How we passively increase conversions on our landing page without moving a finger

One of the hardest parts of selling a product on a landing page is communicating the idea behind the product as clearly as possible. A visitor that sees a landing page for the first time has to be convinced to stay on the page in just a couple of seconds. This is the first hurdle any landing page has to take. When that hurdle is taken and our landing page was able to land a good first impression with the potential customer, the next challenge is to effectively direct the attention of our customer towards the elements that are most interesting to him. We want him to get interested in the product, while giving an accurate description of what can be expected from the product features. This is particularly important in order to reduce chargebacks. More on that later. Now, that we have effectively directed the attention and sparked interest while clearly describing aspects of our product, we present a conversion action.

We have developed pagent to automate this process

All these considerations and even more go into designing a landing page. The combination of text and design ideas are then glued together with a lot of intuition. And when we come up with a new idea for the landing page, we run an a/b test to analyze its impact before showing it to every new visitor. This is in multiple parts subjective and time-intensive process, so we asked ourselves: How can we automate the design, content and a/b test process for our landing page?

How to automate the design, content and a/b test process for our landing page?

We couldn’t let it go and started working on a solution. We started developing an automated way for website optimization. An autonomous agent that is integrated into the page, that learns about the content of the page, learns about the product and market, and, most importantly, learns about the audience the landing page caters to. Let’s take a look step by step:

Understand the status quo of the landing page

To create new ideas for the landing page, the agent first needs to understand the current representation of the landing page. For that purpose, we scraped the whole website that the landing page belongs to. This gives a starting point and an impression of the status quo of communication style and website design.

Understand the product and the market

Utilizing the website data that we scraped before, we can now create an understanding of the product. We want to break down the description of the product to the purest form possible and then create messaging around that, which is accurately tailored to the understanding of the websites’ audience. We put emphasis on this, because we believe that it can help to significantly reduce chargebacks or churn. To gain an understanding of the market, we have to analyze the main keywords in order to find relevant competitors and adjacent products. We analyze the keywords on Google and curate a list of potential competitors and comparable products on the market, we gather information about how similar products are positioned and represented. This analysis helps to generate a brought understanding of the market.

Get an analysis of the market and a variation for you landing page

Understand the audience

Given an adequate baseline, the potentially most impactful method is understanding the audience that our website caters to or “could” cater to. If we understand the audience, we can tailor the website to their preferences. Since we are automating the whole process from generating variations to testing them, we are opening up the opportunity for a scalable approach to fine-grained personalizing for specific user groups. But that’s something for another time. To understand the audience, we analyze the reaction of the audience to different variations of design and content created by the agent. This is done with automated A/B testing. Whenever the agent suggests a new variation for the landing page, this variation is tested against the default. The more tests the agent ran, the greater its understanding of the audience becomes. And any variation that leads to a positive result during testing, is a variation that we stick with and show to all users moving forward.

Continuously improve

Now, every time an a/b test finishes on our page, we evaluate the results, give feedback to the agent, and then let the agent create a new variation for the website. Every new variation builds upon the previous tests and in that way, the page gets more and more refined over time.

With pagent.ai we have fully automated this process

Reducing chargebacks and churn right from the beginning

One thing that is especially important to us when optimizing our landing page is to optimize for high-quality conversions. That means, we want to improve the design and, in particular, the content in such a way, that those people who convert really are interested in the product and have the potential for long term customers. For example, we want to minimize the chance for misconceptions of the product by the customer, which could potentially lead to chargebacks or churn.