blog iPresso Marketing Automation

What Are the Control Groups?

Control groups in marketing are like control groups in medicine. One part of the groups receive a working, active substance and the other one gets a placebo. If you don’t know what control groups are, and why they are important in your daily operations, you risk your business taking a dive. Why are control groups and why they matter?

The problem with control groups is that most marketers approach them with the wrong mindset. Many of them think that you have to test everything and always. It’s not true. The base for testing is A/B tests but they are effective only when comparing results of freshly implemented processes and tools with previously used. When we treat them as a pattern and a mirror for every aspect of your company’s operations, it quickly becomes obvious is that results are worthless. Why is that?


A/B testing relies on parallel comparison of two different versions of the same tool or a process. We create two groups of users: one that will see freshly implemented solutions and a control group that will use older functionalities. The page’s algorithm draw which version of the page can bee seen by whom. Thanks to this mechanism you can directly compare scenarios and understand what works and why.

Let’s imagine a situation: you make changes to the product category’s main page. You have products that need increased sales: classic monitors and curved monitors. The older version of the page had a copy that promoted curved ones. The copy showed them as a technological breakthrough, perfect for content consumption and office work. It didn’t work so well, customers still preferred older hardware. In both cases, sales weren’t that good. In your version of the website you don’t promote one product over the other, you just list their pros and cons, showing personalized client’s approach. Now both types of monitors can reach the client – the one that likes classic solutions and one that wants to experience a new quality of office work or have creative stimulation. What approach will bring more money? This is why control groups exist.

With a statistical survey there comes two important terms: statistical significance and confidence interval. The second one is simply a probability. You can measure it by having a number between 0 and 1. In statistical science, there is an interval between 0,9 and 0,99. In other words – if the number is, let’s say, 0,84 then we are 84% sure that the survey has been done in accordance to the highest standards and methodology. It’s reliable. The confidence interval depends on the number of people that are surveyed. If both versions of the page about the pros and cons of monitors are visited by 100 people, there is not much certainty. If the number is 100,000 visitors, you can say that the A/B testing will be done right. The higher the number, the more reliable the study, and the more accurate conclusions. 

Statistical significance is an acceptable risk of making a mistake. It lets you know above which threshold of disturbance in the number of people being surveyed, the test will show preferable results to the opposed solution. If we have an old and a new version of the ‘monitor page’, then an error above 2% will point towards the success of a new version of the page. The 2% mark s widely accepted and used in statistical surveys.

Control groups are crucial in modern marketing. They are handy not only when you want to compare two versions of the same website but also in many other situations:

· With different versions of mailings, especially sales mailings

· Landing page versions

· Different versions of a copy, supporting the sales of the same product

· When comparing the cost of previous campaigns

A control group is crucial in basic situations:

· When you want to prove a point or contradict one

· When you want to check how a particular hypothesis influence a company’s revenue

· When you want to know what to improve in marketing communication or sales techniques

The mere construction of a control group is only half the battle. You have to know how to construct it correctly and how to run it.

A control group-based study should be longer than the time given for a classic campaign. You can check how the control group functions under different conditions, with multiple communication techniques, variants of the same webpage. For that you can buy a marketing automation tool and use:

· Reports about user conversions

· Reports about campaign effectiveness

· Google Analytics integration

· Integrated tools for monitoring data about the users

· Scoring – the rating of clients’ and attributes and behawior

A lot of companies run diversified marketing campaigns and wonder about different indicators. They also ask questions like:

· Do I offer customers value for free rather than charging them for it?

· What’s my return on investment (ROI)?

· How much money do I spend to get a single client?

A control group will help with all of these. How to make it and what steps should you take?

· Decide on a group you want to offer a promotion to. It can be ‘whales’, customers who buy a lot and often or regular ones, buying from time to time and spending a few dollars. The choice is yours. Everything is based on a strategy you want to take, effects you want to achieve and indicators you want to measure

· Create and send the campaign up to 90% of your potential customers. The remaining 10% will be your control group

· Use the marketing automation tool to measure the results. Focus on the bounce rate and conversion rate. Pay attention to how much money the control group spends in comparison to the rest of the clients.

A control group is a great and effective tool but there are instances when you don’t want to use it:

· When you have something important to say or give clients, like the information about a flash sale

· If the potential control group is too small. Control groups are good when you have at least 1,000 clients.

Remember about these two rules:

· The more clients you have, the smaller control group you need

· For the optimal results, design a campaign that can be easily scalable and repeatable. That way you will have more material to think about.

A control group is a basic tool for the marketer. Use it to increase the effectiveness of your campaigns. It can also be helpful when it comes to offering personalization.

Jarosław Ściślak

Add comment