How to Design Experimentation Into Your Marketing Program

One of the challenges we run into the most with the CMOs and entrepreneurs we talk to is the challenge of having the time and/or resources it takes to build an effective experimentation program. Our clients know they need to be experimenting with different traction channels or distribution strategies, but how does anyone have time to develop the plans, build the experiments and track the results in the midst of other business priorities? 

Start With One Thing

You may have just finished reading Running Lean or Growth Hacker Marketing and you're ready to upend everything. When we train organizations on our rapid experimentation framework, we actually suggest starting with 1 thing. 1 small test. Starting small does a few smart things for you: 

  1. Gives you a laser focus: Sure, you could take on all the UX challenges in your organization, map out the tests and create dozens of tests at one time. But if you're just getting started, you will learn a lot from focusing on 1 challenge or test and being laser focused on how to execute one experiment to determine a direction. Keeping track of multiple experiments at once can be a challenge for even the most seasoned growth scientists (ask us how we know). It's much better to start small.
  2. Helps you establish process: Running a marketing experimentation team is much different than running a typical marketing team. In fact, we've learned much of our processes from our agile software development teams. Agile is not the only way to organize your work processes, but for experimentation projects that go fast, that need adjustment and/or constant monitoring, agile works well. 
  3. Isolates your data for easy analysis: Data management can be one of the trickiest parts of running a marketing experimentation program. Unless you can track a journey from start to finish, it will be hard to determine the success of any given test. Sometimes, if we can't get a hold of an end-to-end data set, we'll design a test outside of the existing data organization. For example, sometimes we can design an experiment with external landing page or marketing automation systems. Depending on what you're trying to learn, there are a million ways to test outside of your organization's primary site.

Where to Find Your First Experiment

The next question we always get when we teach teams our system is 'where do we start?' It can be a challenge to find the 1 test you want to run, so we like to apply a scoring system to help teams see which test to start with. 

Step 1: Get your team in a room and brainstorm all your testing ideas. Don't hold back. You can test product features, messaging strategies, landing page conversions, email marketing tactics, growth hacks...whatever! Use sticky notes and fill the wall.

Step 2: Take one section of the wall and make the grid below (we use duct tape). Group your sticky notes in the quadrants and then you can evaluate which tests you want to tackle first.

Step 3: Armed with a grouping of high impact, low level of effort tests, you're ready to decide how you'll test, how you'll track data and how long your test should run. 

Starting with 1 simple test will help you and your team get into a rhythm of experimentation that you will start to see the results of in your business.