Experimentation / CRO

Experimentation and Testing Programs acknowledge that the future is uncertain. These programs focus on getting better data to product and marketing teams to make better decisions.

Research & Strategy

We believe that research is an integral part of experimentation. Our research projects aim to identify optimization opportunities by uncovering what really matters to your website users and customers.

Data and Analytics

90% of the analytics setups we’ve seen are critically flawed. Our data analytics audit services give you the confidence to make better decisions with data you can trust.

How to find optimization opportunities in analytics data

When it comes to experimentation, nothing’s more important than data. Experience and intuition can only take you so far, so here’s our guide to using Analytics to find optimization opportunities for your website.

Web Analytics analysis

One of the vital pillars of our ResearchXL methodology is web Analytics analysis, which allows us to uncover hidden problem areas or areas of interest for future optimization. 

By following this process you’ll be confident you’re getting the most out of your analytics tools and you’ll uncover new areas for growth opportunities. Read on for 8 detailed guides on what optimization opportunities to look for, when conducting analytics analysis.

1) Know your high impact pages

High volume pages are the most visited pages on the site, while high-value pages are the ones that are contributing the most towards conversions or site revenue. 

  1. High-volume, low-value pages: e.g. an old blog post that brings in a lot of organic search traffic
  1. Low-volume, high-value pages: e.g. checkout pages

You’ll want to optimize both types of pages, but if you have to prioritize one, start with high-value and get those quick wins in the bag. High-value pages are often the ones closer to the ‘converting action’ or down the funnel and provide a huge opportunity for conversions. Running experiments or doing even smaller UX improvements on high-value pages can have the biggest impact on revenue or CR increase. By investing a little down the funnel, you can gain a lot in ROI. (read more about getting quick wins)

Note: You can evaluate the impact of the page based on the actual Google Page Value or Conversion metrics. Page value can be useful if set up properly, and the interpretations made based on that set up are understood correctly. You should talk to your analytics team to understand how the account has been set up and how they have defined this page “value” to decide which approach to take. 
Looking at multiple metrics is recommended, to make a more informed decision on how to prioritise future optimization efforts.

2) Identify potential usability or technical issues 

Nothing will send a user back to Google and into the arms of a competitor like a broken link, bugs or loading issues. The sooner you eliminate these obstacles, the sooner you can start optimizing your actual content. 

Basic usability checks include:

  • Analysing the performance for various browsers, browser versions and operating systems 
  • Reviewing different screen resolutions.
  • Looking at the device split between mobile, desktop or tablet. Also comparing the performance for various mobile devices.
  • Going through pages with high exit rates

This will help you identify if there are any underperforming cohorts which may have bugs or UX issues. Often times, you can find that there are bugs on a specific device type or some of the resolutions are not showing all of the website content/elements that all visitors should see. Review the top devices, browsers or screen resolutions that have high volume, but abnormal data on engagement metrics (e.g. bounce rate, pages/visit.) or low conversion rates. The browser report below shows user traffic by browser (currently segmented for desktop audience).

Firefox is the second most popular browser, however, it has lowest conversion rate among the top 4 browsers.  This suggests UX or technical issues that can affect Firefox users more than others. You can also narrow down your search to review how different versions of the browsers are performing. It’s always better to dig deeper, to find the specific issue faster.

As an example, the screenshot below shows us that the bounce rate for visitors using a 2018 version of Edge is significantly higher than the site average, which would warrant further investigation. 

The data itself won’t tell you what the actual bug or issue is, but it will help you pinpoint the specific browser or version for further investigation.

3) Fix the damn broken links

There’s really no good excuse for broken links, especially when they can be fixed easily. Not only are they ensuring that the users on your website will have a bad experience, but can also have significant affects on your SEO ranking (Google doesn’t like sites with 404 errors, as indicate poor “quality” content). 

Reduce frustration for website users and help your ranking by heading to the Behavior report in Analytics. Search “Page Not Found” in page titles and see which links are giving your users trouble. Some URLs might need an edit or redirect, while others might indicate a more serious issue. 

You can also set up new events to track errors on your site (e.g., “product out of stock,” “invalid login,” “invalid coupon”) to find further opportunities for optimization. A good example is adding error tracking on a product page, where a visitor need to select sizes before proceeding, will let you identify if the UI for these selectors is intuitive and stands out or if it needs further design improvements. 

4) Compare landing page metrics  

Identify who is coming to your site and where they are landing. Comparing conversion or engagement metrics for different landing pages can help us understand how they are performing. If visitors coming through specific ads are not finding the content or products on the landing page that they were originally looking for, this opens up an opportunity for future improvements. 

Top landing pages that indicate issues in engagement metrics (e.g. high bounce or exit rate) or a much lower conversion rate than other similar ones should be investigated further. In the example below, you can see how people landing on a specific page converted much lower than anyone else. (read more about building high converting landing pages)

Doing further analysis into acquisition channels that landed on this page or looking at specific keywords from Google ads allowed us to identify later on that people searching for specific content were not seeing enough relevant products on that landing page.  Digging into this further, we found that people were searching for specific terms from google search and they did not have enough examples of the products they were after, on this specific landing page. 

Looking into traffic channels or specific keywords that are being bidded on, you can find which ones are not converting as well and reevaluate if you are spending your money wisely or if you need to reevaluate your strategy. 

While landing pages and acquisition will directly help you identify pages or channels which are not performing well, you should also have a look at exit pages. This will help you get a better understanding of pages that users saw last, before they decided to abandon their journey or leave the website. Seeing a high exit rate on a specific page is not always a bad thing, for instance if the user already finished their purcha. However there might be opportunities for optimization for pages that users should generally not be leaving. 

Tip: You should also have a look at the navigation summary reports, which show the pages most frequently viewed before and after a particular page.
This can be useful for identifying when users are going around in circles or moving through the site in a way we wouldn’t expect. 

5) Ever tested your site speed?

Slow sites kill conversions. In general, load times should be three seconds or less (read more on page load time statistics). Any page load times that are 10 seconds or more could cost you traffic, conversions and money.
You should look at which pages or segments have slowest load times, to prioritize future improvements. Often times, slow page load time is not an issue on the overall site, but a few specific pages (or even device/browser) that affects the average scores. 

The explorer tab, underneath the Page Timings report in GA, allows you to access information on user engagement and load times for individual pages or segments (e.g. traffic sources, user types, devices/browsers.).
This view includes the following subtabs:

  • Site usage - Basic interaction metrics 
  • e.g. Pageviews, average load time and bounce rate
  • Technical - Network and server metrics
  • e.g. Avg Domain lookup time, Avg server connection time, Avg page download time
  • DOM timings - Document parsing metrics
  • e.g. Avg document interactive time, Avg Document content loaded time. Read more about site speed metrics and what they mean here

In the image below, you can find an example of page load times per page (taken from the Site usage subtab).


In this case, the affiliate page had most issues with page load time and should be further investigated. The example below also highlights the importance of using segments in your analysis.

When comparing page load times for various devices, we found that the issue was mainly evident on mobile devices, which was also impacting the average scores. 

6) Analyse the sales funnel

If you’re not clear on your sales funnel, you can’t get clarity on your customer journey. Conversion reports in Google Analytics will help you see where your funnel is leaking (and how quickly). You can then prioritize actions.

Your report might look a little bit like this:

Or something like this:

So what do we do with this data? We should watch out for the highest abandonment rates between the steps to identify where the biggest optimization opportunities lie. If the biggest drop off is on the product page, then this page should be prioritised for testing or further (qualitative) research. Increasing add to carts on the product page should be one of the primary goals to work towards, to ensure more people are going further down the funnel.

Top tip: 

View the Reverse Goal Path report. This shows you the last three pages the person visited before completing a goal. This can give you more information about your most valuable paths.

7) Review the ‘goldmine’ that is internal search usage

The internal search feature is a goldmine for conversion optimization. This will tell you what users are looking for, how often they’re searching for it and whether they’re finding it. Consider it a gift of content possibilities.

The Site Search report will give you data that will help you better meet your users’ search needs. Below we’ve listed a few common questions you’ll want to ask, and where in Site Search to look for them:

  1. How often do visitors use my website search bar and what are they looking for? Look at Usage stats.
  2. Where do people start searching? Look at Pages stats.
  3. Are website visitors satisfied with the results of the search? Look at the % Search Exit value. If this number is high, it’s likely that visitors are unsatisfied. If the Results Pageviews / Search value is over 1, this suggests people had to dig to find what they were really looking for. (Note: This might not be a bad thing, it all depends on the context.)
  4. How do different cohorts search the site? Use advanced segments

How is my business affected by visitors searching the site? Under Search Terms and Site Search Category (Primary Dimension), select a category, then select the Goal or the Ecommerce tab.

Here’s an example of an internal search term report:

You can also link site search queries to overall conversion rates. Are visitors searching for specific terms more or less likely to convert? 

8) Segment your data to make it more meaningful 

You’ve probably already found a few areas of concern on your site that require fixes or further investigation. With segmentation, you’ll probably extend that list, but you’ll get a much clearer understanding of how your users behave.

Advanced segmentation lets you analyze your users closely by narrowing down data into meaningful groups or cohorts. You could segment by:

  • Traffic sources e.g. Social, Direct etc.
  • Mobile vs Desktop
  • Geographical location
  • New vs Returning visitors
  • Landing/entrance page
  • User behaviour (e.g. did the user use site search, click on live chat or visit a specific page during their journey?) 

You’ll often find your conversion rate is lower when looking at segments. This is mainly because most of your users have already bought from you. Your real conversion rate might be higher if you exclude existing customers from the segment. Remember, averages don’t tell the whole story. In order to extract the maximum value from your analytics analysis, dissect your data to get a specific view, then take a deep-dive into segments and secondary dimensions. 

The example below shows a huge discrepancy between different devices and how they behave on the site.

While almost 80% of the site visitors came to browse on Mobile, majority of the transactions or checkout page visits were done by people on desktop. What does this mean?

Perhaps Mobile visitors are more likely to do research and evaluate the different products, while desktop visitors were more transaction focused. Maybe this means that the same mobile visitors are later returning to the site on desktop to complete their purchase. Alternatively, this can also indicate that the mobile version of the website has huge usability issues. To dive deeper, more qualitative research should be conducted (such as user intent polls, user testing and etc.) but this allows us to see how different cohorts behave on the site and where we should focus additional research efforts. (see conversion research techniques here)

Summary And Key Takeaways

What’s good for one site may not be a marker of success for another. Therefore, it’s important to think about the data you gather in the context of your business, and keep an eye out for data that’s unusual for your site. Seasonality or various campaigns on the site can also have an impact on how users behave on your site, which should be taken into account when conducting analysis.

Smart questions at the start of this process are essential. They’ll help you uncover problem areas faster, so you can develop more meaningful experimentation hypotheses. To summarize, you should: 

  1. Carry out a simple walkthrough of your site. Keep an eye out for “bad” parts of the site and signs of underperforming cohorts. (See: UX Heuristic review, to detect usability issues)
  2. Come up with a list of questions that will result in answers that you can easily act on, for any additional (perhaps more qualitative) research.
  3. Context is essential. You’re looking for irregularities for your site, not anyone else's.
  4. Gain a good understanding of Google Analytics basic reports and how to extract as much valuable data as possible from them.
  5. Use segmentation to dive deeper into your data and get more useful insights.

Related Posts

Who's currently reading The Experimental Revolution?