work.

Weekly experiment reviews

The Instagram User Acquisition team shipped experiments left and right, and sometimes engineers would push something out, only to forget to check back on it later. (We did this at my startup too. 🤦🏻‍♀️)

The problem with shipping and forgetting is you miss the learning. Each week, we’d bring the 20-person team together for a quick review of all currently running experiments. We’d decide what to fully roll out and what to cut and discuss what we’d learned from each of those. This meant that learnings accumulated across the team, because each team member learned what projects drove impact and what didn’t regardless of whether they had worked on them.

For startups

I brought much of this approach over to my startup, Plura, during our early days, and they worked just as well from the seed stage onward as they did at Instagram’s massive scale. Asynchronous brainstorms were the most relevant of these in the pre-metrics phase since they drove more original ideas.

After you achieve product market fit and start to scale, you want to get a good handle on metrics and learn which levers move which numbers. Learning the levers helps you choose where to invest product, engineering, sales, and marketing resources for growth. In the process outlined above, this means you’ll be arming the team with better data to inform future product decisions.

If you want to learn more about this, my coaching practice is nearly full, but reach out for a discovery conversation and we can see if it’s a fit. lunarayemail@gmail.com