How to Apply Growth Hacking for Startups

You often hear about the magic hacks that made today’s startups huge, and maybe you even admired how lucky they were. Well, what if I said that it was not at all about luck but rather a rigorous method that you also can learn? Yes, I’m talking about growth hacking! Compared to what you might think, growth hacking is not about finding the one hack that will revolutionize your business but rather continuous tweaks that make you excel.

These small continuous, but regular hacks, together can have a larger effect than one accidental hack. According to Peep Laja, an expert in conversion rate optimization, 5 % monthly improvements in conversion rate will lead to an 80 % total improvement after a year.

Growth hacking is a great method for startups to grow since all decisions are based on experimentation, it is pretty foolproof and much less risky than traditional marketing and product development. Another thing that differs in growth hacking from traditional marketing, is that it includes the complete user journey, from acquiring to referring. It is not only about bringing in traffic but also making sure that those people discover your product, fall in love with it and stay as loyal users over a long time, which has a bigger impact on your bottom line rather than traffic.

It is all about learning about your users, for every new learning you can improve your product which will yield in your growth. Exactly like scientists have improved life for humans in centuries now, you will as well run experiments and make discoveries to improve the life of your users.

This article has been inspired by the book “Hacking Growth” written by the masterminds Sean Ellis and Morgan Brown, but I’ve adapted it to the needs of an early-stage startup. It also comes with a template to guide you through the disciplined process of Growth Hacking. You find it by clicking here.

The first step in your progress towards growth hacking is to make sure you have reached product/market fit. You can read more about the product/market fit in this article written by Tren Griffin. It is an unclear concept, but in short, you could say that it is about having a product that people love.

The second step will be to get your data infrastructure and reporting set. A good data infrastructure should track all of your users, the channel they came from, and the events that they pursue. To find the right tools to do this, check out my article about tools for data analysis.

Reporting should be in the form of automatic dashboards, rather than time-consuming excel sheets and ad hoc analysis. They are incredibly important to you since they allow you to be in control of what is going on in your startup and follow your historical development.

To find out what to put in your dashboards, I recommend you to check out the article “How to Get Started with Metrics in 5 Steps”. You will learn how to identify your KPIs and how to break these down into your leading indicators. For growth hacking to work, you will need to know which levers to pull, in order to move your KPIs in the right direction.

With the success of these steps, you can get serious about growth hacking. Just as a scientist, you should then move on to identify hypotheses about your users, which you will then prove or disprove. The process can be described as a loop of Analyze — Ideate — Prioritize — Test. As you start off with your tests as a small team, launch one test per week and let it run for a duration of two weeks. As your team and user base grows, you should be able to do more tests in parallel and run weekly loops to increase the speed of your discoveries. Again, to be able to keep your speed up, you will need to be organized and disciplined for it to not end up like a mess.

Let’s look through the 4 steps of growth hacking:

1. Analyze

First of all, we need to identify good assumptions to build hypotheses, this is best done by analyzing your user base or your audience.

To analyze your user base and the usage of your product you will need to categorize your users. Group together your best users, those who are loyal users coming back over and over again. Try and look at what they are doing compared to your average users, which features do they use? When and what do they visit? What are their characteristics, from which channel were they acquired, and what is their demographic background? Next, analyze users abandoning your app or website, what makes them exit and who are they?

You can also analyze your audience’s behaviour outside your product, this might be a great alternative if you lack data on user behaviour or need new ideas. Check out this article to learn how an early-stage startup can do audience research to identify hypotheses to test

2. Ideate

By now, you probably have a bunch of ideas about your users. At this step, you will organize your ideas and brainstorm experiments to test these ideas. The smaller startup and the fewer users you have — the crazier your ideas need to be. To be able to see a clear difference in an experiment, you’ll need to test drastically different versions.

Use a document to map out your experiments and follow up. The experiment name should be an easy but correct description and the assignee will be responsible for the process of the test. The assumption should be a statement that you wonder would be true concerning your users. The next section is your prediction on how your KPIs and other leading indicators will change, of course, this doesn’t have to be an exact science, but it is good to be able to benchmark when you have your results.

Next, you should specify where the test will take place so you don’t accidentally run overlapping experiments. Finally, in part 1, you’ll need to specify the time consumption, which will later help you choose and prioritize between tests.

3. Prioritize

Hopefully, you will now have a long list of tests to run. Exciting! But you will have to choose and prioritize projects — an organized way of doing this is by using the ICE method. It stands for Impact, Confidence, and Ease for which you should give a score of 1, 2, or 3 for each project.

• You score an experiment high (3) on Impact if it is expected to have an important effect on your KPIs.
• You score an experiment high (3) on Confidence if you are very certain that you will reach the expected results.
• You score an experiment high (3) on Ease if it is quick to develop and doesn’t demand any investment.

Rank your ideas and identify the winners. Select not only based on top ranking but take into account where they take place (since they run in parallel) and development time. You can then fill out the last section of part 2, the start date, and the planned duration.

4. Test

The last step will be to run the test. There are different methods for doing this but it is all about proving or disproving your assumption with as much significance as possible. The best is of course to do proper AB tests, which you can read more about how to do here, on final conversion. However, this might be tricky, not many startups have reached the milestone of several hundreds of weekly conversions.

As you await that, you can run split tests on leading indicators that have a higher frequency. In the worst case, you can analyze your performance before and after the feature was added, but this might be highly dependant on other things, such as the day of the week or different ad campaigns.

Do your best to find proof that your hypothesis is right or wrong so you can learn about your user base with confidence. Try to find a method where you can be confident that it was the change you made that was the reason behind your outcome.

Finally, you summarize the results and what you’ve learned. Put down the results you achieved on the predefined KPI and leading indicator to determine if your hypothesis was right or wrong. Finally, you will decide if the feature should be implemented or not.

After finishing this final step, you will restart the loop with analysis. With the results on hand, you will probably get many new ideas to test. If you proved the hypothesis right, then you might want to double-down to try if it works in other areas of your business as well as to maximize your learning. If you proved your hypothesis wrong, you will need to come up with new opposite ideas.

When you’ve started running these experiment loops, it is warmly recommended to do weekly meetings. Even if you’re a small team, it is good to sit down once per week and get some perspective and together analyze the results and plan the next loop.

All right, now you know all to get started, why are you still reading this? Go get growth!

This story is published in The Startup, Medium’s largest entrepreneurship publication followed by + 381,862 people.

Subscribe to receive our top stories here.