title icon

Our approach

Behavioral data analysis

First, we consider the problems and objectives of your company. We review data and add additional relevant data, often times including unstructured data (i.e. data produced by humans for humans, such as articles, social media posts, blogs and research reports). When appropriate, we complement this data with a user opinion survey. We usually find qualitative feedback most useful, and since this is ‘normal’ unstructured data, we love to work with it. Then, we use behavioral insights and run association analysis. After running an in depth data analysis, we understand the factors that are relevant to accomplishing your business’ objectives. In short, we use real data to provide actionable advice for your distinctive company.

Finding the Sweet Spot

Using real world data, we map the customer journey. By applying our knowledge of behavioral economics to this information, we generate an evidence informed hypothesis to improve your product and services. Then, we give you an EXACT definition of the variant that must be tested. We eliminate distracting and fruitless variables from your test, in the process preventing costly and unnecessary testing.

Testing the variants

We realize that what matters most to you is conversion rates, margins and ROI. But these measures move slowly and are influenced by many factors. Because of this, we design a unique set of metrics to assess our tests. This allows us to evaluate whether the adjustments we tested produce the desired results (i.e. confirms the hypothesis).

Most likely, you have heard about A/B testing. Using our specialties, we are able to apply A/B testing to any part of your operation (bid marketing, HR, web design, etc.). Doing so allows us to propose better solutions to increase the overall value for your company.

A/B and Multivariate Testing

Typical A/B testing compares the outcome of two variants. Traditional examples include  testing two web layouts, or two different colored call-to-action buttons on your website. Conversely, multivariate testing evaluates the performance of a much larger number of variables at the same time, providing insight into which combination delivers the best results.  Today, there are several online platforms that help you perform various A/B and multivariate tests. To make the most of your testing we will tailor the hypothesis to be examined based on in-depth research of behavioral studies. Our extensive behavioral insight allows us to propose sound hypotheses based in available evidence. Often, it is wasteful to run tests on all your traffic; usually, it makes sense to analyze only a fraction of it. This test must be designed as shrewdly as possible for it to maintain statistical significance. In addition, we are able to run pre-tests in our unique digital laboratory facilities when necessary. If you already know the critical points that need to be addressed and improved, we make sure it is done in the most efficient and effective way.

“It’s about coming up with your hypothesis first and then developing a test from there. Rather than saying, ‘I want a test button created,’ or ‘I want to try two different layouts,’ you create a hypothesis first, for example, ‘I think green buttons attract more attention and therefore will have a higher clickthrough rate.”

—Michael Burk, Senior Online Product Manager, Electronic Arts (EA)

Testing for non-Internet businesses

Even if your company is a non-Internet business, it still makes sense to use our data driven testing approach. Virtually every company has some online presence, be it a simple web-page or a regular newsletter. Even in B2B settings, people make decisions; managers are prone to the same behavioral biases as everyone else. Thus, there are always opportunities to apply behavioral insights and tests to your business, from issues of employee motivation to product design.

Contact us