Common Conversion Rate Optimization Mistakes
Quick Summary Conversion rate optimization (CRO) is becoming an essential part of every marketer's toolbox. It's a great way to make more sales without needing to spend more money. A great CRO process can help you maximize your site's appeal and convert more visitors into customers.
However, sometimes you don't see the results you wanted, even after implementing CRO best practices and strategies. That can be frustrating and disappointing.
Still, it's not a sign that CRO doesn't work. The optimization process takes time and effort, and small mistakes can tank your results. There are a few common points that can trip up marketers new to CRO. Here are the seven most common CRO mistakes that you might be making, and CRO tips to avoid any problems as you’re optimizing your site's conversion rate.
1. A Lack of Scientific Methodology
Don't be fooled by the term "science." It's not restricted to labs and people in white coats. You can and should apply the scientific method to your CRO process every step of the way. The scientific method includes four major elements:
Observation: Identify something about your site that could be improved, like low conversions on a product page, or low time on page.
Hypothesis: Come up with an idea to improve it, like making the "Add to Cart" button more obvious.
Experiment: Test your hypothesis through A/B testing, like measuring whether a bigger button affects conversions.
Analysis: Look at the results from your experiment and make sure they're statistically significant. Did the new button improve your conversion rate or not?
Once you've gone through the process once, start over with a new observation. It's an iterative process. For example, you could test three distinct changes: using a bigger button, making the button more brightly colored, and moving the button. Then you could try the changes in combination with each other to see if their effects multiply.
The scientific method is an easy way to keep your CRO process intentional. It allows you to collect valuable data and refine your process over time, which means that you make informed choices and avoid wasting time by repeating tests.
2. Disorganized Testing
Disorganized testing is just as bad as failing to test changes at all. It's not enough to just use the scientific method in CRO. If you don't track your results well, then you may as well make changes at random.
The purpose of the scientific method is to refine your experiments over time. If you test a bigger “Add to Cart” button, you need to record three elements of the process:
The subject of the experiment -- what changes did you make?
The experimental methods -- how did you collect data, and what data did you collect?
The outcome -- how were conversions affected?
Do this for every CRO test you perform. You may not remember that you tested changing the Add to Cart button in a year. If you keep track of every test you perform, you can avoid running the same experiments more than once.
Depending on your preferences, you can track your changes and results in a spreadsheet or a roadmap. Either way, make sure you save these files somewhere easy to access, and make sure your naming conventions are clear. Your future self will appreciate this thoroughness.
3. No Consideration for Timing
Every business experiences periods of high and low demand. A cosmetics company may experience high sales in the weeks leading up to the holidays and lower sales in the summer. At the same time, a grilling subscription brand will get more traffic in summer, but it’ll inevitably fall when people are indoors. Those highs and lows don't just affect revenue. They should also impact your CRO schedule.
The timing of your CRO tests is a critical element of your experimental design. You want to perform tests when your site traffic is average or a little high, not when it's low. This helps you get an appropriate sample size and monitor how changes impact conversions during normal traffic levels.
If your site generally gets 10,000 visitors a month, but only 4,000 a month in the summer, don’t run your tests in July. A week-long trial during that slow period would only allow you to sample a thousand individual visitors, which may not be enough to get statistically significant results.
Instead, run tests when your site traffic is regular or increasing. That same week-long test performed in December could involve 2500 unique visitors, giving you a more robust sample size. You'll have a better idea of how your changes affect conversions because you'll have more data.
See how we launch & scale brands on Shopify Plus with a data-backed approach
4. No Attention to Statistical Significance
After you've collected the results of your test, you need to analyze them. However, it's not enough to eyeball your data and make a decision. You need to collect enough data to ensure that your results are statistically significant.
Statistical significance is a critical element of scientific analysis. A trend in data is considered statistically significant if it's more likely that it occurred because of a specific cause than through random chance. You need to collect enough information that randomness doesn't distort your results.
For example, consider a site that receives 10,000 visitors a month. In a month-long A/B test, 5,000 people see option A and 5,000 see option B. Suppose 4,000 people converted after interacting with site option A, but only 3,500 converted after seeing site option B. In that case, the sample size is big enough that it's statistically significant.
Now imagine that the same test is run, but they gather less data. If they tested their site options for one day and saw 400 visitors total, only 200 would see each option. If 160 people converted after seeing option A, and 140 converted when they saw option B, the difference is 20 people. That difference could be explained by sheer random chance, even though the percentages are the same.
Your number may be inconclusive without you even knowing, but there are plenty of statistical significance calculations you can use. However, if you perform CRO with a marketing platform, it will help you determine if a result is significant automatically.
5. Not Separating Traffic
Your website isn't just viewed through a desktop browser. People visit eCommerce sites through tablets and phones about as often as they visit through desktops and laptops. However, the two different platforms can result in very different web experiences. That's why you need to segment your traffic when you're studying CRO results.
If your website isn't optimized for mobile, you may get great results from desktop visitors and poor ones from mobile users. If your site is optimized for mobile and doesn't cater to desktop users, the opposite could be true. If you fail to separate your traffic, the web traffic from each use case averages out, and it appears that you're getting mediocre time-on-page and conversion rates across the board.
Segmenting your traffic lets you avoid that. You can learn whether a change appealed to one set of visitors over another. You can even spot if you're under-optimized for a particular platform if you see comparatively high bounce rates.
You can also segment your visitors by demographics. That lets you learn about customers from specific locations, people who visit after clicking on ads and emails, and more. It's an excellent tool to break down your audience and learn what appeals to different groups.
6. CRO Testing Instead of Revamping
The last and most harmful mistake is simple: CRO testing can't improve your site if the site itself isn't functional. If your website is old, poorly designed, or contains outdated features like Flash, conversion rate optimization isn't going to work.
Why? Because the problem isn't with your copy or the location of a button. Performing CRO tests on your copy and SEO is like putting a new coat of paint on a car that’s missing an engine. Your site needs to be rebuilt from the ground up.
If your site is out of date, then a complete revamp is more efficient and cost-effective. The time and effort you spend on CRO won't fix a fundamentally broken or slow website.
By investing in a completely new site, you can begin with a clean slate. Your company can focus on personalization, e-commerce features, and SEO during the revamp process, then return to A/B testing and CRO afterward. If you think your website needs a total overhaul instead of optimization, we can help. Let us help you design a new, functional, effective website that's optimized from the jump.
The CRO process takes time and effort. Even if you've implemented CRO tips in the past, there's always more to learn about the process. But don't give up on it!
When you commit to CRO, you can see incredible results. As long as you make sure you follow the scientific method, collect statistically significant data, and stay organized, you're on the right path. If you need help, don't hesitate to contact us. We can help you get your CRO efforts back on track or revamp your website from the ground up. Either way, we'll help you get the results you want from your e-commerce site.