For an web analytics analyst or a data-driven marketer, these are words to live by: “Without data, you’re just another person with an opinion.” [Tweet It!]
Optimization isn’t about educated guesses and hunches, no matter how many years you’ve been in the industry. It’s about doing the research, asking the right questions, digging for clues in problem areas, paying attention to the signs when they appear, and running smart A/B tests.
Web analytics analysis is a big part of that. It helps to separate the optimizers from just another person with an opinion.
Why Web Analytics Analysis Is Important
Web analytics analysis is a key part of the ResearchXL model. In case you’re unfamiliar with our conversion research framework, here’s a visual summary…
Typically, when explaining web analytics analysis, I say it’s when you: “Conduct an analytics health check. Set up measurements for your KPIs and identify leaks in your funnel.”
In practice, it’s a little more complex than that. I’m willing to bet quite a few people skip this step altogether. And those who don’t? They’re likely missing out on a lot of hidden opportunities.
7 Ways that Predictive Analytics Is Transforming Ecommerce
Predictive analytics help you understand what your customers are going to buy before they do.
- Ensure your data is reliable and accurate.
- Answer questions that arose during heuristic analysis.
- Uncover new, hidden problem areas / areas of interest, leading to the generation of more questions.
- Find quick wins / fixes and future A/B test ideas.
Questions That Need Answers
Before you can properly complete your web analytics analysis, you should have a list of questions that need answers. For example…
- Is anyone actually using the text size or print buttons?
- When they’re browsing, are they using search?
- What do our most valuable users do differently on our site? Does that matter?
- Are there broken devices or browsers?
- Where are the leaks in the funnel? How can we fix them?
- Which promotions are most effective?
- When they do X, is an event being fired?
- Are error messages being recorded?
- How often are items removed from the cart?
That’s why asking intelligent business questions is such an important skill. (Sorry, but stupid questions do exist.) As Peep explains, asking intelligent questions means the answer you uncover will be actionable…
While working with Peep on a conversion research project, for example, I wrote down multiple questions during a walkthrough. At the end, he asked, “What are you going to change / do based on the answer to this question?” If I couldn’t answer, the question was deleted.
You have a ton of data at your fingertips, so if you don’t know what you’re looking for, you won’t find it. And when you do find it, you need to know what you’re going to do with it. There’s a big difference between data and useful data.
So, before you start web analytics analysis, do the following…
- A simple site walkthrough.
- Look for suspicious / “bad” parts of the site.
- Look for signs of poor tracking (e.g. each flow can’t be measured separately because the URLs are the same).
- Develop a list of questions you need answers to.
- Go through the list of questions and ask yourself, “What are you going to change / do based on the answer to this question?” If you don’t have an answer, remove it.
Chris Mercer of MeasurementMarketing.io adds that asking intelligent questions isn’t quite enough…
Conduct a Google Analytics Health Check
I’ve written an entire article on how to conduct a Google Analytics health check, so I’ll keep this brief. The goal of the health check is to help you answer these three questions…
- Am I collecting all of the data I need?
- Can I trust the data I’m collecting?
- Is anything broken or tracking / reporting incorrectly? Why?
You might look into…
- Is enhanced link attribution turned on?
- Do you have your “Virgin View” and “Working Views” configured properly?
- Have you configured your custom and default channel groupings?
- Is internal site search set up and working correctly?
- Does the site use a third-party cart? If so, do they have cross-domain tracking in place?
Essentially, you want to have a look around and ensure you’re collecting all of the data you can be in a reliable way that ensures accuracy.
Of course, you want to cover your bases, but as you grow and over time, your web analytics setup risks becoming overly complicated and, therefore, will continually require debugging. This isn’t a one time job. As you continue to grow your business and your web analytics maturity, you’ll want to continue to watch our for issues of data integrity, data maturity, and data governance. You’ll begin asking questions like:
- Should we set up filters and views for this particular team?
- Who should have access and control to set up new event tracking?
- What’s our event tracking taxonomy? What’s our UTM campaign tagging taxonomy? Can we clean it up?
- Are there strategic events that we’re currently not tracking? What’s a proper road map to improved tracking?
- How can we streamline certain reporting for stakeholders and executives? How can we improve the process of business questions and answers?
Remember That Context Is King
Is bounce rate useless as an indicator? Of course not. Should you exclusively use page value to indicate success? Also no. What’s important here is that you’re hyperaware of context at all times. Use this same critical mindset when viewing any sort of metric, including click-through-rate and even conversion rate. There are no absolutes in this case, and an analyst’s best strength is a curious, humble, and inquisitive mind.
Remember, you’re looking for anything that’s unusual for your site. Just as there is no “good conversion rate”, there’s no “good bounce rate” or “good page value”.
You’re like a detective digging for clues when you’re conducting web analytics analysis. If you don’t know the history of the case / crime / suspects, you’ll probably put the wrong person behind bars.
Step One: Start with the Highest Value
There are two high value places you can start looking into right away:
- High volume, low value pages. For example, an old blog post that gets a lot of organic search engine traffic because it ranks well for a popular keyword.
- Low volume, high value pages. For example, your checkout page.
Optimizing either of those pages would likely result in big value for your business, right?
Jeff personally likes to start with those high volume, low value pages…
So, for example, here’s a list of top pages (by pageviews) and their bounce rates (by comparison to the average)…
If I were going to go after high volume, low value pages, I’d start with pages 3, 7, 8 and 9. You can change bounce rate to whatever success indicator you’d like, of course.
You might’ve noticed that I’m using an advanced filter. Why? To filter out pages with too small of a sample size. If not enough people have seen the page, the success indicator isn’t reliable.
Pages 3, 7, 8 and 9 should be added to a list of suspicious / “bad” parts of the site. From there, we can examine the pages in-depth.
Is something technically wrong? Is the copy boring or unclear? Is the UX awful? Take a look at similar pages that are doing well. What can the bad pages learn from the good ones?
Tip: The navigation summary is an awesome, lesser-known resource. You can use it to find out where people are going when they leave key pages, which could give you an indication of what’s missing from those pages. You can also identify links that are rarely used and simply distracting from your main goal, your most wanted action.
Here’s an example from a blog…
9.7% of people will leave the blog (page 1) to visit page 2. Another 8.5% will leave to return to the main site. Obviously, this is 10x more helpful for SaaS and eCommerce sites.
Step Two: Web Analytics Quality Assurance
I’ve said it before and I’ll say it again, website quality assurance is vital. Pages and flows that are technically broken are a conversion killer… there’s no way around it.
As Peep explains, cross-browser and cross-device issues are low-hanging fruit for optimizers…
You did a simple walkthrough above, but you should also do an in-depth technical walkthrough where you explore your site (starting, first, with your funnels) using all of the devices and browsers that are even remotely relevant today.
Here are just a few different devices…
- iPhone 5S
- iPhone 6S
- iPhone 6 Plus
- Samsung Galaxy S4
- HTC One X
- Blackberry Z10
- Google Nexus 9
…and a few different browsers…
- Chrome 51
- Chrome 47
- Chrome 21
- Opera 38
- Opera 12.12
- Firefox 40
…but there are literally dozens and dozens more (of both). You can use a tool like BrowserStack to help you with browser and device testing, but it’s still rather time-consuming.
So, while you’re working through all of them, you can choose to focus on the handful that will provide you with the most value.
Chris explains how to identify the biggest opportunities here…
Here’s an example of the browser report…
Approximately 37.8% of Chrome sessions result in a transaction. Safari (33%) and Firefox (39%) are about the same. If they were drastically different, you might have uncovered an issue.
You can also drill down and see which browser versions are most important, which can help you spot issues below the surface…
Take a look at number 7. The bounce rate is through the roof compared to the average. That definitely calls for a closer look, so you’d want to add it to your list.
Don’t leave this quality assurance to your tech team. They will miss things. Check and re-check. Go through every browser, every browser version, every device. In the meantime, you can spot high value problems based on your current audience.
Step Three: Broken Links
A broken link is friction. A visitor linked a link because they wanted X, but instead they got a 404. Disappointing, right?
That’s a big part of the reason you should be looking for, and fixing, your broken links. Sounds pretty simple, but let me ask you this…
- How many people have landed on 404 in the last 30 days?
- Are your 404 errors more often the result of an old list or a link that’s never existed?
Many people won’t be able to answer. That’s because while it’s a well-known fact that broken links are bad, most people don’t know how to / want to do much about it.
Chris suggests one way you can get started…
Before you begin, just ensure that the Google Analytics code is, in fact, on your 404 page. If so, when you search for “Page Not Found” (or whatever the title of your 404 page might be), you’ll see something like this…
When you click through, you’ll see what is, essentially, a prioritized list of 404s to fix…
If you want more info, you can even use a secondary dimension. Open it up and search for “Full Referrer”. This will tell you the URLs that referred traffic. This can be a huge help, especially if you have external sites triggering your 404s.
As Ian Lurie of Portent points out, broken links are also bad for SEO…
Note that form error messages are important, too. We’ve written an entire article on perfecting those and tracking them in Google Analytics, so I suggest taking the time to read it: 4 Common Mistakes With Error Messages (and How to Fix Them).
You can also use Google Analytics events to track general site errors (e.g. “product out of stock”, “invalid login”, “invalid coupon”).
Step Four: Site Speed
While 1-second load time would be nice, if you manage to get a load time under 3 seconds, you’re doing fine. If it’s under 7 seconds, it’s okay too (but you have to try to improve it). Over 10 seconds and you’re losing money in noticeable quantities.
Basically, to avoid losing money in noticeable quantities, you want to keep an eye on your page load time numbers.
Here’s a super basic example where I simply compare the average page load time of my top pages to the site average…
Obviously, there’s some work that can be done here. Again, make note of these pages so that you can dig for more clues later.
But you can get way more insight with a little segmentation. Do certain devices have slower load times? Certain countries? Browsers?
Step Five: The Funnel
You can use the conversion reports in Google Analytics to identify which parts of your funnel are leaking money (and how rapidly). This will tell you where to focus first and can point you in the right direction for further web analytics analysis.
The funnel visualization report looks a little something like this…
Here, it’s clear that the focus should be on the “add to cart” step. In your funnel, you might need to focus on the payment page or the review order page… it’ll be different for everyone.
Remember, the further down the funnel you get, the less of an impact you need to make to get a big return.
For example, if you’re an average eCommerce site and you improve the checkout page even slightly, you will see a big increase in revenue. If you optimize closer to the top of the funnel, a small improvement won’t have as big of an impact.
There are, however, quite a few limitations with the typical goal funnel visualization in Google Analytics (including backfilling, and the lack of segmentation capabilities). Luckily, most of this limitations can be surpassed (if you’ve implemented enhanced ecommerce) by using the horizontal funnel feature.
These are entirely customizable (to the point that you can use them for things like content engagement tracking), and you can analyze by any custom segment you’d like as well as by dimensions.
Another cool feature of this funnel feature is that, if your GA account is integrated with Google AdWords or Doubleclick, you can create an audience off of anyone who abandons at a particular step and use them in your AdWords campaigns. This enables more advanced targeting for your ad campaigns.
You also have your goal flow report, which should look something like this…
So, what’s the difference? Well, there are actually a few differences…
- The goal flow report is more flexible.
- You can apply advanced segments, you can see retroactive data if you add / change a funnel, you can take full advantage of date comparison, etc.
- The goal flow report shows the most accurate path your visitors take before completing a goal.
- It allows loopbacks, which means that if someone goes from step one to step two and then back to step one, you’ll see it (no exit recorded). It also doesn’t backfill steps if the visitor skips a step in the funnel. Finally, it shows you the actual order in which steps of the funnel were viewed.
If you’re interested in learning more about these and the many other differences, click here.
The goal flow report will be able to tell you more about the paths people are taking, where there’s friction and where to prioritize your digging / optimization.
Tip: Always check the reverse goal path report, which shows you the last three pages the person visited before completing the goal. This can essentially be used to tell you about your most valuable paths, some of which you may not even have considered.
Is there any unusual behavior here? Why might that be?
Step Six: Internal Search
If you allow visitors to search your site and you’re not following up on how often they’re using it, what they’re searching for, whether they’re finding it, etc., then you’re leaving money on the table. All over the table.
How frequently do users use my search box and what are they looking for?
Where do people begin searches and what do they find?
Are users satisfied with what they find?
How do different groups of users search my site?
What business outcomes result from users searching my site?
The answers to these questions will open a lot of doors for you to explore beyond. Fortunately, it’s pretty easy to navigate all of this in the site search report…
- Frequency: Site Search > Usage
- Begin: Site Search > Pages
- Satisfaction: Look at the % Search Exit metric. If it’s high, people are leaving immediately after searching, meaning they likely weren’t very satisfied. Also, look at the Results Pageviews / Search. If this is over 1, people had to dig to find what they were really looking for. (Note: This isn’t always a bad thing; remember the importance of context.)
- Different Groups: Use advanced segments.
- Outcomes: Site Search > Search Terms > Site Search Category (Primary Dimension) > Select the category > Select the Goal tab or the Ecommerce tab
Here’s an example of the search term report…
That report is pretty basic, but with a little digging, site search tells you what people are looking for and whether your site is effective in helping them find that. That’s a lot of insight. When you notice something unusual, be sure to make note of it for later.
Samantha Barnes of LunaMetrics gives just a few real world examples of how site search can uncover insights…
Web Analytics Segmentation Is the Real Secret
So, we’ve covered a lot. By now, you probably have a pretty big list of suspicious / “bad” parts of your site to explore further as well as a big list of issues that just plain need to be fixed.
But the truth is that your list will 10x when you really focus on segmentation, which allows you to slice and dice your data to find even more insights.
Jeff gives a great overview of using advanced segments…
I’ve actually written an entire article on advanced segments, How to Setup Google Analytics and Segment Your Data. As mentioned above, Yehoshua talks about segmentation heavily in his section of our Beginner’s Guide to Conversion Rate Optimization as well.
I definitely recommend taking the time to read those two resources before you continue with web analytics analysis.
As Chris Mercer points out, web analytics analysis goes hand-in-hand with the other steps of the ResearchXL model, like qualitative research…
Use Custom Reports
Also, don’t forget that you can put together custom reports to help you with web analytics analysis. So, don’t get stuck thinking you’re limited to the handful of reports that Google Analytics supplies by default.
You can learn about a few key custom reports for optimizers (and how to make your own) by reading 12 Google Analytics Custom Reports to Help You Grow Faster.
Questions lead to more questions, but if you’re asking intelligent ones, eventually they’ll lead you to insights on problem areas and smarter A/B test ideas.
This all gets easier the more often you do it. Why? Because the data itself won’t tell you much of anything… that’s why it’s called web analytics analysis. Spend a few days roaming around Google Analytics, experimenting.
For now, stick to the following process…
Do a simple walkthrough of your site. Look for suspicious / “bad” parts of the site and signs of poor tracking setup.
Create a list of questions that will result in actionable answers.
Conduct a Google Analytics health check to ensure: you’re collecting all of the data you need, you can trust the data you’re collecting, and nothing is broken or reporting incorrectly.
Remember that context is king. You’re a detective digging for clues; you’re looking for irregularities for your site.
Understand the basic Google Analytics reports and how to get the most insight from them.
Understand segmentation and how to dive deeper into your data for even more insights.
Develop your own custom reports to help you.