Running an e-commerce store is all about realizing growth. Optimizing wherever you can to increase the amount of goods sold. You play a game of chess on multiple levels, improving your site performance, creating engaging copy, learning who your customer is and advertise your services to generate more traffic and building a brand.
Optimizing starts with learning through data, to analyze and discover conversion opportunities. Google Analytics has a nifty report, called E-commerce under the Conversions section, a fantastic mix of valuable insights of your online sales performance. In this post we’re going to analyze the Google Merchandise store, so you can follow along and understand how we go from data gathering to actions.
How is the Google Analytics Ecommerce report enabled?
Before we start, note that the e-commerce report is not implemented by default, just like a lot of reports in Google Analytics (GA), that needs to be configured to work properly. If you use Magento, you’re lucky as a GA integration comes out of the box. Very handy for a quick start.
The e-commerce report can be enabled through the documentation provided by Google. I recommend running it through the IT-department to make sure the implementation is done properly. I wouldn’t feel comfortable implementing it as an online marketer. Before you start the actual implementation you need to enable Ecommerce tracking in the Admin section within the View panel.
How to report and perform a good ecommerce analysis?
There are numerous ways to start an ecommerce analysis. Chances are you’ve googled other posts to discover how to find ecommerce optimization opportunities. A lot of those tips and tricks posts fail to convey how important it is to take a step back before starting the actual analysis. The articles jump into the most common errors and mistakes made right away, but it’s geared towards industry professionals and don’t teach you the proper techniques of how to analyze web performance through web analytics tools. Reading these articles aren’t geared towards your specific situation. You’ll looking the same pointers, blinded for what is really causing issues on your website.
If you jump into analyzing straight away, you’ll fail to properly frame the project. I’ve jumped into analysis right away a lot of times at the beginning of my career, thinking I had found a brilliant way to optimize, only to discover it’s so minor or a false positive that I had to go back into the tool again.
The most important elements for a thorough analysis consist of the following:
The first step to a successful analysis is to be very concise with the problem definition. Be specific on why this analysis came to be and what assumptions were made. Why have sales dropped?” is not a valid question, neither is ‘How to increase sales?”. Questions such as, “How to add an extra t-shirt sale when people have one t-shirt in their cart?” steer you in a certain direction and will ensure you’re gathering the correct data.
If you are tasked with the analysis by one of your supervisors, I urge you to first schedule half an hour or so to fire some questions around. You want to at least answer the following questions before you dive into the actual analysis:
- What do you expect from the analysis?
- How would you like the data to be presented to you?
- Who will the data be presented to?
- When will the data be presented?
- What sources do you want me to use for the analysis?
Some questions might be no brainers, but trust me on this one that a lot of times management won’t have a clear picture of what they expect from your analysis as well. This can be because online marketing is not yet fully developed within the company or the hunch is based on gut feeling and will settle down when you have a constructive talk.
What will you be analyzing and what is not part of the analysis? A possible scope would be only to investigate the amount of products in a cart during checkout, focusing on a specific product category that is lacking behind categories on your website. This will help with presenting your results to your stakeholders.
It’s important you define a time frame for the project. What period is used for analysis? In some cases you might have insufficient traffic. In this scenario you will have to take a longer time period such as the last three months, the past six months or even a year. WIth high traffic volume, the last month of last week might suffice.
What data were you unable to analyze? Did the tool have known shortcomings before you started analyzing? It’s important you specify them beforehand. If you’ve just implemented e-commerce tracking, the amount of data is limited, but nonetheless you were tasked with making an analysis. State that data is limited and suggested improvements aren’t scientifically viable to be implemented despite reporting them.
Explain the actions you have taken or going to take based on the findings you are about to reveal to your audience. You might also want to counter possible questions and have run the findings with the team who will be responsible for optimizing the store. Often times you’ll be in control, but you might also discover possible improvements in product categories which are not your responsibility.
The above checklist is not a hard truth, but it can help you a long way. There might be more specific requirements to your analysis. Make sure you have explored them beforehand. Brainstorm sessions where stakeholders can throw their ideas onto the table can also bring interesting topics for analysis. Do not start before you have all uncertainties cleared, or else you’ll deliver an underwhelming report where you’ve poured countless hours into.
If you want a deeper dive into how to report Google Analytics reports, you should read another article I’ve written where I run down the different techniques and tips on how to report web analytics data to different crowds within your company.
Benchmarking for context
Before we start our analysis and deep dive into the e-commerce environment, we have to do some preparation to get a sense of how our store is performing against competitors. We are going to gather data from external sources to get the grips of the e-commerce market as a whole and gather benchmarks. I find this highly relevant to answer questions during the presentation of the results.
Managers will often ask you how they stack up against competitors and it’s good to reassure or affirm their suspicions. It’ll show you are on top of the game and with your findings explain how to stay ahead or get closer to your competitors. In the long run you’ll get a feel for different metrics and you’ll be able to place your own results into context.
In 2019 Smart Insights analyzed, from a multitude of sources, the different conversion rates per region for e-commerce websites. Globally the average conversion rate of desktop devices was 3.90% and for mobile phones 1.82%, tablets 3.49%. In the US desktop managed to get a conversion rate of 4.14%, mobile phones a meager 1.53% and tablets 3.36%. Zooming in on the channels Paid Search has the highest conversion rate with 2.9%, followed by organic search with 2.8%, referral with 2.6%, email with 2.3% and direct traffic with 2.0%
What is in the Google Analytics ecommerce report?
After implementation you’ll have the following report available in the GA e-commerce report.
The overview is a quick glancy that features metrics such as revenue, conversion rate, transaction and average order value.
Shopping behaviour shows how many visitors have had no shopping activity, put nothing into the basket, have abandoned the basket or have abandoned the check out phase. This report will give you a high level overview of user behavior. This is especially handy for management reporting, but not so much for a deep dive into optimization.
Checkout behavior shows the steps in the funnel and abandonments by new and returning visitors. This report zooms in on the shopping behavior check-out abandonment phase. Note that the e-commerce report might have different names and amount of steps as it is a custom build report.
An overview of all individual products. Here you can see which products are most sold (quantity) and for what price (avg. price). You’ll be able to see the product revenue and the amount of unique purchases. This is a valuable report, because we can see which products result in generating the most revenue.
View unique transactions with metricus such as revenue, tax, shipping costs.
Product List Performance
Product divided by category. Useful when having a very extensive product portfolio. Think of Zalando, Best Buy or Walmart.
Using advertising on your own site that links to other sections of your website. It’s just like a mini Google Ads within your website, where you can see how your “ad” performs through the amount of impressions it receives, the amount clicks and the CTR. You can link this data to the amount of transactions and revenue this promotion generated.
Insights into the coupon codes used during the checkout phase. Useful for when you are distributing coupon codes on other websites or through promotional activities.
Has the same function as order coupon, but specified for a specific product, highlighting the revenue generated for the specific product.
Reveals which affiliate websites generated revenue.
The guides for the marketing section are limited, or otherwise unclear. While very useful, they are most valuable when frequently used. I can imagine the average e-commerce store won’t be a hardcore user of the marketing section, but those distributing a lot of coupon codes through discount website, will most benefit from the Marketing report in the e-commerce section.
Analyzing the Google Merchandise Store
To give you an idea of how I would approach an e-commerce analysis, I’m going to break down the different steps and metrics I would be using in a real life scenario. There is no wrong or right approach. Alright, there may be a wrong approach, like I did in the beginning, getting to a conclusion too fast based on too few metrics.
For this analysis I’m going to first take a high level approach of the Google Merchandise store. Having a high level overview will give you pointers towards potential data points you need to deep dive into. As you become familiar with an account, you might not have to do the high level overview approach, but I’m assuming you’re either inheriting an account or are new to analyzing e-commerce reports and want a full breakdown of the process.
I’m going to take the whole of 2019 as my point of reference. You may be reading this well into 2020 or even later. As I stated earlier, selecting a time frame varies from account to account, so it’s not relevant to the actual methodology. I will not add screenshots of the metrics themselves as I’ve learned that Google Analytics has the tendency to change its lay-out regularly and it’s about the core of the analysis, not fancy screenshots. I know, I know, I used some screenshots in other posts.
Also note that as we dive into the Merchandise store the lay-out may already be changed and the recommendations might not be relevant anymore for this particular account. The core principle remains, I want to show you how to analyze an e-commerce store with Google Analytics and find growth opportunities and transition beyond the account to get a feel for the opportunities.
Conversion rate per channel grouping
First I want to zoom out a little and see how the individual channels perform. We won’t be doing channel optimization, but it helps us to get a feel of how the individual channels perform.
|Channel||New Users||E-commerce conversion rate|
Wow. I’ve seen some sad accounts in my career, but this account deserves an Oscar for best drama. There is not one channel that converts above 1%. The amount of new users this website is pulling in though, is amazing and yet it’s unable to convert users to a purchase in this high level overview.
Ecommerce revenue per device
Now we need to have a better look across the different devices as well. Through this we can later on determine which device needs our attention first.
|Device Category||New Users||E-commerce conv. rate||E-commerce revenue|
|Desktop||389,259||0,06 %||$ 24,025.64|
|Mobile||176,624||0,30 %||$ 31,426,16|
|Tablet||14,291||0,59 %||$ 7,899.07|
Again the conversion rates are abysmal, especially on desktop which doesn’t even reach 0,10%. Mobile has a better performance, but also far below the 1,82% we have discovered during our benchmark research.
As we break down the shopping behavior report and get a global view, we can see a few things. Of all the sessions just 0.14% resulted in a transaction. I’m beginning to suspect there are some serious problems on the website that hurt conversions.
|No shopping activity||831,525||639,794||76.94%|
|Sessions w/ product views||187,601||137,620||73.36%|
|Sessions w/ add to basket||44,934||33,648||74.88%|
|Sessions w/ checkout||20,458||19,324||94.46%|
|Sessions w/ transactions||1,139 (0.14%)|
We see that the checkout is plagued with high abandonment rates. This doesn’t surprise me judging by the shopping behavior report and the conversion stats we saw earlier. For the amount of visitors to the website, the conversion rate is very, very low. Something else is curious as well, the amount of users that reaches the review page is lower than the amount of transactions. This can mean there is a faulty implementation.
|Billing and shipping||14,632||9,037||61,76%|
|Sessions w/ transactions||1,139 (7.78%)|
Page Speed analysis
I want to go to the page timings report and look if there’s more going on which might explain the different conversion rates we’ve seen before. The overall load time for the website was 4.03 seconds in the whole of 2019. This is above the recommended holy grail of 3 seconds, but not as bad as I was suspecting.
I keep my reservations for the page timings report in Google Analytics, as there are far better tools out there, but for the sake of argument for this post, I want to stick to what we have at hand.
|Browser||Avg. Page Load Time. (sec)|
I’m not as surprised by the stats here. The performance is doable. It’s not best in class, but there aren’t any problems we can see at first hand. Safari and Edge are a little slow, but nothing big going on that would indicate performance issues on a technical level.
The merchandise store
So far we’ve analyzed the different parts of the website, so it’s time to dive into the store itself. We’ve gotten different pointers which were pretty disastrous to begin with. The amount of traffic this site is pulling in is praiseworthy, yet it was unable to turn this traffic into dollars. We saw that 74.88% of sessions abandoned the website after adding products to the cart and another 94.46% abandoned the check-out. We’re going to do some old school user testing.
Looking at the basket a few things stand out. The desktop view is as follows.
As you might have judged by your own experience, this is one pretty bad page. The desktop version has a color palette which is not engaging to the average user. The design also oozes a year 2000 approach. This is interesting as most of the apps from Google are slick and well thought out. Modern.
Another element is the plethora of menu-items during check-out. It looks like the checkout is an overlay of the actual website. The color scheme is always way too muted in some cases. Just look at the continue shopping button which is almost just like the background. The continue checkout buttons are also misaligned, which is just an eye soar. The mobile view is even worse than the desktop version. The color scheme is way stale, cold. It doesn’t invite me to move to the checkout.
But the true culprit is the page after this one, where registering is necessary to complete the purchase. We’ve seen this years ago when webshops were coming up and registering was mandatory to complete a purchase. We now know that with the rise of webshops and the discontent with passwords and account management, users abandoned check-outs faster than ever before. More conversions can be generated instantly by allowing users to buy without logging in or creating an account.
A lot to improve
We’ve seen that through some basic analysis that the numbers were staggering. The amount of traffic the merchandise store was pulling in would make any average webshop owner envious, but the conversion was really bad. From the numbers itself it was difficult to determine where the real faults were. Nonetheless it gave us clues that while the website was technically on par with most webshops, users were not converting at all. We had to take a step back and look at the actual website and there we saw there was a lot going wrong in all kinds of places, especially during checkout.
Now I’m guessing you are wondering what’s next? This wholly depends on the situation you are in. If you are within a larger company where multiple departments are dependent on the same checkout environment, you’ll have to present your findings and possible solutions. In this scenario it would be to create a guest checkout option.
If you are responsible for the whole website, you can go your merry way and improve on the go. But you need to do this methodically and not go about this lightheartedly.