Stop guessing with your landing pages
You can follow best practices. You can look at heat maps. You can listen to peoples’ opinions and anecdotal feedback. But the only way to truly know what works and what doesn’t on you landing page is to split test.
The problem with best practices, qualitative data and even tools like heatmaps is that they leave too many variables. What works for some businesses may not work for yours. Different audiences respond differently. And clicks, views and other engagement metrics don’t necessarily equate to conversions.
Split testing takes the guesswork and assumptions out of the equation and provides measurable proof beyond a doubt (when done correctly). User actions speak louder than words, after all. And split tests allow you to isolate variables to see the effect on conversion rates.
That’s not to say that the methods above aren’t above useful. They are great for forming hypotheses. Then you should test those to prove or disprove them, even if you are confident in your assumptions. Even experts with a proven track record are sometimes surprised by test results.
You need some real traffic for split testing. A lot of indies would not see a big benefit in my opinion.
That's true, if you don't have the traffic necessary to get statistically valid results, then you're not ready for testing yet. In those cases, though, I don't think conversion rate optimization should be the priority. Those businesses should focus first on identifying and defining their target audience and achieving product-market fit, then in bringing in traffic.
Very true but nowadays you can run ads on a tailored landing page and run a test on that page to boost sessions. In that way you are proving a concept and gaining validated learnings without burning much cash or having the initial traffic to start with.
I think there are a few hacks to go around this as an indie
I agree with your advice Sean, but it is a bit like saying "the only way to truly know what car works for you is to drive them all". It's true, but it doesn't mean everyone should do it. As @strzibnyj mentioned, you need traffic. It also takes work to form good hypotheses – user interviews, for example, which you mention in your post (good read!). Time to test them. Expertise to analyze them. And so on.
Working with many solopreneurs and marketing teams that consist of 1-3 people I can guarantee that most of them should listen to best practices first and foremost. And we shouldn't be slacking them for not doing multivariate testing.
Firstly, because mistakes people make are really, really basic. Like smashing multiple CTAs, showing too many offers, not having customer journey in mind at all. Secondly, because they can "Pareto" it, and achieve 80% of results with 20% of resources. Then, with time and money, comes time for thorough testing.
But for that they should reach out to experts like yourself anyway :)
I wasn't trying to slack anybody. I wanted to grab people with the headline but perhaps it is a bit too strong. I admit that not everyone will be at a stage where testing is feasible.
I don't think this analogy works. Testing is about trying everything. You can't. There are unless options. Instead it's about proving or disproving a hypothesis. To use your analogy, it would be more like saying "the only way to know if car A or car B works better for you is to drive both" or, if you want to get more accurate, "the only way to know if this new card will work better for you than your current car is to test drive both".
True, your analogy works better. I guess we agree that listening to feedback, learning best practices, and looking at heatmaps is not an alternative to split testing, it's just an earlier step. Hence my response to pitting them against each other in this post. But I get it, it definitely grabbed my attention and got me to comment :) Thanks for the contribution!
I agree, A/B testing is not figuring out what might work. You use other tools, experience and intuition to predict what is going to work. Then with A/B testing you just test the hypothesis.
You can A/B test 100 things that are all bad, so you will never find a better solution, you might find the best one out of the bad ones though.
As an indie hacker, I do customer discovery (call potential customers as I validate the problem im solving) and use their exact words in my landing page. It’s not sexy, but it keeps me from writing stuff that I think sounds good, and literally speaks to the pain points (real or felt) that my customers have.
good advice. this is one of those things that is easy to neglect but can be so valuable.
Just FYI, the linked page renders as blank for me. I tried loading in Firefox and Safari. There's page source, but all I see is white.
Really? That's strange, it's working fine on my computer. I'll have to do some investigating to find out what's going on.
Thanks for letting me know.
I always have this in the back of my mind, but never put it into practice. This is my reminder. Thank you!
Thanks for this article. Do you have any recommendations other than VWO and Optimizely for A/B testing? We don't have enough traffic for Optimizely, I just started testing VWO and Google Optimize feels like too much work to configure. I'm looking for a lightweight affordable solution. Thanks!
It's hard to give a recommendation without knowing your situation. There are a ton of options with different features which you may or may not need or be looking for. For instance, there are a lot of solutions that are page builders with testing capabilities, which could be a positive or a negative.
I would make a list of the features and integrations that will be most important to you based on your goals. That should help to narrow things down a bit.
With your comment about Optimizely, you might also want to calculate how large of a sample size you'll need to get statistically valid results before spending money on a testing solution.
Hope that helps. If you want to discuss this in more detail, feel free to reach out to me.
This is true, but you can also A/B test with heatmaps, and see if your design changes lead the user attention where you want them to.
Also, A/B tests tell you what is working and what is not. Heatmaps tell you where the user attention is focused, so you can direct it (possibly using A/B testing) where you want it to.
True, except the purpose of a landing page isn't to get the user to pay attention to a certain area of the page. It's to get the user to take action. You could direct users to what you think is important and actually decrease conversions if it doesn't resonate or adds friction.
It's better to use heatmaps to develop your hypothesis, then test it.
I'm addicted to AB testing. And I was never satisfied with doing one test at a time, one after another. It would be amazing to hear your thoughts about an article I wrote about that: https://croct.com/blog/post/ab-tests-personalization :)
This comment was deleted 2 years ago
Yes, in general, A/B testing requires a large volume of data or a long time to run the experiment with.
Yes, testing takes time. But don't let that discourage you. You shouldn't think of it as a one-time solution but rather as an ongoing process where you iterate improvement.
This comment was deleted 2 years ago