1. Build a Data Baseline
Of all the marketing disciplines, conversion rate optimization is perhaps the most scientific. This is mostly a research job.
And doing your research right means developing a baseline for how your website converts. That takes data. Trying to optimize your site without understanding your data is like building a house on sand. All your work will just wash away in the end.
So where do you get this data? Google Analytics is the best place to start. There are a lot of other options that do things Google Analytics can’t — VWO, Optimizely, HotJar — but for now let’s stick with GA. It’s free, and you probably already have it installed.
So what kind of data are you looking for exactly? In most cases it will be conversion rate data — you know, because that’s what you’re trying to optimize.
Look at the raw conversion rate for your most popular pages that directly lead to conversions. Typically that means your homepage and product pages.
Your blog might have a lot of traffic too, but that’s rarely a direct route to revenue. The best opportunities are those that deliver a significant lift in revenue, so start with the pages closest to the money.
Record your traffic numbers, bounce rates, conversion rates, and revenue numbers. Voila, now you have a baseline. Whenever you conduct a test, you’ll always attempt to beat the baseline. That’s how you know you have a winner.
Look for Quick Wins
Now that you have a handle on your conversion rate at the page level, let’s do some segmentation.
A little later, we’ll cover how to formulate and test a hypothesis to improve conversions, but in the beginning simply slicing and dicing your data can reveal opportunities hidden in plain site. These issues are often technical in nature, so just fixing them will raise your conversions.
The out-of-the-box reports in Google Analytics work nicely for this type of analysis. Here are two to start with:
Alt: Browser report in Google Analytics
It may surprise you to learn that not all web browsers work the same. Alright maybe not, but it is surprising how often cross-browser compatibility is overlooked. If your site is broken in Firefox or Chrome, then your conversion rate is probably tanking.
Go to Audience > Technology > Browser & OS.
Now check out your conversion rates for each browser. If one is really low, then it’s worth your time to run some tests to see how your site is rendering in that browser.
Page Load Speed
Alt: Page Load Speed report in Google Analytics.
Page load speed is another technical opportunity that can instantly increase your conversions once fixed.
The reasoning is simple: If your site is slow, people leave. That’s an absolute truth about the internet. So you really want to make sure your pages are loading quickly — or at least quickly enough to where it isn’t costing you conversions.
There are a couple of different ways to look at this, which Yehoshua Coren articulates here. However you choose to look at it, don’t sleep on page load speed. If you ignore it, you’re just refusing to earn more money.
II. Know Your Funnel. Optimise Your Funnel.
Alright, so you’ve got baselines for your data, and you’ve tackled any easy wins. Now it’s time to analyze how your marketing funnel works, isolate priority areas, and start to form some hypotheses for improvement.
This requires that you gain a clear understanding of your marketing funnel, i.e., the pages people visit and the steps they take on their way to converting. This requires that you set up a funnel report in Google Analytics.
If you don’t have that set up, you should stop what you’re doing and make it happen. Here are a couple of resources that explain how:
How to Set Up Analytics and Measure the Right Stuff
How to Set Up Goal Funnel Visualization Reports in Google Analytics
Now, assuming you have your goal reports set up, here’s how to use them for your conversion audit.
Find Underperformers and Drop Offs in a Funnel Report
Setting up your funnel reports properly delivers the raw data you need to figure out where people are falling out of the conversion process.
For ecommerce stores, conversion rate suffers during the checkout process. Customers put items into their carts — and they may even enter their credit card information — but they don’t complete the purchase.
For B2B brands, this could be someone clicking through from the homepage to a product page and requesting a demo.
Either way, having a funnel report will show you exactly which point in the process people are getting cold feet. That allows you to pinpoint the exact page where you should focus your optimization efforts.
Here’s a review of slicing and dicing data in funnel reports:
Alt: 3 ways to view funnel reports in Google Analytics
Remember, you’ll probably find multiple opportunities for optimization. Focus on the areas at the bottom of the funnel first. You’ll not only improve your revenue in the short term, but you’ll also ensure the improvements you make further up the funnel compound the good work you’ve already done.
Figure out why with qualitative data
Up to this point, we’ve only looked at quantitative data, the hard numbers. This information is critical for uncovering what needs improvement. But it’s not that good at telling you why that particular page needs to improve.
That’s where qualitative data comes in.
There’s a lot of different ways to obtain qualitative data. Here’s a short list:
- Heatmap tools
- Pop-up questionnaires on your site
- Customer surveys
- Customer interviews
- User experience research
- Interviewing your customer success and sales team
Now, some of these options are more reasonable than others. For example, interviewing customers requires a lot of time investment. Meanwhile, implementing a heatmap tool like Hotjar continuously collects data at a relatively small cost.
Since qualitative data can be subject, it’s best to combine multiple sources of qualitative data.
So getting heatmap data will allow you to get more or less firsthand knowledge at how people are using your site. If they’re on your SaaS product page, hover around the request demo form, but never fill it out, you can assume they don’t feel comfortable entering their data.
But that’s still an assumption. Now, say you launch a one-question pop-up survey on that page that asks “What do you want to know that’s not on this page?” Now you’re getting more specific answers about what’s tripping people up in that moment.
Appcues takes a slightly different approach by providing options for the user to choose (instead of going fully open-ended) but it’s still a worthwhile example:
Alt: Pop-up question on Appcues
Finally, interviewing your colleagues in sales and customer success is always a good idea. They’re in direct conversation with customers all day, every day. So these team members should have insight like
- What questions customers ask most often about the product
- What outcomes they’re looking for from using the product
- What hesitations they have about committing
The combination of these feedback loops should give you the raw material you need to formulate some experiments to run.
III. Hypothesis, Test, and Validate
At this point, you’re done with the conversion rate audit. Now it’s time to run some tests.
There are a lot of different ways to run experiments on your website. You could change the copy to better describe your product or implement some design changes to make your ecomm store more compelling.
Let the qualitative data be your guide. Look for patterns in the responses you got from your surveys and your interviews with sales or customer success. It’s critical that you have a solid hypothesis that you can prove or disprove with your tests.
Also, most tests won’t be significant. I know that’s difficult to accept, but it’s the truth. You’re not going to be finding gold every time you implement a test. But an insignificant result isn’t a failure, because you’re learning. And the more you learn, the closer you get to launching one of those tests that really will rake in the revenue.
There are other things to consider too. You must run tests until you have statistical significance. Tools like Optimizely and VWO will automatically calculate this for you, but that doesn’t always mean you should stop testing. It depends on your sample size.
Statistical significance is a somewhat complicated topic, so I recommend reading this post about statistical significance and validity.
Clearly, there’s a lot that goes into conversion rate optimization. From developing a data baseline to conducting qualitative research to find solutions, the process isn’t easy.
But as digital marketers, we’ve been granted an astounding opportunity to continuously improve how our websites perform and our businesses run. Our predecessor would have killed for that kind of resource.
So even if it takes you awhile — and it probably will — start building your conversion rate optimization process. It’s the best chance you have at making sure your website keeps up and eventually surpasses your competition.
Otherwise, you could be easily left in the dust by more data-savvy marketers. And no one wants that.