Sure, you have Google Analytics set up on your website to help you determine how many people visit your site, where they came from, and what they're viewing while there. But it's likely you have questions about whether you are using it as effectively as you can to boost conversions and sales on your website.
In this Q&A interview with EcommerceBytes, Web Analyst Sayf Sharif of LunaMetrics shares his insights on the most common mistakes website owners make with this tool, and how they can start using it to change the future and not simply report the past. Plus, find out what you should be reading on a regular basis, why testing is so important, and why "You are wrong" is the first rule of testing.
EcommerceBytes: What's the biggest mistake online sellers make with using Google Analytics?
Sayf Sharif: Assuming that their websites are correctly configured. It's that they never try and use Google Analytics to proactively make their websites better and improve their conversions. There is so much to Google Analytics, and most people don't scratch the surface.
Untrained "analysts" will look at their Unique Visitors, their Pageviews, and their general ecommerce data, but they won't jump into Google Analytics and look for insights. What email campaigns led to more conversions? Where in the shopping cart flow are people exiting rather than purchasing?
There are a million questions to ask, and insights to gain, and too many people simply don't do any real analysis, often because they're ignorant of the metrics and how to do it. They collect data, and report the past in a black and white way, rather than using it to change the future.
Above: a screenshot of a Lunametrics Funnel Visualization report, which visualizes the steps leading up to a Goal.
The Importance of Revenue vs. Conversion Rates
EcommerceBytes: In this blog post, you talked about the importance of conversions. Is that something you find your clients don't measure enough, and why do you think that is? Are they difficult to track?
Sayf Sharif: A conversion for one site is an ecommerce transaction, and for another it's a form submission. It can be hard for some people to figure out what their goals are. One of the first things I do with clients is try to get them to identify their KPI's (Key Performance Indicators) and then use them to create goals on the site. How exactly are we defining success?
Once those are determined, it's important to set up Goals in Google Analytics to help measure them and gain insights. I think for the most part people don't use them because either they have a hard time defining their KPI's or they are just ignorant of the existence of the Goals feature in Google Analytics. If you know about Goals, and you know what your Goals are, I just can't imagine you not setting them up.
EcommerceBytes: Which Google Analytics metrics are the most important and why?
Sayf Sharif: It depends on the site, but I don't know that I'd say there is any one specific metric to rule them all. For ecommerce sites, revenue is a big one. That's generally the be-all end-all of success on the site.
Some people will get caught up in Ecommerce Conversion Rate, but I think that's a mistake. Conversion Rate can be affected by many things, and is it a problem if your Ecommerce Conversion Rate dropped, but your revenue and Unique Visitors increased? Maybe you got a big new source of traffic that just doesn't convert as frequently as your previous average, but it's still more traffic and more revenue. You should optimize for Revenue, not for your Conversion Rate.
EcommerceBytes: How important is measuring/looking at incoming links?
Sayf Sharif: It can be important. The sources of your traffic and how they end up converting can be an important place to gain insight. Are your Facebook referrals far more likely to convert? That might be something you want to work on. Are you getting a high volume of traffic from another source because they are linking to you? It might be a relationship you want to take care of so they don't remove their link. It's just another area to look at to measure failures and successes, and get ideas for the future.
Keeping Up with Google's Changes
EcommerceBytes: Google changes things regularly. What do you recommend your clients do to keep on top of that?
Sayf Sharif: If your job is to use Google Analytics, I think you'd be smart to subscribe to the Google Analytics blog itself and read it regularly. They are constantly adding new posts, features, admin tools, and also linking to great new articles across the industry from Google Analytics Certified Providers.
Also the LunaMetrics blog is actually pretty great. We publish three days a week on Analytics, Conversion, SEO, PPC, and other aspects of Digital Marketing. I wouldn't wait three or six months to find out about new features you could be using right now to improve your website. Especially with the new Universal Analytics upgrade to Google Analytics, there are going to be tremendous changes over the next year.
Stay on top of things! If you can't do that, try and schedule an intensive training at least once a year. We've had plenty of repeat students at our Google Analytics training who just want to get brought back up to speed.
EcommerceBytes: Do you find many of your clients do not put the analytics code on enough of their pages? Or do you see mistakes with the way the code is embedded?
Sayf Sharif: When people come to us for audits, we see all sorts of problems. Deprecated code and old ways of doing things that don't work anymore. Ineffective ways of doing things or avoiding them altogether. More and more these days we don't see random pages missing the tracking code, probably because more and more sites are "dynamic" and that tracking code gets delivered to every page. I've never gotten an audit where their setup was perfect, but they WERE coming to us for an audit, so that's probably not surprising.
The most common error probably has to do with subdomain and cross-domain tracking. Many sites these days are using subdomains (like say secure.domainname.com for their cart rather than www.domainname.com for their main site) and they configure that wrong, or they have cross-domains with part of their site on something completely different like oursite.otherdomain.com and www.oursite.com and they haven't correctly tied the two domains together in Google Analytics to track traffic back and forth between them.
I'd say for anyone with more than one domain, to make SURE they are tracking subdomains and cross domains correctly, and that you don't have a ton of self-referrals (that is a referral source on traffic with your own website as the source).
The Importance of Goals
EcommerceBytes: Which Google Analytics metric is the least useful, or the one online retailers should worry about least, and why?
Sayf Sharif: The one that has nothing to do with your Goals. If you don't do paid ads, then the advertising metrics are going to be pointless to you.
Nearly every metric that you'll see in the standard reports is useful at one time or another. You might not generally care what mobile devices people are using, but you also might do an analysis that shows which mobile devices or operating systems end up getting people to spend more money via their devices, and then you might want to look into why.
The hostname metric usually is pretty meaningless, but if you get some weird traffic patterns, you might check there to see if someone has scraped the content from your site and has reproduced it elsewhere (along with your Google Analytics tracking code), etc.
Why You Should "Ignore Your Gut"
EcommerceBytes: In a blog post about "Ignore Your Gut," you say, "You are wrong. This is the first rule. You think you know what the right thing to do on your website is? You are wrong." Can you elaborate on that and what you mean by that?
Sayf Sharif: When people want to try and improve something, I'll often tell them to think up a few good ideas, and a few bad ideas, and to test at least one of both. People just innately think that their opinion is right, and at least with conversion and usability, it's almost always demonstrably not. Just because you like something, doesn't mean everyone else (does).
Could you be right? Sure, you're gonna be right on occasion. But if you go into it assuming that you are wrong, you're more likely to divorce yourself from a bias which could flavor your tests. If your bias says "well green is obviously the best color for this, let's just test size" then you might not test and find out that if you made it red it would improve your conversions by 100%. I find it best to just get people to start from a humble position. Sometimes the best ideas are the worst ones.
How Testing Can Reveal Surprising Truths
EcommerceBytes: You also talk about the importance of testing new ideas. Web design, email marketing, etc. Can you give some examples of testing from real-life clients or even the Lunametrics site?
Sayf Sharif: Most tests tend to be boring. "Hey, we increased conversion by 4% by changing the image size!" They tend to be incremental rather than huge and groundbreaking, but those tests happen too.
As an example of ignoring your gut, we one time decided to test a new banner at the top of every page on the LunaMetrics site. I thought it looked great, and that it would improve click through. We did some user tests and people didn't even see it. There was a complete ad blindness to it.
When we pointed it out, people all liked it, but NOBODY, literally nobody, saw it without it being pointed out to them. We learned from that and changed what we were doing. Usually you're so close to something that you just can't even see it with fresh eyes. You need new people to do the user testing, so that you can see through their eyes how someone sees it for the first time.
As far as why don't people do it? There is a wide variety of reasons. Sometimes it's laziness, other times people just can't wait for a test to run and want to change the site fast and often.
More often than not there is a technical limitation, where people don't have access to change their code frequently, etc. That can often be solved by using third party tools that hook into Google Analytics like Optimizely, but people still shy away from it.
Also, there can often be political problems where you have multiple "owners" of the website, and they disagree on which direction to go. Or sometimes people are worried that some visitors to the website will see one page one way, and others another way, etc. It's rare that people truly understand the significance, or they'd be testing more.
EcommerceBytes: I noticed when I came to the Lunametrics site I got a pop-up that asked what I came to the site to find today. Then it asked "Were you able to do it successfully?" - yes or no. Has this pop-up been effective? Do most of your visitors answer it?
Sayf Sharif: Right, that's Qualaroo, which used to be known as Kiss Insights. I'd say it's effective. By no means does everyone use it, but we have often gotten good feedback from it. It's the user testing equivalent of walking across a wood floor and hearing a squeak. It can be so much more than that too, but in generally I think it's a great way to get some feedback on what you're doing from actual qualified visitors to your site, rather than a semi-random test subject.
EcommerceBytes: What about tracking YouTube videos? How can sellers do that?
Sayf Sharif: I actually just recently put together some code to easily track YouTube videos on websites without a ton of integration coding - see this blog post.
Is it important? Well, that depends on the site. I think at the very least if you're embedding videos on your site, it's important to track whether people are playing the videos, and watching them through to the end. That lets you know if making that content is even worth your time, and whether it impacts your conversions.
If you're spending X hours on your videos, and they create Y conversions or Z revenue, whether it is worth it becomes a simple math problem. If you don't know how to connect X to Y or Z, you have an unsolvable math problem.
EcommerceBytes: What other advice would you give online sellers about testing their websites with Google Analytics?
Sayf Sharif: Do it! If you're not always running user and conversion tests, you're not doing your jobs. That might seem harsh, but it's the truth.