The conversion rate. If you’re anything like me, it’s the single most important metric you look at next to revenue when checking analytics. But buried in that seemingly straightforward number are all sorts of nasty deceptions.
In this post I’m going to tell you why you should never take your average conversion rate at face value. You’ll learn how to uncover hidden insights, remove misleading “noise” and ultimately get a better understanding of the health of your business.
A Quick Definition
A site’s average conversion rate is defined as the total number of goals completed (usually purchases for eCommerce stores) divided by total visitors. So if you make 1 sale from 100 people visiting your site, you’d have a conversion rate of 1%. 1/100 = 1.00%.
Sounds pretty straightforward, right? What could be misleading? It turns out quite a bit.
Reason #1 – Everyone Can’t Buy From You
I used to own a store that sold trolling motors. Because the motors were so enormous and expensive to ship, we didn’t sell to customers outside of the U.S. Even if a customer from, say, Canada wanted to purchase, our website wouldn’t allow them to. So when Google Analytics spit out our “average” conversion rate, it was really artificially low because it counted all the people who couldn’t purchase even if they wanted to based on where they lived.
If you get a meaningful amount of traffic from outside of your business area, you should really account for this discrepancy. Otherwise, you’ll be working from a conversion rate that underreports your actual performance. The best way to do this is to create a custom segment in Google Analytic that includes results only for countries you ship products to and regularly track the conversion rate of this group.
Reason #2 – People Access Your Site in Different Ways
As I’ve written about, 2014 was a rough period for our radio business with the conversion rate declining throughout the year. At first, I thought there was something wrong with our business until I started doing some digging and discovered the problem: mobile traffic.
I realized that visitors using the desktop version of the site were converting at a rate similar to previous years. It was the massive growth in mobile traffic (along with an abysmal mobile site) that was bringing down the average. The problem was with our poor mobile site, but without segmenting our visitors I never would have known that by looking purely at the average conversion rate.
Equipped with that knowledge, we made the design to invest heavily in a new mobile redesign which ultimately helped us double our mobile conversion rate.
By looking at conversions by devices type, we would have caught a bug that costs us thousands in lost sales.
Looking at conversion rates across browsers is also critically important. Browser specific incompatibilities aren’t as bad as they used to be (curse you Internet Explorer 6!), but it’s still not uncommon for a site to work flawlessly in Firefox while a bug prevents visitors from checking out via Chrome – or vice versa. The same principle applies to looking at different rates of conversion across operating systems.
After looking at conversion rates across browsers last year, we discovered that conversions on iOS devices were a fraction that of Android. After digging, we learned that our checkout button on iPhones and iPads wasn’t working – the ultimate eCommerce fail.
We fixed the problem, but only after the bug had cost us thousands of dollars over multiple months. If we’d been tracking our conversion rates by device more closely as opposed to just looking at our average rate, we would have caught it much sooner.
Reason #3 – You Don’t Account for all the Noise
As you’re reviewing your end-of-month analytics, you come across a horrifying fact: your conversion rate has precipitously dropped by 40% from the previous month. Your mind immediately fills with visions of your hungry children, foreclosed-on house and the beater van you’ll be forced to move into.
One of the worst things about your average conversion rate is that it’s often filled with “noise” – data that skews your figures but isn’t meaningful.
Here’s an example: Let’s say your store gets feature in the New York Times, your traffic explodes for a short period and you end up with 10x the amount of monthly visitors. Because most of these visitors are just passing through (as opposed to potential customers) almost none of them buy and your conversion rate drops by 90%.
This traffic spike would be considered “noise”. Looking at your average conversion rate in a vacuum leads to unnecessary panic.
That’s why it can be helpful to track something I call “core conversion”. This is a sub-segment of your visitors that are searching for your core offering, and subsequently isn’t affected by spikes in traffic or others non-meaningful data.
In the past, I’ve tracked core conversion by measuring purchases only for visitors who came to the site after searching for our top 5 to 6 high-level, broad but commercially applicable keywords. Sadly, we can’t do this as much anymore with Google gutting organic referral data but you can still do it if you’re running paid AdWords campaigns.
While not quite as good, you can also track it by segmenting visitors who land on your top category level pages from organic searches. The majority of these visitors will likely have reached that page by a high-level keyword search query.
By measuring your core conversion, you’ll get a much clearer sense of how your business is performing without all the noise. If this metrics dips, you know there are some problems with your business – or perhaps the industry as a whole – that you need to dig into.
Reason #4 – New vs. Returning Visitors
If you depend on a lot of paid traffic, understanding the difference in conversion between new and returning visitors can make or break your campaigns. Here’s why:
When running paid campaigns, it’s crucial to understand your cost per acquisition (CPA). This is the amount it costs you to acquire a customer.
When planning campaigns, a lot of people will look at their average site conversion rate and their average CPC (cost per click) to estimate their CPA. If their overall conversion rate is 5% and they spend roughly $3 per click to get a visitor, they assume it costs on average $60 to acquire a customer. (It takes 20 visitors to get a sale at a 5% conversion rate. 20 * $3 per click is $60).
Repeat visitors almost always convert at a significantly higher rate than 1st time visitors.
But here’s the problem: repeat visitors almost always convert at a significantly higher rate than 1st time visitors. So in our example, repeat visitors might convert at 6% while first time visitors convert at only 2.5%.
When you’re buying paid traffic, you’ll be sending traffic that’s almost entirely 1st time visitors to your site. This is traffic that will almost certainly convert at a lower rate than your average, driving up your acquisition costs and wrecking havoc with your ROI assumptions made under “average” conversion rates.
To avoid this problem, make sure you’re using your 1st time visitor conversion rate when making any estimates and/or planning for paid campaigns. Credit goes to Bill D’Alessandro for this point, which we discussed up on our recent podcast episode about essential eCommerce metrics to track.
Reason #5 – Not Everyone is Trying to Buy
On my store Right Channel Radios, we carry a large selection of radio equipment for sale. But we also host a large technical library with dozens of technical articles. Four of our ten top most viewed pages are pieces of information content. And our #4 all-time most visited page – right behind our home page and 2 major category pages – is an information piece on installation troubleshooting that has a very small likelihood of resulting in a sale.
Excluding visits solely to pages like this from your conversion rate can be really helpful in getting a more accurate sense of what your true rate really is. This can be tricky to do, but it’s possible. If you have a URL structure that has a specific tag for informational content (“/blog/”, for example) you can create a custom segment that excludes any visitors who land on that page. And if you’re using Enhanced Google Analytics for eCommerce, you can look at the percent of people who actually view a product page and progress through checkout to filter out informational only visitors.
Moving Beyond Averages
If you’re not sure how your average conversion rate breaks down, head on over to Google Analytics right now to find out. Understanding the difference between New vs. Returning visitors, Desktop vs. Mobile and different devices types is really easy as Google Analytics calculates all those metrics for you automatically.
For other metrics like tracking core conversion and informational visits, you’ll need to create your own custom segments. It’s pretty straightforward to do and you can find a great tutorial to get started right here.
If you’re a private forum member, swap notes with other established store owners about conversion rates in our ongoing discussions on what rates other store owners are seeing, this one on desktop and mobile conversion rates and many others. Not a member? Apply today to join our community of hundreds of vetted, experience store owners.
Regardless, just make sure not to take your average conversion rate at face value. As you now know, there’s a much deeper story going on beneath the surface.
Photos by shazz and David Blackwell.
Post by Andrew Youderian
Ah contact forms….such as a standard feature of a website aren’t they? One would think that WordPress would have tons of great, easy-to-implement choices for this. And indeed there are TONS of choices, it’s just that most of them have… 
In the past few years a new breed of web hosts has emerged that focus specifically on hosting WordPress web sites. Some of the key players are WP Engine, FlyWheel, Pressable, Synthesis and Page.ly. These web hosts tend to be a little more… 
SMBCEO.com just published a guest post I did for them – “5 Ways To Use Twitter Strategically For Your Business” which gives you a few ideas on how to get the most out of Twitter from a business perspective. By…