Analytics Market Chart

Increase Website Conversion with Google Analytics

Google Analytics can tell you what's wrong with your site.

Google Analytics is a powerful tool for improving online conversions. It helps to answer two fundamental questions:

  1. How is my website performing?
  2. How is my marketing performing?

Understanding what is actually happening and why is critical. This article focuses on getting a comprehensive view of how your website is performing and what to change to increase conversions once visitors get to your site.

How is the Website Performing?

Regardless of the type of site, the steps below outline the pattern every site owner must use to improve their efforts.

Step 1: Define Success

Every site is different and exists for a different reason. For example, the goal for an e-commerce site is getting the visitor to finally lay down some money. Within an organization, there are also teams with separate objectives. For the acquisition team in an organization, the goal might be to get the visitor past the beautiful landing page.

Beyond a primary objective, every site has multiple measures of success. Before making changes, we have to know what we want those changes to affect so that we can measure whether it was worth the effort to change things. We also need to know whether we even changed things in the right way!

To do this, we need to clearly identify what the goals are. What determines whether this visit was a success or whether it was a drain on resources?

Possible success points:

  • purchase
  • make a comment
  • add an item to the shopping cart
  • view a product page
  • sign up for an email list
  • read an article
  • view the contact page
  • not view the contact page

These actions are not to be confused with the 20 conversion goals in GA. Only some of these should be set up as goals like that. They should all be watched, but it's not always appropriate to track them that way.

Also, think long and hard about why visitors are coming to your site in the first place. What has to happen for them to feel like the visit was worthwhile? Browsing through keywords and Site Search reports can help with this. (Also pay attention to what kinds of searches aren't there that you assumed would be.) Surveys give a sure-fire indication. You may decide after reviewing this list that you have designed your site all wrong, because you're attracting visitors for the wrong reasons. Or you may decide that you've been too focused on certain conversion points while overlooking important — and potentially profitable — site activities.

Step 2: Find Trouble Areas

Now is the time to look at each of those activities a visitor should do and determine whether they actually are doing them. If not, why not? What is holding them back?

Since every site's goals are different, finding the trouble areas will be a unique process. There are some common places to start looking, though.

Measure each goal individually to evaluate whether it is being achieved. Determine what can be measured that would indicate success. This might require going back into the site to add a snippet of code to track a click or capture time on page.

Look at landing page reports to find pages with high bounce rates. If the page was supposed to draw the visitor deeper into the site, or if the visitor's intention was clearly to do something that took longer than one page, this is a trouble area. When looking at landing pages, though, remember that not all of the blame for a landing page's performance can be attributed to the page itself. Did the advertisement or traffic source give the wrong impression? (Here is one example of how it can be difficult to disentangle the site's performance from the marketing's performance.)

Look for pages with high exit rates that visitors shouldn't be leaving from. Are there pages in the checkout process or some other natural funnel that visitors abandon?

Also, look for pages where visitors should clearly be progressing to a particular page. Where are they going instead? That can tell you a lot about what they expected or why they left the expected flow.

You need to find out not just whether or not a visitor is doing what they meant, but you also need to get some possible explanations. If hordes of visitors are going to your contact page from the middle of your checkout process, is it because something is broken on your site? Is the process confusing?

Create an advanced segment for unfinished goals to see what brought those visitors to you in the first place and what other pages they viewed. This will give you insight into why they are there and what they expect.

Step 3: Test

After cataloging all the things that are wrong with your site, you need to start making some changes. But, not so fast! Unless something is broken, it's not usually wise to change everything all at once, nor is it always wise to change it immediately. We should make sure whatever change we make is actually a change for the better, and also that you are changing the right things.

Test areas of the site that have the largest impact first. If your home page seems to be affecting conversions negatively, you might start there if it has the most traffic. If another page on your site seems to have a more adverse affect on conversions you might start there instead, even if it receives a little less traffic.

From there, using the possible explanations you gathered in the last step, make some hypotheses about why that page is hurting you. Is the color scheme is all wrong? Is the call to action unclear? Is the article you wrote just boring? Is the page getting traffic for unrelated search terms? Is it some combination of all of those?

Usually, it's not clear exactly what element of the page is causing the problems. In this case, run a multivariate test in Google Website Optimizer on each element of the page that you have doubts about. The main principle with all testing is to work from big issues to small. Pick as few elements of the page as you can and then just run one or two variations of each element. Test broad principles first by testing very different variations. (If the variations are too similar, the experiment may take longer to complete and you may not be able to tell what about the winning variation made the difference.)

If the call to action button isn't getting enough clicks, start by testing a very large version against a very small one. If there is no conversion difference between the two sizes, you know that you chose the wrong variable. After that, try very different wording. Then different colors. Then different shapes. Then different locations.

Once you've figured out which variable (or set of variables) is responsible, test different variations of each. If you learned that the shape of the call to action button had an impact, what is the very best shape to use? In the end, you should be able to pinpoint exactly what made an impact, so that you can apply the lesson later in similar situations.

The genius of running a multivariate test is that you can see whether it was just one element of the page holding back conversions, or whether it was a combination.

This process requires data, which takes time to gather. That's one reason for starting with very simple tests. They take less time to give clear results.

Step 4: Optimize

Once a series of tests has been completed, take action. Change the elements of the page to reflect what you learned from your experiments. If there are similar situations elsewhere on the site, you are probably safe to make the same kinds of changes there, too.

If you are doing exhaustive testing and optimization, you will find that you end up retesting elements of the site. As you optimize the site, smaller trouble areas will start to surface that never caught your attention before. After a round of optimization, you may find that the variation of a page is no longer as effective as it seemed before. This may be because it doesn't match the rest of the site as well anymore or because the traffic itself has changed or just because the rest of the site is so much more polished that your first improvements prove to be less effective in comparison.

In any case, this kind of data-driven change is good!