Finding growth opportunities in your product with data

The first step to optimizing your funnel is learning what is happening today. To do this, you should track all of the key events along the journey for a customer to discover, engage, and eventually pay you. Sounds like a lot, but it usually is four or five events for any given business. For our sneaker app example, we want to know who is downloading the app, who is opening it, who is viewing sneakers, adding them to cart, and checking out? These events would be App Downloaded, User Logged In, Viewed Product, Added to Cart, and Checked out. We have lots more resources on choosing your event names.

Analytics tools like Google Analytics, Amplitude, and Mixpanel are great for seeing these basic funnels. You can easily see the breakdown of how users are going through your funnel, and even slice and dice by attributes like the channel they came in from or their company size.

Example of a funnel in Amplitude

You've probably heard the term "leaky funnel." In an ideal world, all of your users are converting 100 percent between steps, and you don't have a triangular funnel ... it's rectangular! If you think this is you, it's highly likely your tracking is broken :).

Sadly, there is always drop-off through a funnel.

Take the above screenshot, from SearchSong to Play Song to ShareSong, nearly 25% of people fall away. This is a leak.

Your job is to figure out which are the most egregious leaks in your funnel and fix them. Get to work!

Focus in on one part of the funnel

With the complete picture of your funnel, you have to decide which leak to plug first. You may be lacking context at this point, so a common question to ask is "What's a good conversion rate?" The main thing to remember is there is always room for improvement. No matter what your conversion rate is, it can always go up.

That said, here are some areas to think through to help you prioritize which funnel step to go after.

• Reach: The higher up in the funnel you focus, the more users you can affect with your optimizations. And with more users, your optimizations will have compounding effects. For example, let's say your site gets 100,000 viewers a month, and five percent of those (5,000), sign up. If you can increase that to 10 percent, you'll have 10,000 people going through the rest of the funnel. If you focused on improving the billing flow, where only 0.05 percent (500) people hit each month, your change won't be affecting that many people.

• Company Goals: However, it's also important to layer on the metrics that matter to your company in your decisions. If your team is focused on driving up paid accounts and revenue over total user growth, then it might make sense to do some more bottom of the funnel experiments. If growth at all costs is your strategic goal, then activation and acquisition may be more important than getting people to pay.

• Common Sense: You may be looking for better advice here, but the reality is that you'll have a good feel for what might have the most impact by looking at the numbers. This gets harder as you pick off the low hanging fruit, but to start, it will be pretty obvious what are big win opportunities. For example, if you notice that nearly only five percent of the people who download your app actually sign up, that's a big opportunity.

Investigate the problem

Now that you've picked your funnel step to focus on, the next thing to do is to really dig into that problem. Why are people abandoning the signup page? Are we asking for too much information? Is the flow broken? Is it hard to interact with? Is it too bare bones to be persuasive? The great news is there are many strategies to help answer these questions, learn more about your customers, and measure product usability!

User session recordings: See where users are clicking and where they are getting stuck. Using tools like FullStory or HotJar, you can see their full sessions. At Segment, we use these all the time, often uncovering bugs or quirks in flows we didn't know existed.

Live chat: Pop up a chat window on the affected screens or pages. See what questions customers ask, or even reach out to them proactively. Consumers are pretty used to this type of hands on support from companies these days, and typically respond well to live chat. Some great live chat tools are Drift and Intercom. We've used both at Segment in our app and on our public site.

Surveys: Similar to a live chat, you can pop up little surveys at the bottom of pages to ask tailored questions about the experience. You could also email your customers a survey. Good tools include Olark, Delighted, Satismeter and Pendo, but there are many more out there, too.

Customer interviews: Often the most helpful thing to do is pull a list of customers who got stuck in that part of the funnel, and reach out to them directly to have an email chat or a quick call to learn about their experience. You should be able to export this list of customers from your analytics tool.

Cross-company knowledge swap: If you have a dedicated success or sales team, you should ask them about what they've been hearing from customers around this issue. Even search through your help desk to see if you can find some tickets that address the issue.

Measure your benchmark and set a goal

Before you dive in to creating ideas to test, you must measure your baseline. Imagine going on a diet, but not weighing yourself first. Be wise! Think ahead. You want to know if your efforts have been successful so you know what to do next, and to do this you need to know if you've moved the baseline.

What conversion rate are you starting from? Then, set a goal for what success would look like. Taking a "let's wait and see" approach may be tempting because setting goals are hard, but it will help you think clearly about what to do next. When to invest more in solving the same problem, and when to move on. Setting goals is also a great way to build momentum and focus on a team. People love to hit goals.

Brainstorm and test solutions

With the problems identified and your goals set, now it's time to brainstorm experiments to push those conversion rates up. In our sneakers app, maybe adding more social proof to the signup page will help. Or adding social login (Login with Google, Facebook or LinkedIn). Or just taking down the login altogether? Maybe the "exclusive sneaker community" we've been trying to create just isn't working, and we want all the users we can get?

A simple way to prioritize ideas is to rank each idea based on potential impact and time to execute. You'll want to crank through the ideas with high potential impact and low time investment first, then move on to others.

One way to test the impact of your ideas is to run A/B tests. Launch your new feature or experiment to a percentage of users, while keeping the experience the same for everyone else. However, don't fall into the easy trap of thinking that all improvements must be A/B tested. Getting an A/B testing tool set up, getting experiments launched, and waiting for statistical significance takes time and resources. As an example, at Segment, we almost never run A/B/ tests in our app. As a data infrastructure product, we don't have a high volume of daily users in our app. This means that running A/B tests long enough to get statistical significance often would take weeks of waiting. Time we don't have to waste. In contrast, on our signup page, we run experiments all the time. We have the volume of users, and it's such an important conversion point to optimize, that it makes a ton of sense.

Reforge has a great article outlining how to think about A/B testing and a flowchart to help you make decisions.

Google Optimize, Optimizely, Visual Website Optimizer, Leanplum, and Apptimize are all popular tools for running A/B tests.

See if your ideas worked

Once you've let your experiment run long enough, go back to your analytics tools to see if your experiments worked. Did you hit your goal for improving the conversion rate?

If you got close, you'll want to implement the change and perhaps move on to your next growth idea to increase that number. Many growth teams focus on moving up a single funnel step with a series of small experiments before moving on to the next!

Rinse and repeat

Learning the steps behind optimizing your funnel isn't hard. The trick is sticking to the process.

Image from Avinash Kaushik's blog.

You can't just use quantitative data to evaluate drop-offs and measure performance improvements. If you want to achieve truly powerful lifts, you need to talk to your customers as well. They will tell you what's going wrong and give you the best ideas for how to improve the experience. Collecting qualitative data is just as important as using a quantitative lense to approach problem solving and funnel optimization.