Building a Growth Engine

Justin Hilliard
7 min readMay 25, 2020

So a product’s been built. The products demonstrated product market fit. Now What? It’s time to grow!

Who am I? I’m a data and growth guy. I started my career doing growth at a Series A start-up, then moved to Sonos to be a data scientist in the go-to-market org. I currently work on growth at Facebook.

Growth is tricky. Scaling a product, while maintaining quality and efficiently deploying growth capital is not as simple as burning more dollars 💸 on, the all to well known, ads market. Building an efficient growth engine requires a solid grasp of 3 things. Goal Metrics, Funnel Metrics, and a viral strategy.

Goal Metrics

Goal metrics are different for every business, but they ultimately boil down to 4 different kinds of metrics. To illustrate the point, we are going to use someone throwing a party as an analogy.

The first is a Top-of-Funnel metric. The Top-of-Funnel metric is a measure of awareness. If you are throwing a party, this would be the number of people who knew there was a party going on. The invitees are aware of the location, time, and theme. The goal of this metric is to get a high level sense of the number of people who are aware of the party. This metric is a bit of a vanity metric, but allows you to track your potential TAM that could convert. For tech businesses, this metric is usually website traffic. Website traffic is defined by the unique number of users who visit a site on any given day. For consumer goods, this is qualified traffic. Qualified traffic is traffic of people considering a product. So maybe people visiting the shop or a product description page.

Frankly, growing the top-of-funnel metric is the hardest and least certain metric to grow. It is actually the metric you probably want to grow last. After all, there is no sense in pouring water into a leaky bucket. A few growth channels include media and press, word of mouth, and ads. Despite what it may seem, all of these can be choreographed. When ready, A good PR contact can help get placements. Word of mouth can be increase by really thinking through your viral loop ( well get into this later).

The second is a goal metric. This metric is what you are trying to get the user to do. This metric is usually directly tied to revenue and is therefore your most important metric. Continuing with the party analogy, this is the number of people you get to show up to your party. This is the main growth goal. Once a user is aware of your product, what are you driving them to do. If your product truly has product market fit (as measured by a healthy J-curve), this metric will make or break your product. After all, one can send tons of invitations to a party, but if no one shows up, did the party really happen? For tech business, this is the main content generation action. Did the user add a friend? Did they make a post? Did they make or join a stream? For consumer products, this is the purchase of the product.

Growing this metric is actually pretty easy. It is all about funnel arbitrage. Map out the user funnel then close the gaps. There are many proven methods to bump this metric. The first is decreasing friction. Decreasing friction is making something easier. You can take out unnecessary funnel steps. You can add automation for user actions. Another method is chaining. This is putting a desired action in a flow a user is already doing. For example, an app can have users specify interests in the account creation flow (pintrest) or add friends while creating an account (facebook).

The third is a quality metric. In the party analogy, this is the number of users who have fun at the party. As a product grows, companies will always get lower intent users into the funnel. It is a function of getting bigger. The goal is to try to keep the ratio of users who are taking the goal action, measured by the goal metric, to the quality action, measured by the quality metric, relatively consistent as you get expand. The quality goal metric is something that is the earliest indicator that is most correlated with long term retention. This metric should be validated with good data analysis or, where data is not available, sound logic. If you have 10 people come to a party, you want 7 to have fun. Simple as that. For tech businesses, it is the user action that demonstrates user investment. This can be did they watch a full stream? did they interact with a friend? Did they make a post?

You can largely use the same funnel arbtrage techniques to grow this metric. Beware, This is a measure of if the product has product market fit. So if this is going down significantly, your product probably needs some work, not your growth engine. Companies probably need to re-evaluate the core service offering or if you got your core service offering right. Often times users will be using your product to do thing that you don’t expect. If this is case, pivot and lean into what is working.

The last metric is a retention metric. In the party analogy, this is the number of people who come to your next party. This metric is important because you “can always buy users, but you can’t buy retention”. For tech companies, this is usually measured by a J-curve for your goal and quality metric. For consumer product companies, this is the repurchase rates.

Contrary to popular belief, there are ways to hack the retention metric. But it really all has to do with how good you are at CRM.

Funnel Metrics

Funnel metrics are important. If the goal is to get users to accomplish a task, everything can be viewed through the steps a user needs to go through to accomplish that task. IE a funnel. So why does looking through user actions through as a funnel matter? Well if you know all the necessary actions, there is clarity on what to work on next. Simply put, it allows you identify the largest area of opportunity. Therefore, when time is your most precious resource, you can effectively manage where you work. At FB, looking at a funnel to discover where you are going to spend resources is called Impact Estimation.

Impact Estimation

This is something at Facebook called impact estimation. It is the way that FB prioritizes engineering roadmaps.

It is easiest to explain it by way of example, let’s take an example sign up funnel with 3 steps. 1) click sign up button. 2) fill out the sign up form 3) confirm email. Let’s take 100 users land on the site. 50 click the sign up. 25 finish the form. 20 confirm email. What would you try to improve? For every additional user, we get to click on sign up, we get .4 accounts (20 / 50). For ever addition user to finish the form, we get .8 accounts(20/25). For every additional user to confirm email, we get 1 account (20/20).

If we think we can improve every step by at most 10%, then which step should we try to improve? If we tried to improve the number of users who click sign up, we would get 1 account ((50*.1)*.2). If we tried to improve the number of users who complete the sign up form, we would get 2 accounts((25*.1)*.8). If we tried to improve the number of users who verify their email, we would also be able to get 2 accounts ((20*1)*1).

So now we now know there is an equal impact to our goal metric if we tried to try to get users to complete the sign up form or confirm their email. How do we resolve the tie? Let’s look relative. Since the percent of users to next step is 80% (20/25) and 50% (25/50) respectively I would probably try to close the gap for the sign up form. Why? Well what if we actually got that 10% number wrong? there is more room for a positive mistake if we work on the step with more opportunity.

Viral Strategy

The third and final growth component is a viral strategy. A viral strategy is your plan to grow your viral coefficient. At FB, it is called the viral loop. At its core, it is the plan for one user action to turn into many subsequent user actions. Since the goal of a party, isn’t to get all the people in the world to show up. This is where the party analogy begins to break down, but it is basically the average number of friends each invitee brings. A main question is how are you going to leverage current user actions to get more actions on the platform.

You can do this a few ways. Some examples include social context, gamification, referral programs, social media share prompts, and other social engagement prompts. The main question is in a frictionless and non-spammy way, How can we get users, to engage their network? I’m sure you’ve heard of the amazon flywheel. This is basically the same. How can you leverage your core value proposition to help get your customers to help you grow.

--

--

Justin Hilliard
Justin Hilliard

Written by Justin Hilliard

Growth/Data Scientist @ Facebook and Sonos, ex. Banker JPM, Carnegie Mellon Grad

No responses yet