author img

Ewelina Łapińska

Web Analytics for e-commerce: 5 Most Common Mistakes

Even though web analytics is nothing new, some e-commercers are still making the basic mistakes, rendering it impossible to draw useful conclusions from data. In this article we take a closer look at five such mistakes and we tell you how to avoid making them.

1. Using Only the Basic Tracking Code

The implementation of web analytics begins with adding a basic tracking code to your website. It counts each view of your online store. But for some this is also the end of their adventure with web analytics. Meanwhile, this basic implementation (i.e. adding only the pageview tracking code) will tell you very little. You will know that a user visited your website but you will not be able to tell what he or she did there. Did they scroll it down? Did they click on an interactive element? Or perhaps they watched the embedded movie? To be able to tell that you need to implement additional events.

This is particularly important for Single Page Application (SPA) websites which don’t simply load content but work like an application – the contents is added dynamically (as in this example: https://www.meandem.com/). Google Analytics’ basic tracking module often counts only the very first view of a SPA website and all the other interactions (e.g. clicking tabs) are ignored.

You should also track your website goals, both in a macro scale (e.g. transaction) and in a micro scale (e.g. signing up for a newsletter). Also, monitor actions your users take right before achieving goals. This information is available in the Flow Visualization report or Goal Flow report in Google Analytics, or using Funnel: Bookmarks in Facebook Analytics.

In an online store it is also important to track product add to cart and start checkout events, as well as the efficiency of product lists. You can track these events using the Enhanced E-commerce module and – partly – Facebook Pixel. The information you gather can be used when measuring the efficiency of your campaigns on different channels.

2. Neglecting the Quality of Data You Collect

If you’ve implemented both the basic code and additional events, make sure you use them smartly. First of all: use proper naming and stick to your system so they are intuitive and easy to decipher by your colleagues (especially when tagging the PPC campaigns with UTM tags). Bear in mind that your modifications will not affect your past data - changing the name of a campaign or adding a new event will only affect the data you will collect after that particular modification.

Also, you need to remember about the technical aspects of event configuration. Decide when events should be recorded by your tracking tools and if they should affect the bounce rate. Example: a subscription to a newsletter can be recorded in your Google Analytics or Facebook Pixel as soon as the user hits the Submit button, instead of when they get a confirmation message. But this can distort your data – the number of subscriptions will not reflect reality when for example a user will not fill in an mandatory box and hit Submit. In some online stores an extended tracking module has been implemented and it allows us to learn about e.g. products that have been displayed. But - as is often the case in web analytics - the devil is in the detail. In case of a standard implementation of Product Impression events are only recorded when the page is loaded and not when products are actually displayed for the user. And since many users never scroll pages all the way down, the data on product displays are often overstated.

When configuring events, decide if they should affect the bounce rate – depending on how you want to interpret user intentions. Say you’re presenting a new product on a landing page and you want to collect a list of contacts interested in its premiere. But there is also a short movie embedded in the page. You don’t want to count as bounces users that watched the movie but didn’t enter their email address. Remember that all events affecting the bounce rate should be triggered by the user and should be purposeful (e.g. clicking play to watch a movie).

It is also crucial to precisely filter out the traffic generated by your team and any subcontractors you hire – this applies also to the testing version of your website (provided that you’ve embedded the same tracking code there). This is important for websites that enjoy smaller traffic – data on several dozen visits can greatly distort your statistics.

3. Not Understanding the Metrics and Comparing Them Between Systems

How often do you compare ROAS or click rate of the same campaign in Google and Facebook? Even though the metric is the same, the result will be different in each of these tools as they use different methods of assigning credit for conversion to a specific channel. For Facebook it’s 1 day since the ad has been displayed and 28 days since the ad has been clicked on, while for Google Analytics the credit is assigned to the last indirect source of visit before the conversion.

Bounce rate and time spent on the page are other misinterpreted data.

„In Google Analytics, only the sessions that resulted in one server request being submitted are counted as bounces, e.g. when user views a single page and then exits it without submitting any other request to Google Analytics server in that particular session.”

This means that a user that doesn’t visit more than one page and doesn’t trigger any of the predefined events related to the interaction in your online store will be counted as a bounce with session duration of 0 seconds. Make sure you remember about it especially in your landing pages – at first users typically browse it without making a conversion.

4. Neglecting the Context and Visualization of Data

Is the average order value (AOV) of $100 a lot? You will not know it unless you put it in a context, by asking the following questions: was the AOV similar in previous time periods? What percentage of your customers have a higher and a lower AOV? Focusing only on the average value can make you blind to other important information. It may be that a segment of your clients generates much greater AOV. But do they buy more expensive products? Or do they buy cheap ones but in big quantities? Data segmentation and its presentation on a timeline will help you understand how your online store is really performing.

As you can see in the below example adding context to your data allows you to draw more conclusions. And you’re still using the same data! The way your data is displayed is also important – picking the right diagram will help you see the truth hidden in your numbers.

5. Spending Too Little Time On Data Analysis

Every web analysis should begin with a business question. If you don’t know what problem you’re hoping to solve, how will you know what data to analyze and where to look for it? Whether you’re using Google Analytics, Yandex Metrica or Facebook Analytics, it all boils down to using the data to improve your business. Therefore it is important to work out a method of collecting and analyzing data. For example, you can generate specific reports within the tool or create interactive dashboards (e.g. in Google Data Studio) that collect data from different platforms, allowing you to spot any change in the metrics. It will also simplify cooperation within the team and between you and your client, as both sides will be referring to the same data.

Bear in mind that data analysis gives you knowledge on what is happening to your online store. And this is the key for finding some good experiment ideas for making improvements. Your statistics may seem fine at first but once you start inspecting them closely you’ll notice where you lose money. And only then you’ll be able to work out ways of stopping it.

Of course, the above five common mistakes are just a tip of an iceberg, so we will be discussing the topic further. Make sure you visit our blog regularly!

Do you want to build a store on Magento 2?

Read also: