August 30, 2019

The Growth Marketing Process: 5 Steps for Running an Experiment

Growth marketing is a data-driven method to optimize specific areas of your business through experimentation. Growth marketers typically focus their efforts on the six “pirate” metric areas, awareness, acquisition, activation, revenue, retention and referral (AAARRR). 

Here are five steps for conducting a growth marketing experiment at your company. 

Step 1: Identify Which Area Needs the Most Attention

Before you can start a growth marketing experiment, you need to have a reporting system in place that can track the impact of every change you make. You also need to have a baseline for how your business is performing in the six pirate metrics areas

Once you have that baseline data, you can identify which areas you’re doing well in and where you need to improve. Once you establish the high-level area you want to focus on, break it down into micro areas of opportunity. 

For example, a couple years ago, New Breed noticed room for improvement in our acquisition efforts: specifically, our MQL-to-SQL conversion rate was lower than it should have been. We decided to further research potential causes for the problem and develop an experiment to improve our conversions.

Download our Marketing Goals Calculator Template to find out just how many  contacts you need at each stage of the funnel.

Step 2: Design an Experiment Around Your Designated Focus Area

The first step for designing your growth marketing experiment is to develop a hypothesis. After you know what you want to change, you need to come up with ways to make that improvement. That change could be as small as rewriting a web page headline or as large as re-designing that entire page.

Determine what action you think will cause the biggest change and then craft your hypothesis by filling in these blanks: “Given [our current data], we believe taking [this action] will impact [these results] by [this much].” 

Keep the SMART goal framework in mind when crafting your hypothesis. If your hypothesis isn’t specific, measurable, attainable, relevant and timely, you won’t be able to obtain actionable results from your experiment.

Then, outline an experiment that will test your hypothesis. Your experiment will actually need to influence your hypothesis and should include some form of baseline against which you can compare your results.

For example, if you’re optimizing an existing element or page, you can A/B test the new version against the old version to see which performs better. If you’re implementing a new strategy, you can compare the data collected after that change to the data collected beforehand. 

When designing our experiment to improve our MQL-to-SQL conversion rate, we noticed that our MQL score was based on both fit and interest. That lead to people from unqualified companies becoming marked as MQLs due to their engagement with our content. 

We hypothesized that we could increase MQL-to-SQL conversions by changed our MQL criteria to only account for fit. We believed that would lead to more high-fit leads getting passed to sales, making it easier for them to convert them.

In addition to increasing our MQL-to-SQL conversion rate, we were also expecting less MQLs to be marked as unqualified.

Step 3: Implement the Experiment

Run your experiment for a designated timeframe or until your results have statistical significance. Setting guidelines in place prior to starting your experiment will help prevent you from jumping to conclusions too early. You need to be patient.

It’s normal to see drastic differences immediately after implementing a change, but those results will even out over time. 

Additionally, try to test one thing at a time. The more variables you add, the more difficult it is to conclusively identify the reason for the results you’re seeing. If you’re already running one experiment in an area, wait until it’s over before starting additional tests.

In our experiment, we changed our lead scoring system. We separated fit and interest into two separate stages of qualification. Under the new system, our leads were identified solely on interest. Then, we created a list in our CRM using all the minimum requirements to be a good fit for us. Anyone who became a member of that list was qualified as an MQL.

We conducted this experiment over the course of a year, and because we couldn’t just change our lead qualifications for just a portion of our database, we didn’t have a control but instead compared the results after the end of the year to the data collected before the experiment started. 

Step 4: Analyze Your Results

When your experiment is over, compare your results to your initial hypothesis. Did you reach your goals? 

If your experiment underperformed, did it do so poorly that you want to stop pursuing that avenue for improvement? Or, did you just see insufficient results that require further testing?

If your experiment validated your hypothesis or exceeded expectations, how can you use your findings to improve your company’s processes even further?

When analyzing your results, make sure your control or baseline is directly related to your experiment. If there’s a seasonality to the subject of your experiment, you should be comparing your results to the equivalent season. 

For example, a company that sells and services air conditioner units will probably close the most deals in the summer when it’s hottest. If they’re experimenting on their sales process during that season, then they should compare the results to sales during previous summers as opposed to the spring and winter leading up to summer. 

On top of that, you should look at trends in the industry during the duration of your experiment. If everyone in your field saw the same change, those results were probably caused by other factors. 

At the end of our experiment, we looked at the data points we had identified in our hypothesis: MQL-TO-SQL conversion rate and the number of disqualified MQLs. We saw a drastic decrease in the amount of disqualified MQLs, but the number of SQLs we were converting didn’t increase the way we had expected. 

Step 5: Start Your Next Experiment

The purpose of a growth marketing experiment is to influence a key performance indicator (KPI) metric which impacts your overall pirate metric area. If the results of your experiment sufficiently improved your company’s performance in that area, then you can start over from step 1 and identify the next area for improvement.

If your experiment didn’t get you completely to your goal, then you should start over from step 2 and look at different ways to impact your KPI or different metrics that can impact your overall performance in the area where you’re experimenting. 

Because our SQL conversion rate didn’t increase the way we wanted it to, we decided to examine other factors that could be involved, like follow-up and response time.

The Takeaway: 

Growth marketing is a methodical, data-driven way to improve the performance of a given area within your business. Growth marketing, or “growth hacking” as it’s sometimes called, has helped many companies optimize their strategies — that’s why it’s trendy.

However, just because you’re working in marketing doesn’t mean you need to do growth marketing. There are many parts of marketing that you should just be doing, and you don’t need growth marketing to tell you what they are. If you don’t have defined buyer personas, you don’t need to run an experiment to know that establishing a target audience will help you create more effective marketing materials. 

You don’t start off with growth marketing. Growth marketing comes into play after you’re already following best practices when you’re working to take all those efforts one step further.

New call-to-action

Guido Bartolacci

Guido is Head of Product and Growth Strategy for New Breed. He specializes in running in-depth demand generation programs internally while assisting account managers in running them for our clients.

cta-pat

Ready to jumpstart your acquisition, retention and expansion efforts?

Request Assessment