Tracking: costly mistakes and how to avoid them

This article is a long-form read based on one of our tracking webinars. If you prefer to listen to or watch the full webinar, you can access the recording here. 

As an online merchant, your insights, flow, and performance can only be as good as the data you are basing your decisions on.

It’s easy to get sidelined when this data turns out to be unreliable due to issues originating from bad data quality, data duplication, and messy Google Tag Manager setups.

This also impacts us at Revend, where we rely on clean, well-behaved data to build helpful operational monitoring models.

A good tracking setup sets you up (get it? ;)) for success. And success is what we want for you.

So we decided to share our best practices. In this article, you'll learn:

‍

1. What event-based tracking is, and why is it useful for your company. 
2. 5 reasons why your tracking setup might be broken.
3. 5 tips to improve your tracking starting today. 

‍

What does team Revend know about tracking? 

Our CTO Alex has ton of first-hand experience with setting up tracking. 

Some context: Before co-founding Revend, Alex ran 4 e-commerce websites for over 10 years in both the retail and direct-to-consumer segment. He developed with frameworks like Magento, Shopify, Spree, and WooCommerce for nearly two decades. He was also the CTO at online auction house Vavato for 3 years where he made sure the website could endure up to 300,000 daily visits without running into issues. 

Getting tracking right was an important part of his responsibilities, because in the absence of proper tracking Alex knew he could not count on the reliable reporting needed to optimise the platforms he oversaw with his teams. Most of what we share below was recounted from his experiences enriched by discussions we had with numerous specialist in the field.

The what, how, and why of event-based tracking.  

Let's start at the beginning. What is event-based tracking exactly? 

In one sentence: Event-based tracking is the collection of events based on user interactions. Examples of user interactions: view a product, add something to cart, take a checkout step, and a successfully check out. 

How is this tracking done? Generally, there are two ways to do event-based tracking. 

1. Client-side tracking: you generate data on the client-side, for example on the browser, and send it to the server. This is third party data, and will become considerably harder once more restrictions on cookies collecting this data will come into effect in the Summer of 2023. 
2. Server-side tracking: you generate data on the server-side based on your own business activity, for example purchases and add to carts. 

There's a lot more detail available to dive into, but for this article we will leave it at this high-level definition. 

Now that we have established what it is and how it is done, why should you perform event-based tracking? 

Because measuring these interactions lets you understand behaviour of customers on your webshop. And this understanding is a crucial ingredient to enable essential tactical and strategic actions for your business, such as:

• Reporting
• Revenue attribution
• Conversion rate optimisation
• Conversion-based AI-driven ad campaigns
• Personalisation
• CPA and affiliate marketing campaigns

When tracking is done right, it also makes it possible to monitor and adjust your operations in real-time. If a monitoring tool like Revend can base its models on accurately represented data, it can perform way better. 

There seems to be no doubt: as an e-commerce operator, event-based tracking should be on top of your priorities list!

Common tracking challenges

We have come across a variety of tracking challenges while helping our users set up their real-time monitoring script (in Google Tag Manager, DataLayer, or via Revend's API). While Revend has no impact on tracking whatsoever, our solution is highly dependent on getting clean and complete data to build good models. Here are the most common tracking challenges we have encountered so far:

• Breaking Javascript: If there is one little piece of Javascript in a Google Tag Manager container that is broken, then that whole container will not load. We have seen this happen to a number of our customers, and it severely impacted both their tracking abilities and Revend's ability to track their key metrics accurately.
• The 3rd party cookie apocalypse: In the 'old days', which in this case will mean the time between the invention of the digital cookie up to the Summer of 2023, it was quite easy to track an internet user's behaviour throughout their different sessions across the web. Now, this will become much harder. Contributing factors are the new rules in iOS which limit and disable third party cookie sharing and also Google's expected ruling to the same effect. 
• Developers needed: If you want to improve your tracking, but you are not a developer, you will be dependent on one. Maybe you use an agency, or maybe you have in-house developers you can count on. Whatever the case, developers are an expensive resource, and can be a challenge to come by in a timely manner. 
• Pixel overload: The number of channels to receive data has grown immensely. And with those channels come pixels, lots and lots of pixels. Facebook, Tiktok, Klayvio, Hubspot, etc. all offer you their very own pixel. Each implementation of a pixel takes its toll on your tracking setup, as it represents one more data source to reconcile. 
• Duplicated data: Next to breaking Javascript, this is the most common challenge we encounter. When multiple data sources (for example your ERP and CRM system) send the same kind of event into your tracking setup, you risk counting these events double (or in some cases even triple or quadruple!). A simple example tells us why this can be catastrophic: if you count your website visits double, then associated metrics like conversion rate and bounce rate are proportionally impacted as well, completely distorting your view on the reality of your performance. 
• Vendor Lock-in: If you have ever tried switching analytics vendors, you will immediately understand this challenge. Moving away from, for example, Google Analytics, is extremely difficult because your data is often kept in a consolidated format. Access to your raw data can be very hard (or impossible), making the move to another vendor very burdensome. Setting up tracking goes hand in hand with having good access to your data, making vendor lock-in extra painful in these cases. 
• Sampling: Once you reach a certain scale, analytics tools will start doing sampling of your data. This means your data will not be very accurate or representative anymore on an individual scale. You'll feel most of the impact of this inaccurate tracking when you are trying to figure out conversion rates on parts of your audiences. Luckily, Google Analytics allows you to circumvent sampling when you set specific conversion goals yourself, but otherwise your data will mostly be an approximation of reality rather than the actual raw data. 

What are 5 reasons your tracking setup might be bad? 

If you are reading this article, you are probably open to some constructive criticism about your tracking. 

Here are five reasons why your tracking might be open to improvement:

1. You use browser-side tracking

We see some of our customers using standard e-commerce platform plugins that use browser-side tracking. Popular examples are the standard Shopify or Google Analytics integrations. 

These solutions offer tracking, so what's the problem? 

• E-commerce vendors offering browser-side tracking do not want you having good data, they want to upsell you their reporting tools. So there is an incentive there to not give you the best tracking experience from the start. 
• Theme developers often do not add tracking code, forget to add it, or do not maintain it pro-actively. It does not help them sell themes, and you probably never explicitly told your freelance developer you wanted a tracking setup. 
• Browser-side tracking needs cookie consent. Cookie consent has gotten harder to obtain, and without it you will have to depend on Google Analytics or other third party solutions with server-side tracking to fill that void and keep an eye on the top of your funnel. 

2. You use a Tag Manager and generate events manually

If you are using a Tag Manager and you configured certain events or clicks on your website to generate synthetic events using Google Tag Manager (GTM), this means you probably do not completely trust the plugins you have available on your e-commerce platform or you do not have developer in house to put the event-generation code inside of your theme.

Nothing wrong with that of course, but we do see that this can lead to a number of practical challenges:

• You might have missed a number of use cases you did not think of when manually setting up your click and URL-based triggers, causing events to go missing. For example: checkouts triggered from pop-ups or via old outstanding orders are sometimes not counted as an event trigger by mistake)
• You can do a lot with dimensions, but you can not track products. Which means it becomes very hard to track performance overall. 
• Rage clicks (multiple clicks in quick succession) and unforeseen UI circumstances can lead to duplicated or incorrect event tracking.
• We have come across a lot of GTM setups where GTM does not trigger without first obtaining cookie consent. Because of this you risk losing a lot of non-personally identifiable information that you could have tracked even without a cookie. 

3. You use a different trigger for different pixels

If you use a click trigger for one pixel, and a pageview trigger for another, or any other combination, it will be very hard to get a clear view on your data between and across the different tools that you are using. 

• Data will rarely match up between GA, GTM, Facebook, Google Ads, affiliate platforms, CDP, and CRM if you are feeding them different data at different triggers. 
  
• The previous point directly affects conversion and ad performance as well. How can you compare the performance of different channels if they are not reporting on the same data?
  
• Your own reporting will also not match up with the platforms' reporting, creating potential confusion, frustration, and delays while making important investment decisions. 

4. You are on Shopify, but not on Shopify Plus, and can not track checkout

As the title suggests, this reason is specific to those of you using Shopify, but not Shopify Plus (yet). In this case, you can run into a number of issues that affect your tracking. 

• Shopify does not track conversions by default in their GA integration. This is an obvious problem if you want to use conversion as a factor to personalise your ads. 
  
• You are not able to inject GTM tags or any javascript in the checkout flow.
  
• However, you can track visits to the "Thank you" page and track your successful purchases that way. You need to be careful in your implementation, however, and make sure you only trigger an event once to avoid generating multiple purchasing events and skewing your attribution. An alternative or supplement to this method is adding events when someone starts the checkout on the cart page.

5 . You have a mix of dataLayer, GTM, and GA-based tracking

This final point is something we have observed in nearly every setup we encountered when setting up or discussing Revend for mid-sized or large e-commerce customers. There is often a historical mix of using dataLayer, GTM, and GA-based tracking all together. It can be done, of course. But it is very tricky, and can become too complicated to ensure tracking accuracy. 

• Multiple data sources risk duplicating events, or publishing the same types of events with different or missing data (which is called schema skew). Unfortunately, GA does not automatically deduplicates data when this occurs. 
• Mixing these tools can result in no one knowing where to find which event, since they do not consolidate properly into Google Analytics or your first-party analytics solution.
• Pixels are fed haphazardly and can miss important information.
• As a result, your reporting risks being partly inaccurate due to missing or duplicated information.

So, what is the right mix? We will get into that below, but to give you a sneak peek: We suggest using the dataLayer as the master for everything you track in combination with using the theme to trigger events instead of GTM is the golden standard. This goes for everything that you need to track through the browser, but it is even better if you can use server-side tracking (as we saw in point 1 above). 

Let's dive into some actions you can take to improve your tracking today. 

What are 5 tips to improve your tracking today? 

Here are 5 practical recommendations to get your tracking to the next level, starting today. 

1. Use the dataLayer to your advantage

The dataLayer should be the basis for everything that you track in the browser. Having the dataLayer filled in in a way that is 'standard', i.e. in a way that is compliant with the standard e-commerce format suggested by Google, will allow you to feed that data into every pixel, tool, and analytics solution that you intend to use. 

One thing to keep in mind is that you need to set up your Dimensions in the dataLayer in order to use it properly. We suggest creating a schema catalogue for all of your events, to keep things clear. It comes down to making sure you always send the same type of data for each event from everywhere in your theme code and platform(s). 

If you are a Segment user, you can enable the Protocols feature so event schemes are enforced in an automated way. 

2. Use server-side tracking

As established above, server-side tracking is something that can increase the reliability of your tracking considerably. 

When you create a server-side container that runs tracking code outside of the browser, it gives you a number of great benefits:

• It makes your website a lot faster. You are not loading any third-party pixels or Javascript to do tracking, only GTM. GTM will then call the rest on the server-side (not on someone else's laptop or smartphone). The data flows faster, does not get rerouted through any third parties, and is not at risk of getting blocked. 
• It is easy to set up, and readily available. Following the first tip above, you can get everything going into the dataLayer to flow server-side. Then you can still use Javascript to send data out to any pixels or analytics tools you prefer. 
• Another useful aspect of server-side tracking is platform tracking. This means that rather than getting a conversion event from the browser, which can be very unreliable due to refresh rates or gateway issues, you can make sure your server records the order and pushes the order to your tracking storage. This is often the most reliable way to track your conversions. There is a possible downside to this, as you might lose some of the attribution data that comes from the browser. There is a way to avoid this downside, however, which we will cover in a next tip.

3. Feed pixels and third parties through a server-side container

This tip is the logical result of the previous ones, but it bears repeating. Once you have the server-side container set up, you can remove every pixel you have running from your website, and run them through your server-side container. This will speed up your website, make tracking much more reliable, and make it very easy to add a new pixel. Rather than having to set up new pixels through a different GTM setup each time, you can do it on the server-side GTM container through the dataLayer. 

Alternatively, and perhaps even better, you can use a dedicated tracking solution that can distribute events to other parties. Solutions like Segment and Converge come to mind, both are great tools to solve this pixel challenge for you. 

4. Track campaigns in the browser, conversions on the server

Tracking your campaigns in the browser, collecting data on which GTM tags were triggered to guide people to your website, is a good practice. Once you are interested in revenue attribution, however, you also need to include the conversion data that is collected on the server and link both data sources together. In order to realise this, make sure you have a matching customer ID for each tracked user on both the browser- and server-side. This way, consolidation is simple and tracking becomes more effective. 

5. Monitor your data and your operations

Last but not least, we strongly advise you to monitor your data and operations. 

• Data: Having a great tracking setup is important, making sure your tracking setup continues to run great even more so. Companies like Segment offer tools like Protocols to monitor your data for inconsistencies in schema and quality. In the near future, Revend will also extend its solution to include data quality monitoring to offer insights on how to improve the reliability of your tracking results. 
• Operations: With your data tracking in prime condition, you will be sitting on a goldmine of reliable data. Levering this data to drive business decisions will give your store a huge competitive advantage. If you are still trying to figure out how to set this up, Revend can help. Revend continuously monitors your e-commerce operations in near real-time to detect potential issues before they impact revenue. You can take a look at what type of issues Revend can help spot for you here. 

Closing thoughts

We are all proud geeks at Revend, so it is only fitting to let Yoda have the last word in the image below. 

If this article drives home one point, let it be this: Better tracking is the path to the revenue side. 

Thanks for reading this article. If you still have questions about how to improve your tracking or how real-time operational monitoring can help you detect conversion stoppers and increase your bottom line, please get in touch or check out more resources below.