Many web optimization projects are NOT based on data, but instead on hunches and guesses. Digital Analytics is all about taking the guesswork out of optimization. Separating facts from opinions. This means doing research, asking questions, digging for clues in problem areas, and paying attention to the signs when they appear and performing Intelligent A/B tests.
Setting the stage
A critical prerequisite of data-centered optimization, with digital analytics, is about making sure the data in your analytics tool is setup correctly, otherwise you won’t be able to draw accurate conclusions from your data and might even end up making decisions that hurt your business. Check out this blog post from Freshegg, on how to do a Google Analytics health check. It’s a great read that will give you everything you need to optimize your Analytics setup and avoid the problems that arise from poor implementation.
Making sure your web analytics tool is properly set up can be a tedious process, where you actually have to go through each page and check whether pageviews are being counted correctly and relevant events and goals are being triggered.
Once you verify that your data collection is on point i.e., your digital analytics tool is tracking everything that you need on your site, and you can now trust the accuracy of your data, it is time to move on to the next step.
Asking the RIGHT Questions
Where do I look? What am I even looking for? That’s how most people feel when they first start dabbling with web analytics. Many marketers get overwhelmed when they first start looking into digital analytics, whether that be Google Analytics, Adobe or any other tool out there. The sheer amount of data, reports, and not to mention almost unlimited segmentation possibilities, can lead to a feeling of not knowing where to begin, and that’s okay!
Luckily, there is a simple solution to this problem. The solution is basically approaching Google Analytics with a specific problem or problems in mind. My personal mantra when it comes to GA is,” If you don’t know what you’re looking for, then you might as well not look at all”. That’s why I always recommend starting with formulating relevant business questions before even going into GA. This makes it much easier to choose the right sort of reports and data to look at, so you don’t end up aimlessly jumping between reports, scratching your head, and wondering where do I go from here?
But notice the very important and distinct difference between asking questions and asking “the right questions”.
Asking the right questions simply means, asking questions that are actionable. This means there should be a clear path of action to each question. Put in another way, you should be able to answer the question: “What am I going to change or do, based on the answer to this question?”
This process is very effective at helping you sift through the large amount of data, at your disposal and focus on extracting useful information for your optimization process.
But in order for you to find these questions, you first need to go through your website while trying to point out and note potential issues that might be negatively affecting your business. This process will help you form relevant hypotheses, that can then be used to form specific questions/problems, that can be further analyzed via your digital analytics tool. Try using the following steps as a guide:
- Start by going through your site, this means pages – both regular and checkout (If e-commerce) and look for things that might be causing an issue. Be very suspicious and try to come up with as many questions as possible.
- Look for potential measurement problems, like pages with no tracking code, missing or faulty events, multiple pageview events being tracked etc.
- Write it all down and ask yourself, what am I going to change based on these answers?
Once you’re done with that, it’s time to delve into Google Analytics and start investigating some of the preliminary hypotheses, that you might have conjured up while formulating your research questions.
Always look at the bigger picture
An important thing to keep in mind, before delving into GA, is that context is everything. This means, that you should always keep in mind that whatever metric you’re looking at, has a broader picture that might help you understand it better.
Jeff Sauer, Marketer & President of Data U, likes to use bounce rate as an example for this:
“….bounce rate counts the number of people who came to your website and only recorded one pageview. They could have left your site in a huff or their browser could have crashed. Or they could have spent 10 minutes on your site.
Whether a high bounce rate is bad or not depends on what you want the person to do. Do you want someone to view your site, find their answer, and leave? Or do you want them to stick around for a long time and window-shop all day long?
Are you focused on converting the people who can be converted? Then your bounce rate will probably be very high. And so will your conversion rate.
There are online millionaires with 95% bounce rates. There are websites with 5% bounce rates that are going out of business. Bounce rate is almost completely meaningless without business context.”
Does this mean that bounce rate is completely useless? Of course not, it just means that you should always keep the context in mind and consider the reasons that can lead to it being high or low for your site and the specific page that you are analyzing.
Think about your overall goal for this page, what are you trying to get people to do and as Jeff says, it might be that a high bounce rate is perfectly reasonable for that page.
This, by the way, applies to almost every other metric you can think of, everything from conversion rate, click-through rate, or time on page, is not set in stone. That’s why I’m not too fond of comparing websites in terms of their metrics because the truth is, most websites have very little in common. In fact, the only thing they do have in common is the variability of factors affecting the performance of each site, making a cross-site comparison a futile activity (for the most part).
Now that we’ve gotten that out of the way, it’s time for a step-by-step, breakdown of some of the key areas I usually look into, with digital analytics, when searching for optimization and growth opportunities.
1. Technical issues
This is, without doubt, one of my favorite places to look and where I always find conversion killers. It’s also one of the places that very few people are aware of and where the problems are fairly straightforward to fix.
So what do we mean when we say technical issues? What is meant here are cross-device and cross-browser issues. What we are trying to do in this step, is answer the following questions with our web analytics data:
- How does my site perform with different browser types?
- Are there browser versions, within each browser type, that are performing worse than others?
- Is my site working well with all device types?
- How does the user experience look like for each device type?
Once you define what you’re looking for – remember what we said earlier, always have questions! – it’s time now to open up your analytics and look for the answers. Let’s do this part together.
I’ll start by pulling up the browser report. Choose Audiences > Technology > Browser & OS. This will give you the following report.
On the left-hand side we have all the different browsers & operating systems, and on the right, you have all the accompanying metrics. We can quickly observe that most of the traffic is coming from Chrome followed by Safari, so those are the two major ones. Edge and Firefox are pretty much the same.
A quick glance over at the metrics, shows us that most users entering the site are doing so on a Chrome browser. Let’s keep in mind that this is only a demo GA account and the data is based on Google’s own ecommerce store – conveniently called Google Store. The reason why we’re mostly seeing Chrome traffic is probably to do with Chrome being a Google product and since we’re looking at the Google store, it makes sense why most users are on a Chrome browser.
You should, generally expect user and session counts to be more “fairly” distributed – between Chrome, Safari, Edge, and Firefox – on most other sites.
Just looking at the e-commerce conversion rates, we notice that Chrome is – once again – massively outperforming the other browsers. Chrome has an e-commerce conversion rate of 0.32%, while Safari is at 0.12% and the rest have 0%. Massive difference, right? Obviously, we must keep in mind that more than 5X as many users are entering the store via Chrome compared to Safari, but since the conversion rate is still based on 7.000+ Safari sessions, we shouldn’t be ignoring these numbers.
It could easily be that there was a technical issue behind these numbers, maybe certain elements of the checkout page are not loading properly on Safari, either way, it definitely deserves our attention. You could, for instance, visit the same pages via Firefox, Chrome, Edge, and Safari and compare the cross-browser experience.
Another thing we could do here would be to drill further down and look at how each browser version is performing. You can do this by creating a custom report or by just clicking on any of the browser types. It might be that a particular version of Chrome or Firefox, was performing much worse than the others. Knowing this might save you a lot of money in the long run.
Another important point is that you shouldn’t limit this process to browsers and browser versions, but you should also be checking out device performance and device versions, to find out how they’re performing individually. Optimizing for mobile is hugely important these days and knowing whether a site is mobile ready or not, can make or break conversions.
2. Site Speed
Most of us have, at some point, experienced a slow site, and know by now, that it’s definitely not a pleasant experience. It’s annoying and you feel like you want to leave as soon as possible. So it shouldn’t be surprising to know that most other users also feel the same way. In fact, multiple studies have shown that site speed impacts both conversion rate and sales.
Among them, is this study by Portent, which concluded that “When pages load in less than 1 second, the average conversion rate is almost 32%. At a 1-second load time, the conversion rate already drops to 20%. At 2 seconds, the conversion rate begins to level off at 12-13% and reaches it’s lowest at a 5-second load time.”
So, with that said, let’s pull up a custom report and have a look at how page speed is affecting our demo account.
This report compares the average page load speed for bounced users – users that left the site without any interaction – and users that chose to stay on the site. We can easily see that users who left the site, upon entering, experienced a slower avg. page load speed than the ones who chose to stay. This might lead you to think, “well this doesn’t mean, that this is the only reason why they left”. True, but you have to assume that it probably played a significant role, considering what we discussed earlier and also because of the significant difference in avg. page load of +3 seconds. So this is definitely something that should be looked into.
3. Funnel Analysis
The purpose of the funnel analysis report is to help you recognize and understand the parts of your sales funnel that are having the largest negative impact on your conversions. This is where you get to visualize the specific steps, in the funnel, where customers are dropping off, either by leaving the site entirely or visiting different pages.
You also get to see, exactly which pages, they visit after dropping off. But that’s in a different report, which we can have a look at, some other time.
You can pull up the funnel report, by going to Conversions >> Goals >> Funnel Visualization:
It can be quite disheartening to see this report, because all of a sudden you’re seeing how the majority of users, never even make it to the final page. Depending on how leaky your funnel is, you might even end up with 0 customers at the final stage. Nevertheless it is very critical to understand user behavior in order to make proper adjustments.
So, if we look at the users, beginning from the very first step. We see that 6.912 people entered the cart page. We also notice that the biggest drop off, in the entire funnel, happens between the cart and the second step, which is the billing & information page, where 65% never even make it. After that, it seems like the dop off slows down quite a bit.
We only have a 26% drop-off between the billing stage and payment, and another 25% between payment and order review. Finally, we only have a 7% drop-off between order review and confirmation.
It’s very common for the drop-off to decrease as customers move closer towards completion, because the deeper they get down the funnel, the more determined you would expect them to be. Even though we sometimes do see, in some cases, heavy drop-off at the latter stages, due to a variety of things, which we will be looking at, in a later post.
So in this example, we see that the conversion rate is about 18%, for people that entered this funnel. Nevertheless, we are also seeing some issues during the various steps, that seem to be hindering users in progressing through the funnel. So now that we know this, it’s time to figure out what’s going on.
Figuring out where to optimize first
But how do we decide on which page to optimize first? A good rule of thumb, that I always use is to simply “follow the money”. In this case, I would look at the overall drop-off rate at every step, to identify the steps that have the biggest up-lift potential on the overall conversion rate (18%). So starting from the top, we can do a little exercise, and ask ourselves: What happens if we can decrease the drop-off level on the first step with just 5%, so instead of a 65% drop-off we only get 60%? This would mean that 2.764,8 people would enter the second stage, instead of 2.446, which also means that 1.432 people would convert, giving us an additional 165 conversions.
The point here is to always target the areas with the largest potential increase in mind, so a 5% improvement will always have a bigger impact on 6.912 than it would on 1.267. You can eventually, move on to the rest of the pages in the funnel, but start by focusing your attention on the biggest gains possible.
Another important point to keep in mind is, the further down the funnel you get, the less of an impact you need, to get a big up-lift. This means that, once you get to the lower levels of the funnel, there’s a big chance that it’s only small aspects of your pages that are negatively affecting your conversions, as opposed to higher up.
On the higher levels of the funnel, you would typically need to make bigger changes to generate an uplift. For example, slight improvements to the checkout page would likely result in 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.
4. Segmentation
So we’ve covered a lot of ground so far and been through plenty of different areas, using digital analytics to improve conversion rates and sales. Next up is segmentation, which is one of my personal favorites, and something I consider a prerequisite for getting the most out of web analytics tools. Because we can’t really say we’ve gone deep on digital analytics, without mentioning segmentation, can we?
Segmentation is, in my opinion, THE most important concept that you should be taking away from this post.
Advanced segments give you the opportunity to isolate your data into meaningful groups, that you can then analyze and compare based on their behavior. As Avinash Kaushik says, “All data in aggregate is crap.”
A good example of a simple segment, that anyone can easily make, and which can provide great insights for your marketing activities, is when you segment conversions based on demographics.
So what you do here is, pull up your demographics report and segment the report based on converters vs non-converters. This will slice your demographics data and reveal the demographic groups that are actually buying from you. So instead of looking at conversions at an aggregate level, you’re now getting actionable insights that you can instantly apply to your targeting settings on Facebook, Google Ads or any of the other platforms, that you normally use to drive traffic to your site.
Knowing this, you could even go a step deeper and look at specific channels. Let’s say you’re driving traffic to your site via Influencers on YouTube and you wanted to know whether they were sending the right demographic to your site. In this case, you could pull up the source medium or campaign report and once again slice the data based on demographics and… voila. This should give you an overview of the different demographics entering the site from YT, and if you have goals setup you can even see how each group is performing, for each goal.
Hopefully this will have given you a taste of the magic of segmentation with digital analytics, keep an eye out for future posts, where I’ll be going more in-depth into this topic, building custom reports and using specific examples from the Google Store.
Conclusion
Analyzing large amounts of data, in digital analytics, and going through reports can sometimes feel frustrating and might at times feel like looking for a needle in a haystack. This is why its so important to formulate specific business questions before actually diving into your web analytics tool because it helps create a path for analysis and narrows down on the data that really matter to your business. The more you do it, the easier it gets and eventually you’ll become more comfortable doing data analyses via marketing analytics.
