As part of our SEJ interview series, Paul Ryazanov of Promodo joins us to discuss how you can minimize your marketing budget with the help of A/B testing.
A/B testing is a way to determine whether or not your website is providing a good user experience. By learning more about how users interact with your site, and comparing the data, you’ll know exactly where to spend the majority of your marketing budget.
Here are some key takeaways from the video:
What You Should Test on Your Website
John:
Hi, I’m John Rampton with Search Engine Journal here with Paul Ryazanov from Promodo. Today, we’re going to be talking about A/B testing. So first question — what exactly are you testing?
Paul:
We test everything, basically. We test the user interface, how people interact with the site, and how they perform different actions. We analyze heat maps, Google Analytics data, and the types of traffic a website receives.
The goal is to understand user behavior and identify what works best for improving performance and conversions.
Why Business Owners Should Use A/B Testing
John:
I’m a business owner — why should I even bother doing this?
Paul:
You absolutely should do it because A/B testing helps you optimize your marketing spend.
You want to minimize your marketing budget while getting the best results possible. A/B testing helps you compare different versions of a page or experience so you can see which one converts better.
By identifying the winner, you can focus all your traffic and marketing efforts on the version that performs best.
Easy A/B Testing Tools for Small Businesses
John:
Let’s say I’m a business owner with a small budget. What are two simple tools I can use to A/B test my website?
Paul:
One tool I recommend is Optimizely.
It’s very easy to use. You basically add a JavaScript snippet to your website, and then you can start running experiments. You test different versions of a page and send traffic to both versions.
Once you gather enough data, you can determine the winner and move forward with the best-performing version.
How to Split Traffic During A/B Tests
John:
If I have enough traffic, how should I divide it? Is there a recommended number of clicks or impressions?
Paul:
When we test something like new landing pages, we usually split traffic 50/50 between the two versions.
As results come in, the system can dynamically shift more traffic toward the page that performs better.
However, if you already have a live interface that people actively use, I don’t recommend sending all traffic into the test immediately. Instead, start with a small percentage of traffic to see how the new version performs before making a full switch.
What Elements Should You A/B Test?
John:
What exactly should I be testing on my site? Buttons? Pages? What should I focus on?
Paul:
One of the first things I recommend testing is the header messaging, especially the text above the fold.
This is the first thing users see, so it needs to communicate your value clearly.
Another important area is the call-to-action message. Instead of something generic like “Sign up now,” try messaging that explains the benefit, for example:
“Sign up now to get expert advice on medical treatment.”
This helps users understand exactly what they’ll get.
You should also test:
Sometimes, simply replacing one image with another can significantly increase conversions.
How Often You Should Run A/B Tests
John:
How often should I be testing? Is this something I should always be doing?
Paul:
Many marketers say “always be testing,” and that’s generally true.
But it also depends on your business size and resources.
If you’re a well-established brand, I would recommend running five or six tests per month.
If you’re a new or smaller business, you might run one or two tests, pick the winning version, and stick with it for a while until you have more resources to experiment again.
Testing requires time and expertise, so you need to balance experimentation with available resources.
Using Heat Maps to Understand User Behavior
John:
You mentioned another tool earlier, besides A/B testing platforms.
Paul:
Yes. One tool I really like for analyzing user behavior is Yandex Metrica.
You can find it at metric.yandex.com. It’s not as well-known in the US, but it’s very powerful.
It provides heat maps, click tracking, and another feature that’s extremely valuable: session recording.
Why Session Recordings Are Powerful
John:
How does that help compared to heat maps?
Paul:
Heat maps are useful, but they require a lot of traffic to become meaningful.
With session recordings, you can actually watch how individual users interact with your site.
The tool records every visit and lets you replay it like a video. You can see where users click, how they move around the page, and where they get stuck.
This is especially useful if your website doesn’t have a lot of traffic yet.
You can observe user behavior, form hypotheses, and then run A/B tests based on what you learn.
The Basic Analytics Stack for Website Testing
John:
So if someone is just starting, what tools should they have?
Paul:
The basic stack I recommend includes:
With those tools, you can analyze user behavior, run experiments, and continuously improve your website.
