A Framework For Running Design Experiments To Improve Conversions

A Framework For Running Design Experiments To Improve Conversions

A Framework For Running Design Experiments To Improve Conversions

If you want to start using design to optimize your website or product, we applaud you. This is the first step in increasing conversion rates, which often times is directly correlated to the amount of revenue you are generating. Though you can get very technical with this, we will give you some high level pieces of advice before sharing a framework for running your design experiments. For the purpose of this post, we are going to make the assumption that you run an e-commerce site, have already set up engagement tracking software throughout your site, and have already collected a baseline of data. 

Smart Simple

You don’t need to do anything drastic. Making simple changes like the color of a button, the image of a product, or even changing the font weight of a call to action, are all great starting points to test. Sometimes the most basic of changes can make the biggest impact. 

Be Incremental

An important aspect of running these tests, is knowing what exactly causes a change in user behavior. If you completely redo an entire homepage all at once, you will not know what changes worked, and which ones didn’t. Keep things simple and change just a single variable at a time. If you do this, you’ll get a much clearer picture of what’s actually working, and be able to apply the awesome findings to other parts of your website. 

Be Scaleable

The changes you make will change the percentage of conversions you have – for better or worse. The more visitors you have on your site, the more impact these changes have. If you can apply the optimized solutions to different areas of your site, company, and beyond, you can realize even more value from them. 

A Framework For Testing Design Changes

Now that we know to keep things simple, incremental (and trackable), and scalable, let’s jump into a simple framework your organization can adopt to run tests and optimize your website through design changes. 

How to run tests will be the same across any industry, but the tests you run will vary depending on the type of business you run, and what you are optimizing (a website, a mobile app, a physical store, a restaurant menu, or anything). The key to success here is to establish a repeatable framework that will allow you to easily make apples-to-apples comparisons. 

Start With An Insight

Using the baseline of data you’ve collected on the example e-commerce site we mentioned above, you can pretty easily pinpoint a few problem areas on your site. Your site, essentially, is a sales funnel. Every interaction, from the second they land on the site, to the point of purchase, is a step in that funnel. Let’s say you have a 5 step funnel (A – E), and at Step C, there is a spike in people leaving the site, not making it to Step D. This is a great area to start your testing. 

Create Hypotheses

You now know there is an area on your site that is not performing well. Why do you think that is? You can conduct user interviews, discuss amongst your team, or research industry trends to help you form a hypothesis. 

Generate Ideas

You now need to brainstorm all the ways in which your hypothesis can be tested. Don’t be afraid of coming up with bad or crazy ideas. If the hypothesis is that the product page is not intuitive enough, and the user doesn’t fully understand what they will get, you should generate ideas on how to make that product page crystal clear. Maybe adding an unboxing video of the product, a more detailed product description, or some other innovative ideas to see what helps solve the hypothesized issue. 

Prioritize Your Ideas

In order to know what changes are making the best impact on your conversions, you need to tweak one variable at a time, so as to avoid creating unclear data. Try prioritizing based off which ideas you anticipate will take the least amount of time to implement, that also have the greatest impact potential. 

Create A Rapid Testing Process

The process for running a test should not take months – remember, the key is to test fast and test often. Ideally, you can build, test, analyze, then deploy over the course of a day to a couple weeks. Once you have your process, stick to it to ensure you have accurate data across all tests. 

There are some great A/B testing tools available that can help you quickly make testable tweaks to your site, collect data, and then analyze it. 

Test Small Changes

This framework works best with small changes. And you will find a greater ROI from your changes when you focus on small changes vs massive overhauls and complex feature additions that take months to develop and test. Not only do the smaller tests take less time and money to implement, they also allow you to run more tests, leading to greater increases in conversion over time.


Do it over, and over, and over again! The more you test, the better your site will be, and the better you will understand your customers.