P L E A S E  W A I T...

Understanding the Difference between A/B & Multivariate Testing and Usability Testing

What is the difference between A/B & Multivariate Testingand Usability Testing?

Remember the last time you were involved in making a critical decision about your site? You may have faced an internal struggle to find the optimum balance between functional requirements and the business goals. You want your site to be the go-to destination for your customers and prospects. You also want your site to meet the business targets you have set and generate ROI.

What did you learn from that experience?

  • Did your users actually behave the way you thought they will?
  • Did the site generate the revenues and ROI that you were expecting?
  • Had your decision been different if there was a way by which you could actually test the impact of your decision before you implemented it?

If you responded with an emphatic “yes” to the final question, then all you will have to do is to commit resources towards “testing”!

You may have come across “testing for optimization” and of “usability testing”.But if you are juggling between deciding which form of ‘test to choose, then this post aims to answer this crucial question by demystifying both these types of tests.

Testing for Optimization

Let us say, you need to choose between two or more design variations of the same page; but you are not sure which one will be preferred by your audience.In this scenario, it would be risky to upload the wrong design.Making frequent changes to identify the design visitors prefer will be even more disastrous. This is where you would use an A/B Test. Here A/B means you have two variations of the page. So, if you had N design variations of a page, then potentially you can run an A/B/C/D/N Test. Multivariate Tests are similar to A/B Tests, but they test a higher number of variables and reveal more information about how these variables interact with one another. Testing measures and reports quantitative information like change in engagement metrics,changes in conversion rate, change in cart abandonment rates, etc.

All tests for optimization are designed as experiments and are conducted by splitting audience and exposing them to the variations. Here is a classic example of this type of experiment. Two variations of an email campaign, variant-A, and variant-B, are sent to different sets of recipients. The response generated by each variant is analyzed to choose the better option. In some cases, variant-A and variant-B are distributed equally between the two sets of recipient sets. In other cases, variant-A would be standard mailer (the control version) and be delivered to a larger set of the recipients; while variant-B (the test version) would only be delivered to a small share of the recipients. If variant-B delivers better results compared to variant-A ceteris paribus, then we can declare variant-B as the “winner of the test” and safely switch to it from the next mailing cycle.

Testing for Usability

Your visitors expect your website to be easy to use, easy to navigate and easy to find what they are looking for. The best way for you to find what is “easy”, and what is “not easy”, for your visitors is through Usability Testing (also known as User Experience Testing).

The concept behind testing for usability is like the rehearsal before the actual stage performance. In this case, your audience is not present and you can rehearse till you are certain that your show is a success. Usability tests are conducted by keeping in mind the experience that needs to be delivered to your end users. The tests involve interactions with at least five human users, whose reactions and recommendations are in essence of the test result. Usability testing deals with perceptions of your site’s visitors. The tests rely on identifying a small sample that is a good representation of your target audience. This form of testing can be a life saver as it answers questions such as:

  • Why are visitors not clicking on certain links?
  • Why are visitors navigating through paths that are different from your expectation?
  • Why the product views are not resulting in a conversion?

During the test, a usability specialist monitors, and analyzes the behavior of the “test group” to arrive at the answers to questions above.Going back to the previous example of two email variants, while A/B Testing can conclude that variant-B worked better compared to variant-A, usability testing can answer, “why” variant-B delivered better results.

Test Objectives

The goal of testing for optimization is to improve conversions, be it macro or micro goals. A site’s macro goals could be to generate sales, newsletter signups, whitepaper downloads, video views, etc. Examples of micro goals include adding products to the cart, using the internal search box, etc. Each of these goals has a specific metric associated with it. Strategies for executing A/B Tests or Multivariate Tests are designed to measure and report the impact of the tests on these metrics.

The field of Human-Computer Interaction plays a major role in usability studies. Usability testing has more permeable goals than the short-term impact of design changes. Usability testing focuses on how to draw, engage and retain the attention of people. Therefore, strategies for usability studies are designed to give decision makers an understanding of the internal responses of users to a website.

Which Type of Test?

Here’s a ready reckoner that can help decide which type of test to take up and execute.

Quantitative Qualitative
Can take long for statistically significant results Results can be summarized quickly
Conducted on part or all of the users Conducted on a representative sample
Requires initial setup and the test can run on their own without manual intervention Dependent on a facilitator, who is required throughout the test
Arrangements Arrangements
  • Scope the problem and audit available data
  • Identify test segment
  • Define test hypothesis control values
  • Create the test variants
  • Configure tools and start test
  • Analyze the test results and declare winner; if test results are not statistically significant, then repeat process with changes
  • Prepare the parameters to Test the prototype
  • Arrange for Test participants
  • Debrief Test participants about expected objectives
  • Observations by the participants
  • After the Test, Fix problems; but Test again if everything else works perfectly along with identified problems
Tools Available Tools Available
Visual Website Optimizer, Optimizely, Adobe Test & Target, Monetate Loop11, Open Hallway, TryMyUI, Qualaroo, Clicktale, Crazyegg

We hope this post was helpful clarifying the major differences between “testing for optimization” and of “usability testing”. If you need assistance with defining your test strategy and executing them, then why not get in touch with one of Nabler’s testing specialists via filling up a contact us form.So, which test should you use while making a critical decision about your site? The short answer is that you will eventually need both types of tests to make informed choices. The long answer is that the type of test you will run depends on the problem you are trying to solve. If you are trying to identify what delivers better results from the conversion perspective, then you should use an A/B Test or Multivariate Test. If your objective is to improve the experience of visitors by understanding how they interact with the site, then you should opt for usability testing.

Drive better results by understanding customer data