How to Use HubSpot Workflows to A/B Test Emails Using Automation
At Aira, we always want to improve what we are doing. That’s why we tend to test as much as we can when creating anything, from landing pages to calls to action and emails. Today, I want to talk you through a clever HubSpot hack, which allows you to A/B test whilst also automating emails.
What is A/B testing?
A/B testing is also commonly known as split testing. It helps us understand what the visitors to our website, or readers of our email resonate with so that we can improve what we share. The idea is that this leads to an increased click-through or, even better, conversion rate.
The better optimised your website, calls to action and emails, the higher the chance that those visitors convert into leads or customers.
A/B testing specifically is the practice of showing two variants of the same asset (landing page, CTA or email) to different segments of your audience at the same time and comparing which of those two drives more conversions.
Sometimes having a different colour CTA could make the difference between visitors noticing that there even is a (sign-up) form and them mindlessly scrolling past without even being aware of it.
So…what do you test?
What the results of each test look like differ between websites and also between each test. They could be very tangible in the sale of items on an ecommerce site. For B2B businesses, it could be more leads. You can even A/B test which SEO changes to your website result in an uplift to your organic search traffic.
It depends on the purpose of the asset that is subjected to the test. First, it’s important to identify the goal of the asset. Goals can be anything from email signups to purchasing on the website and clicking a certain button. This will help you set a hypothesis for the test. For example, ‘If I change the colour of my CTA, I expect more email signups.’
In this example, we are focusing on email A/B testing through automation. In HubSpot, it is possible to test with regular emails that are sent as a one-off. You can create a B version within the original email to specifically pick what you are testing for. A different colour CTA or a different subject, perhaps different wording or including an emoji might make a difference to the open rate of your email. Once it has been sent, HubSpot will record for a set time period all the results and then offer you a neat report to show you what performed better.
At the moment, HubSpot doesn't offer A/B testing for automated emails. But in order to improve what we share with our leads - from confirmation emails, to other email journeys - it’s still important to test!
Not to worry, the HubSpot Hack below will help you A/B test, even if some of the reporting is a bit more manual.
How do we create this magical HubSpot hack?
Let’s get started with the requirements. In order to set this up, we will need to have access to HubSpot Sales or Service, Pro or Enterprise. This will allow us to create lead rotation and in this case, A/B testing!
The example scenario
Your sales team rings each lead to understand their requirements and whether they are a potential fit for the company. After that, you want to send an email and find out whether a change to the call to action may increase the click through rate.
- HubSpot Custom Property, which is a ‘HubSpot User’ property. I’d highly recommend making sure that this property cannot be used in forms and to name it something that makes its function very clear. In this example, I’ve named it ‘Custom Owner - A/B Automation Use Only’. That should make it clear to anyone using the portal that this property has a specific use.
- A list, or other requirements that a contact needs to meet, in order to be enrolled in the workflow. In this case, I’ve used a static list called ‘Static List - A/B Email Automation’.
But Maret - why would you use a static list?
Contacts can be added to static lists from the contact record so when the sales executive is talking to the contact and keeping note in their contact record, they can simply add them to the static list at the end of the call.
- Two automated emails, an A and a B version. Make sure that you only test one thing at a time. Otherwise the results get muddled and we won’t be able to assess what feature works better for your readers. In this case, we will change one of the CTAs in the email and give it a different colour to see whether that increases the click-through rate (we could also change the CTA text, subject, images or body of the email).
- A workflow. The workflow brings all of the above together. It means that once everything has been set up, the workflow and the A/B test can run until you have enough data to see what works best for your readers. Let me explain a bit further below...
This is what the workflow looks like as a whole. At the top, you can see the trigger requirements. They can change to match your requirements.
Following that, you may notice that there’s a 15 minute delay. That’s to give a bit of space between the phone call and the email sent. Otherwise the email would immediately be sent and that could come across as a bit much.
This is where the secret lies in this HubSpot hack. We create lead rotation for the custom property. In this example we have used Harry and Maret as the two options, but it is a good idea to assign this to two people who don’t normally deal with contacts - people who are not in the sales team. This avoids any confusion while contacts are assigned this custom property. The lead rotation will rotate leads evenly (50/50) between the two HubSpot users.
Once the lead has been rotated, we create an if/then branch. This will help decide which email gets sent. In this example, having Harry as the custom owner property means the lead will get version A of the automated email. If the lead rotation assigned Maret to the custom owner property, the lead will get version B.
Once the email starts sending, you will get a bit of information about open rates in this workflow but you can also open the emails next to each other to go more in-depth and compare how the two are performing.
Finally, we clear the custom property, and in this case we also remove the contact from the static list. That means that the contact can be enrolled to another A/B Email Automation workflow at a later date.
We love this little hack in order to set up automated emails for our clients, and still being able to continuously test and improve on the emails that are sent out.
Have you tried implementing A/B email automation? I’d love to know how it’s working for you.