marquage-analytics-1

This article was first published on October 28, 2015.

The implementation phase is a critical step in a web analytics project. We regularly see web analytics implementations that were disregarded until the very end of a website project and were not adequately tested. But it’s crucial to avoid taking any risks that might affect your data. Poor tagging can result in two things: loss of data, and/or incoherent data. And in both cases, all actions (like data analysis) done after implementation will be questionable, because these actions will have been based on faulty data. To avoid this, here are 6 tips for simpler, quicker and more reliable digital analytics tagging to ensure dependable data.

 

1. Have a finalised tagging plan

All web analytics implementations should be shaped by a tagging plan. This tagging plan will allow the developers in charge of the implementation to successfully execute the project and know exactly where they’re headed.

It’s important to take the time to reflect during this phase, before launching the implementation, for two reasons. Firstly, the technical teams will be able to gauge the work required – and therefore the timing and dates of deployment. (I say “dates” plural, as it’s entirely possible to iterate releases based on the tagging plan objectives.)

The second advantage is time-savings. As the thought process is done upstream, you’ll now only need to execute the technical tasks, and not have to question formats, whether certain buttons should be tagged, etc.

 

2. Identify the web technologies used

Today, there are a multitude of different technologies available for building websites. It’s possible to find relatively standard sites based on JavaScript, as well as sites based on more interactive languages like Flash in Angular JS or emberJS frameworks, which are popular right now.

These technologies will have a significant impact on the necessary tagging, so it’s imperative to take them into account when estimating the time required. In JavaScript, you just have to call a file, then use its contained functions. This is generally very simple and quick to implement.

With AngularJS, the efforts required are completely different. Without going into too much detail, this technology stems from MVC pattern and is based on components that must first be created before being included in the web application. So, that’s two steps, versus a single step for “simple” JavaScript tagging; this can therefore have strong repercussions on tagging time. Explanations of these specific formats are available directly in our online help to simplify the creation of directives.

 

3. Don’t stop at the first obstacle

Whenever we talk about “tagging”, we’re actually talking about web development. And in this area, it’s common to face technology-related difficulties. Often, we encounter web analyst teams who tell us it’s not possible to measure certain KPIs due to technical complexity. But it would be unfortunate to stop there. While measuring certain information can indeed be difficult to implement, it’s very rare that it’s completely impossible (after all, “Anything is possible with technology!”). In this kind of situation, it’s a good idea to get in touch directly with your digital analytics service provider’s technical teams – it’s unlikely that you’re the only ones with this problem, and even if it is an isolated case, your service provider’s expertise (combined with help from your developers) can help you work around the obstacle!

 

4. Choose the right tagging methods

When you measure something, you generally have several means to your end. Let’s take a simple example: tagging impressions of an ad banner. To measure this event, there are many different methods, such as passing via an ad server before calling the banner, using JavaScript to retrieve information at the moment data is sent, or even JavaScript functions called on precise events.

These three methods are completely different, each with its advantages and disadvantages. We advise you to carefully choose the tagging methods used, according to your priorities.

 

5. Don’t forget to think about site performance

Strong website performance is an absolute necessity for many companies. These days, when a browser loads a site, a great number of calls are made: certain sites largely exceed 1,000 calls per page. While a call to a digital analytics tool rarely exceeds a few tenths of a second, there can be hundreds on the site, translating to several seconds of waiting time for the user. Nonetheless, things aren’t so simplistic: Site performance does not just depend on the total number of calls.

Techniques exist for lightening the load. For example, asynchronous loading (loading several scripts at the same time) and loading scripts at the bottom of the page can generally help improve site performance, even if the calls are still present. But, in certain cases, we can’t wait until the end of page loading in order to be able to send data. It’s therefore critical to implement your digital analytics solution in the lightest way possible.

If you don’t have any videos on your site, why would you need to load the code allowing for the measurement of videos? There is no reason to do so, and we therefore suggest lightening the code you use. Additionally, if you have 30 banners on your site, it’s often simpler and lighter to send 30 pieces of information together at once, rather than in 30 separate sends. These tips will help you improve the performance of your digital analytics tools – and by extension, your site.

 

6. Test, test, and test again

We can’t insist enough on this point. This is the most important step in tagging.

As soon as implementation is done, you must test to be sure of two things:

  • First, that data is sent when you want it to be sent. If data is not sent to our servers, it will be lost, and you won’t be able to recuperate it. For this reason, you must be certain that everything works perfectly as it should.
  • Second, that the right data is being pulled into your digital analytics solution, and that data is consistent. Once all data is sent, it still might not be sufficient. If, for example, the page name is incorrect, it will be complicated to find the information you’re looking for.

Each site and each kind of technology has its own particularities. For this reason, a code that worked on 30 of your other sites may not work on your 31st site! So even when you implement a code that you already know works, it is still a very good idea to test again to be 100% sure that everything functions.

 

The very last point: Even when carrying out a simple update to your website, we advise that you check the tagging. Even changing a button or any other element can have an impact on tagging. Also, during site updates, it’s not rare for the web analytics script to be deleted for various reasons. To help you with your testing, we suggest you put routines into place with your technical teams to verify that the tags are still present. Today, tools like Selenium let you create testing scenarios to be launched when you wish. Playback each of these tests before deploying a new release. You’ll avoid backsliding and losing data.

 

Author

Product Manager Before assuming his role as Product Manager, Florian spent 3 years as a digital analytics consultant here at AT Internet. A self-taught learner, he dabbles in web development, computer graphics and hardware. What else sparks his interest? New technologies like the Internet of Things, web development technology, and digital analytics, of course.

Comments are closed.