How A/B Testing Helps Lift Form Page Conversions Rates
The lead-gen form. The necessary evil of doing business on the internet.
But they are necessary, for both company and customer. The customer needs your service; you need that customer. The form is how you meet.
What’s going wrong with lead-gen forms is the ask-to-value ratio of most of these forms. Most ask for a ton of information—name, email, company, team size, firstborn’s blood type—but give the lead little value in return.
To make them work, you have to tweak this balance. You need to give the lead more value from the form. The more value you provide them, the more you can ask from them. Phrasing form conversions like this switches this to a pure experimentation problem. You need to find out what your leads value. To do that, you need to gather information about how they interact with your site, what matters to them, and test different options with your leads, iterating to get great leads for you and great value for them.
How Ylopo went from transactional to personal with their forms
“Our registration form on these home search websites was very transactional. It was the equivalent of me saying to you ‘Hey, Ray, welcome to the home search website that you just clicked off a Google Ad, can I have your social security number, please?’”
That’s Brady Gillerlain, Lead Product Manager at Real Estate Tech company Ylopo, during our webinar on putting customer goals first. Everyone has seen a form like this. Always in a right-hand column, always a ton of questions, and always zero reasons for a lead to fill it out. Bad forms like this lead to two problems:
- The friction is too high to fill them in, so you get fewer leads.
- For those that do, the form is so generic the leads are lower quality.
A potential customer who clicked through from that Google Ad with high intent will bounce because they can’t get what they want.
What Ylopo did is what a huge amount of companies should do–reorient their lead forms around what the lead wants and their preferences. In their case, it was about shifting the form to help them look for the right home, and personalizing the form to the customers’ journey. But the specifics of what they did are less important than how they did it: Experimentation.
Using a combination of Freshpaint and Mixpanel, Ypolo used their customers’ data to understand how they were using the site and what mattered most to build up a picture of the customer experience.
With Freshpaint, Ylopo was able to do this quickly using the concept of Minimum Viable Analytics. They got Freshpaint so they could immediately start gathering data on customer actions across the platform. Then they were able to share this data with Mixpanel to visualize and analyze.
This initial step meant that any subsequent experimentation was grounded in data, rather than solely relying on gut. This is as important in conversion experiments as it is in activation experiments. Brady could use that data to define his A/B tests.
A/B tests are when you split your site visitors into two cohorts, and show each different variations of your site (or product or platform). Cohort A might see one piece of copy on the homepage, whereas cohort B will see another. You can then see which cohort has a higher conversion rate (or activation rate or retention rate) and then push that version of the site to all users, leading to higher conversion rates overall. Then start on the next experiment. Done correctly, A/B testing can give you a statistically significant underpinning of the changes you make to your site.
For Brady at Ylopo, the emphasis was around A/B testing different registration forms until they found one that 6X’ed their initial lead registration count and improved lead quality by 20%. This testing led them to a form that was centered on the leads:
[The leads] wanted a bespoke search experience that was kind of tailored to their preferences and what they were actually looking for in a home. So we built that directly into the registration form to make it more personalized and more dynamic to their responses so that at the end they get exactly the kind of search results that they were telling us that they wanted from the get go.
And what happened? A better form was better not just for the customer, but also for the company:
When you actually ask for the PII it's a much lower barrier to entry and the people who didn't want to register dropped off. Now you have these very high intent leads that are coming through our registration funnel along with the wealth of preference data that you can send to the client for when they pick up the phone and call that lead.
All this starts by making it easier to run experimentation. Once you can easily get the data, you can dedicate your time to analysis, testing, and improvement.
How you can create better experiences for your customers with Freshpaint, Mixpanel, and A/B testing
Let’s run through how you can set up something similar on your own site or product. In this scenario there are three key parts:
- Automated data collection through Freshpaint
- Data analysis through Mixpanel
- Experimental testing through an A/B testing platform, such as Taplytics (though you can do this manually).
Data collection
This is the easiest part. You can follow this guide to set up Freshpaint’s Autotrack on your site. Auotrack allows you to start tracking every user interaction on your site immediately without having to wait for engineering to instrument every single component of your site. This is extremely useful when iterating on customer experience results in frequent changes to your UX.
Once installed, you can use the Visual Tagger to name and configure events then send them to Mixpanel and any other destinations.
Here we have a simple form that asks for an email address with a single button. But your form could have multiple input fields and buttons.
To connect mixpanel, you can go to Destinations -> Mixpanel -> Configure Mixpanel and add your credentials:
Then all you have to do is head back to your library and switch on Mixpanel for the events you want to share:
Notice that if you’ve had Freshpaint running for a while before you start using Mixpanel, you can backfill all your previous data so you can run the analysis right away.
Data analysis
In Mixpanel, you can create an insights report to start to understand how your users are interacting with your site.
Here we just see the button submissions, but a more sophisticated form would have more events to analyze. Within Mixpanel, you’re looking for what users are doing on your site, and in particular, on your form. You can start to develop a conversion thesis around why people are or aren’t converting.
In this case, our thesis might be that the copy around the email submission form isn’t very compelling–it doesn’t tell people how this is going to be good for them, the value of it. But with more data collection and analysis, you can create a thesis backed entirely by data.
A/B testing
Once we have our thesis, we can start experimenting. There are many automated A/B testing platforms available, including Taplytics which integrates with Freshpaint. Ultimately, what you would do is:
- Set up two variations of the form you wanted to test. The first would be the control: how it is right now. The second is the test, a change you think will improve conversions. In our case that might be changing the copy to “Get your analytics report when you enter your email”
- The A/B platform would then randomly assign your visitors to one of these two conditions, taking care to make sure that each is seen by an equal number of people. They would route your traffic to the different versions, then measure the conversion rate of each.
- After you have reached your sample size (which you can compute manually, or have the platform do it), one variation will be declared the winner:
Then you can push that version live throughout your entire site, and start testing on improvements to that version.
Rapid data means more experimentation
You can see another version of the Ylopo story playing out. In that story they don’t get the data quickly, it takes weeks to instrument the registration forms, weeks to gather data, and then weeks to iterate on their experiment. They don’t see the step change in their business. No increased conversion rates, no high-quality leads.
Having quick access to the data from your product about how exactly your users are using your product makes everything downstream easier. With that, PMs are freed to do what they need to do–analyze and experiment–to deliver better results for the company and better experience for their customers.
That’s what Brady did at Ylopo. That’s what you can do as well.
