What I Learnt About Funnels

Every product has a funnel, a series of steps customers go through to use the product and pay you money. Here’s what I learnt about the funnel:

First, if you can’t immediately tell me the stages in your funnel, it means you don’t have a good handle on your business. For example, our app, Futurecam, has the following funnel:

App store impression (someone sees the app in the store) > Download > First open > Finish onboarding > Grant camera and photos permission > Capture > Capture properly > Capture enough that some modes lock > In-app purchase

Notice the difference between Capture and Capture Properly. Our app has multiple features, some of which need a tripod. Others are designed for use at night. If the user doesn’t use these properly, he doesn’t get a good photo. This is why we track Captured and Captured Properly as different stages in our funnel.

Our app offers three free photos in each mode: three light trails, three timelapses, three bursts, etc. After that, the mode locks. This is the “Captured enough that some modes lock” stage in the funnel. To capture more, an in-app purchase is required, which is the last stage in the funnel.

Notice that every product has a slightly different funnel. You should know yours like the back of your hand.

Second, you lose a lot of users at each stage. This can make your business unviable, since if 1000 people download your app, a third don’t open it at all after downloading, 85% of the remaining don’t open your app a second time, 2% of the remaining people start a free trial, and half of them convert, that means that just 1of 1000 people are paying you. And these 1000 people were hard to acquire in the first place. This is enough to kill many startups. Startups live and die by their funnel.

Third, collect analytics at each stage. Otherwise, you don’t know how your business is doing, or where things are going wrong, for you to fix. Before I was rigorous about this, I used to jump to a conclusion like “I think things are not working out because our permission flow is not great”, spend a lot of time improving it, and find no improvement. When I looked again, I found that we weren’t losing users there, so improving it is not worthwhile. Instead, you might be losing people in the onboarding, so you should work on that.

Not only should you collect analytics, but you should be able to see how many users are at each stage in your funnel, at a glance, in your dashboard without having to run an SQL query, or do any kind of filtering or calculation. Because people won’t do that consistently, not even ourselves in the future, no matter what we tell ourselves. So just have the data available at a glance.

Fourth, look up benchmarks for attrition at each stage. For example, I read that a third of people who download an app don’t open it even once. If that’s the case for you, don’t be concerned thinking something is really wrong and stop what you’re doing to investigate it.

Fifth, no startup has perfect metrics at each stage of the funnel, so don’t worry about that. Improve whatever you can, yes, but worry, no.

Sixth, you may have some stages where you don’t see significant attrition. When you receive feedback on these, ignore them. For example, 96% of our users granted the permissions necessary for our app to work. I occasionally receive feedback saying, “Why don’t you postpone asking for photos permission till I save the photo? That would be a better UX” My answer is, “It’s not a problem to be solved.”

Seventh, spend some time trying to improve each stage of the funnel where you have significant loss, say 20% or more. Typically, a little goes a long way. Which means that instead of spending a month trying to improve one stage of the funnel, you typically get more benefit by spending a week each on four different stages. If you try to improve all stages of your funnel where you have problems, you are likely to succeed at at least some.

Eighth, metrics at earlier stages of the funnel are called vanity metrics. Don’t pay attention to them, except to diagnose and fix problems in your funnel. Don’t share these metrics on your Slack saying, “Hey guys, we have 10x the downloads today!” Unless it translates into a meanginful metric, like photos captured properly, this is irrelevant. For example, when we made our app free to download, a deals newsletter talked about us, and we had 3000 users overnight, hardly any of whom used the app or paid. They were just noise. So it’s easy to be misled tracking vanity metrics. Even if we think we’re logical, we get carried away. Plus there’s limited space in our brains to track metrics, and you don’t want vanity metrics to eject the real ones from your mental cache.

Ninth, have a stage in your funnel that corresponds to an event where the user sees value. For an e-commerce site, that would be placing an order. For a chat app, it would be sending a message. For an image editor, it would be editing an image and saving it.

But sometimes this is more subtle. In Futurecam, the action that generates value for the user is capturing something properly. Some of our modes require a tripod. If the user is using such a mode, he should use a tripod. If a mode is meant for night photography, the user should use it at night. If a mode requires capturing for at least a minute to produce a good result, he should capture for at least a minute. This is the Captured Properly event we talked about earlier.

This is the point at which a user who installed Futurecam gets something useful out of it.

Needless to say, this is a key event to track.

Tenth, the number you’ll have the bottom of the funnel is the product of the percentages of people that remain after each stage. If 50% of people from the first stage move to the second, and 25% of people from the second stage move to the third, and 10% of people from the third stage move to the final one, then you have 50% * 25% * 10% which is 1% of the users who pass through your funnel.

You can use this information in various ways. If you need 30 people every day to buy your app to break even, that means you need 3000 people to download it every day, so you can plan your marketing efforts with that target in mind.

This percentage also tells you why paid marketing doesn’t work for early-stage startups. If you make $5 per user, since 1% of users pass through your funnel successfully, you can’t afford to spend more than 5 cents per user. You can’t do any kind of paid marketing, whether ads or influencers, for such a low CAC.

Since what matters is the product of the success rate at each stage, if you have a 25% rate at a certain stage in your funnel, and you improve it to 50%, that has the same effect as getting twice as many people into your funnel in the first place. Now you can think of how easy each of them is, and do the easier thing. This way, you can make an intelligent decision between getting more people into your funnel vs improving your funnel.

What did you learn about funnels? Leave a comment below.

Also read What I Learnt About Effective Use of Analytics.

On-demand Leader. Earlier: IIT | Google | Solopreneur | Founder | CTO | Advisor