Defining Your A/B Testing Goals

Do you ever make a small change or update to your website and wonder what impact it will have on your visitors?

Will it help increase sales?

Will it improve the usability of your site?

By defining goals for your A/B testing, mapping out user flows and running high-impact tests, you can acquire data to back up your changes and provide answers to your questions.

Follow these three steps for effective A/B testing.

1. Define your goals

To start A/B testing you need to define what success means for you online.

In ecommerce there are several areas that companies often define as success. For example:

• A customer completing a sale
• Reducing basket abandonment
• Increasing mailing list sign-ups

2. Map out user flow

Map out the user flow that’s involved in each of your goals. Take a note of the pages that a user must view.

For example, a customer completing a sale might involve the following sequence of steps:

1. A visitor lands on your home page.
2. Views offers.
3. Views product page.
4. Adds to basket.
5. Goes to checkout.
6. Logs in.
7. Enters address.
8. Enters postage details.
9. Pays.
10. Receives an order confirmation.

A customer abandoning their basket at the login page would then involve this sequence:

1. A visitor lands on your home page.
2. Views offers.
3. Views product page.
4. Adds to basket.
5. Goes to checkout.
6. Logs in.
7. Leaves site.
8. Receives dropped basket email.

3. Run tests

Use the user-flow steps to run some high-impact tests. Test the steps on your site and note any ideas for areas that you could change or improve. You can set up visualisation funnels in Google Analytics to monitor traffic and drop-outs at each stage.

For example, using the visualisation funnel you may see a large drop-out of customers leaving your site at the log-in stage, or checking out as a guest. You could run an A/B test to examine how to reduce the drop-outs and to see what has the biggest impact on increasing conversions.

Say for example that you have an of average 500 visitors leaving the site at this stage each month. You could aim to reduce this figure by 10%. If you predict that a high percentage of customers going to the next stage will complete the sale with an average order value of £100, then your overall figures could be:

• Monthly: 50 (customers) x £100 (average order value) = £5000
• Yearly: £5000 x 12 = £60,000

So by running one A/B test you could significantly increase your overall sales.

There could be a few reasons for the drop-outs, for example:

• Customers have too many choices
• Customers have forgotten their login details

Here is an example of how you could change your website and test the change in an A/B test:

Option A — Standard IRP login page

Option B — Simplified login page with fewer options

Summary

• A/B testing takes the guesswork out of any changes to your website.
• Defining your goals is essential for effective A/B testing.
• When you run A/B tests on your site you acquire instant results and feedback. You can see which option resulted in more or less conversions. You can see which version of your site the customer preferred and which allowed them to convert more easily.
• Running a single A/B test could potentially result in a significant increase in your overall sales.