Facebook Marketplace launched in 2016 to make it easier for people to buy and sell items within their local communities. Today, it’s a top visited online marketplace used by hundreds of millions of people all over the world. As the product leader for Marketplace since it started, I’m often asked how we grew it. The good news is that many of the principles used to grow Marketplace can be applied to any product or company looking to expand.
What Growth Is Not
Sometimes growth work gets a bad reputation in the industry. Growth work is a powerful tool, but it’s important to remember what it’s not.
Growth does not drive product-market fit. A growth team focused on growing a product with poor retention is fundamentally flawed. Great growth efforts on top of a leaky funnel mean more people having a poor experience. This burns through potential customers who don’t find value in your product. Instead, wait for retention to stabilize or start increasing before investing in growth.
Growth is not about manipulating people. In testing growth tactics, it is important to stay true to the job the product does. People signing up and using a product that they don’t understand is a long term negative for both your brand and those who use it. Using tactics like hiding a “dismiss” link or s-out may help your growth metrics in the short term; but they will not help you attract and engage people who actually want to use your product, which is what you need to build a healthy product over the long haul.
Growth is not scattershot “growth hacking.” Spraying and praying can yield short-term gains and move metrics, but without a clear hypothesis of why something works these ephemeral gains will likely fade over time or be unsustainable. Think of growth work as an investment you are making to improve your long term product performance. At its core, growth work is about making it easy for people to use your product and get value out of it.
Best Growth Practices
Create a growth team for each product. Once there’s product-market fit, a growth team at a minimum should have a product manager, data scientist, and three to five engineers. If you have additional capacity, adding a growth-oriented designer or content strategist can help drive incremental gains. The goal of the team should be achieving a certain outcome (e.g. grow transactions to X). On Marketplace, there are product teams focused on buyer growth and seller growth. They take numerical goals in driving product usage with the ability to test across the breadth of the product experience, wherever they estimate growth opportunities would be most fruitful.
Create a team of growth experts across your company. If you are at a large company, create a centralized Product Growth team. At Facebook, the Product Growth team embeds analysts in the Facebook app, Instagram, WhatsApp, Oculus, Ads, and many other areas. Since Facebook now has more than 50,000 employees, it’s impossible for any product team to keep up with what growth wins are happening even in adjacent products, much less in other apps. The centralized Product Growth team allows the company to cross-pollinate wins across multiple products vs. each product team re-inventing the wheel. For example, the team running a popular unit in News Feed made some design changes that increased engagement. They shared this win with the Marketplace product growth team, who then quickly leveraged those improvements in Marketplace’s feed unit as well.
Be thoughtful in selecting a goal metric for the team. One of the more important strategic decisions is what metric to set for the team that best represents your overall business goals in a healthy way. First, the goal should be easy for teams to understand. Avoid fancy composite metrics that combine multiple different actions since it will be difficult for teams to understand why the metric is moving up or down. Make sure your metric is also tough for teams to “game” (intentionally or unintentionally) with bad growth practices. Earlier in Marketplace’s history, the team was goaled on getting a buyer to send a single message to a seller. We saw a lot of growth, but we learned many of the incremental interactions weren’t high intent. So we updated the metric to require multiple back-and-forth interactions between a buyer and a seller, resulting in more meaningful interactions on listings.
Be willing to reconsider all your assumptions. A growth-oriented mindset is all about looking at possibilities. At Facebook, we had a saying on our team called "no sacred cats". When the original members of the team took on more products and started leading multiple teams, the newer team members went back and tested many of the growth experiments. We saw something interesting happen. Features and tests we ran before with a negative result ended up with a positive one from the new team. To celebrate those moments, we started the hashtag #nosacredcats in our Workplace groups each time we had a growth win that was revisiting an old idea. Many teams have a sense that products are immutable, but things that have been tested can be iterated on with fresh eyes. Proactively rotating team members helps bring in new perspectives. A growth mindset means being willing to challenge the status quo and seek improvement without prejudice.
Think of growth as a game of inches. If your product is more established, you’re better off working on growth activities that drive smaller, yet somewhat predictable, gains vs. big “whale” projects such as re-doing your entire interface. In Marketplace, our most common growth wins were from things like taking a feature that worked on our Android app and making it work in Facebook Lite; improving the targeting of an existing notification, expanding eligibility for using the product, and extending the reach of known notification channels. The Growth team’s impact for the half consists of more than a dozen modest sized wins. While the Marketplace Growth team also drives some strategic efforts, they do not bet an entire half’s contribution on large, unproven efforts that might never ship. When the team wants to invest in a whole new experience, they typically ask a more traditional product team to build it, and then invest growth resources in it once it ships and shows promise.
Focus on People
Reduce friction in flows, but be careful not to go too far. One of the tried-and-true growth tips at Facebook is to taking friction out of flows to increase conversion. When we first started offering sellers the ability to ship their products in Marketplace, we had a pretty complex onboarding experience where the seller was required to enter their banking information for payout before they had listed a single item. After we saw limited seller signups we decided to make this information optional in the onboarding flow (deferring it until later) and saw big gains in seller activation. But you have to be careful not to go too far in reducing friction. A few years ago we made it easy for buyers to indicate their interest in a product by clicking just one button. While this was a large metrics win there were some people clicking the button accidentally or not understanding what it meant. We ended up having to further iterate on this feature to increase comprehension, which cost us some growth but was better for people overall.
Listen to consumer reports and take action. It can be hard work to listen to customer feedback, particularly when you’ve got millions of people using the product. But looking into this feedback and being responsive can yield customer experience wins and growth wins together. When Marketplace started we had some pretty restrictive rules about who was eligible to use it in order to avoid fraudsters before our integrity flows were fully-featured. Then Marketplace grew dramatically over the years, and we never re-reviewed those rules. We started getting a lot of feedback from people who really wanted to use Marketplace and couldn’t. Initially it was easy to dismiss as “yes, that’s to prevent fraud,” but closer inspection revealed that we could actually update certain rules to let more people have access while continuing to mitigate fraud. This led to a win for both the consumer experience and Marketplace Growth.
No idea is too small to test if a lot of people use that flow. In Marketplace, we learned that sometimes people miss out on opportunities to buy or sell items just because they didn’t see a message. We made a small change where we added an administrative prompt to the message thread if the seller did not reply to the buyer’s message after many hours, noting that the buyer was still awaiting the response. This small tweak drove significant growth in transactions.
Use opportunity sizing to make decisions on what to test. In the prior example the experience changes we made were tiny. But they had a significant impact because these flows have a lot of traffic. You should make sure that opportunity sizing is a required part of your growth roadmapping vs. prioritizing ideas based primarily on intuition. In the absence of sizing, the team ends up focusing on the most recent idea suggested or the most interesting change to build vs. the one that would have the biggest impact. Growth roadmaps are built in a spreadsheet, not in a PowerPoint or a Google Doc.
Develop the right infrastructure for testing. Most companies interested in growth already know they need to do A/B testing, and clearly that’s critical. However, something that people forget about is creating long-term holdouts. Facebook creates long-term holdouts for key features and promotional channels, while being clear not to degrade the product experience. For years we had a holdout where 1% of people globally couldn’t access Marketplace. While we were growing the product and making investment cases within the company this holdout was critical for proving that Marketplace had a positive impact on growing overall Facebook daily active visitors. Over time we opened up the holdout a bit to allow all eligible people access via a bookmark to avoid bad experiences. But we still maintain long-term holdouts where all our promotional growth levers are turned off, and where each growth lever is turned off individually. This allows us to use data to understand where we are getting the biggest growth wins, and to observe the changes in the productivity of these channels over time.
Test the maximum product change that is shippable. If you test small changes one at a time you will move the needle in small, sometimes non-stat significant ways. But if you test the maximum change in the experience that you would be willing to put out into the world and it doesn’t move the needle, you can close the door on that change and move on. Rather than taking months to make changes you can make one massive change and understand the maximum difference you can make.
Think about edge cases. We made a product change where we used a machine learning model to select which photo to make default in a Marketplace listing versus using the first one entered by the seller. This change grew clickthrough rates and product detail page views and it was one of the biggest improvements in our metrics that half. But in one category, art in home decor, we received reports that sellers were posting all of their art not showing the image first. That seemed odd, so we looked into it. It turned out that our auto-selector picked the photos of the back of the wall art, since if those were shown people would click on them to reveal the front, artificially boosting clickthrough rate. As a result, if you went to search for art on Marketplace, the feed was full of images of the backs of artwork. This particular issue was tricky to catch in testing, but thankfully we were able to quickly react based on consumer reports and resolve this edge case.
A great growth team can be a huge benefit across any product with consumer engagement, but growth is a skill like any other; it’s something you learn, iterate, and then integrate into your product.