Time for Marketing header image 1

Time for Marketing

2
Mar 2020

#32 - Jente De Ridder - A Vision for Sustainable Analytics Implementations

March 2, 2020

Superweek is a Hungarian conference hosted on the top of a hill, you can't run away, you can't hide. But Jente sais, he liked that, because everyone has to talk to everyone. To each his own :D Jente is on Linkedin here and if you would like to talk shop, chat him up on the business website.

Here are the links to the things we talked about on the podcast:

And here is the full presentation from his talk on Superweek so that you can follow along with the podcast.

 
Here is the transcript of the talk we had:

Jente: The framework has been implemented. What we've done is we've decided to make the framework open-source. It's available for everyone who wants to use it, it can be shared in the notes afterwards.

Peter: This is time for marketing. The marketing podcast that will tell you everything you've missed when you didn't attend the marketing conference. Hello. Welcome to the time for marketing podcast, the podcast that brings you the best marketing conference speakers directly to your podcast listening app. My name is Peter and this episode number 32. Well, we will be going to a conference in Hungary. Before we do that, as you know, podcasts are usually things that people should listen to. You, yes you, can help me to get more people to listen to this podcast. If you like the speakers that I had in the past, and I know you will love today's speaker, just tell anyone. Just people that you've heard that are using podcast. Tell them time4marketing.com is a great website where you can get an interesting podcast. Now, we go directly to Belgium. With me today is Jente De Ridder. Jente, hello and welcome to the podcast.

Jente: Hi, Peter. Thanks for having me here.

Peter: How are you doing? How is Belgium? I've always imagined Belgium as one of the European cold countries, is this so?

Jente: Well, we have global warming also here so it's getting better. [chuckles] It's true, we have a lot of rain but we do have our nice days as well.

Peter: And loads of chocolates, everything is better in Belgium. Do you also have a lot of fries or is that only a Dutch thing? The fires [inaudible 00:02:06]?

Jente: It's definitely a Belgium thing. We have the best fries in the world, the best chocolates, and also over 100 very good beers so for all those things, you should come to Belgium.

Peter: All right. You should be paid by your tourist community to help promote Belgium. Jente, you are the managing partner and a digital analyst at a company called Stitched. Tell us a bit about the company, and more interesting tell us a bit more about what you do. What is your everyday work like?

Jente: I'll start with Stitched. Stitched is a digital analytics boutique consultancy firm active in Belgium and in the Netherlands. What we do is we help enterprise clients to get more value out of their data. We are mostly focused on their digital data so our mission is actually to help those companies make use of the data they're gathering in tools like web analytics. Because what we often see is that those companies that have BI team or data scientists in-house that those teams are used to working with CRM data, point of sale data, but they don't really understand how the digital data is gathered.

Because digital data it's imperfect data, of course, and this can be quite hard for them to get their minds around. What we do with Stitched is, from our experience in the digital data, we team up with those internal BI teams or the data scientists and we integrate the digital data in the entire data sets, the entire data warehouse of the company. We mainly focus on challenges like how do you cope with identification in a digital environment and those kinds of things.

Peter: How did you get into analytics?

Jente: I started analytics over eight years ago now by working in a online marketing agency. I learnt everything involved in online marketing there, the advertising part, social, content creation, search optimization, and also analytics. It was really that data part that motivate me the most. After a year, I decided to switch to another company I could pick up a full-time web analyst role and I've been building a team within that company since then.

Mainly, everything that I know about digital analytics I learned it myself by reaching out to the measure community, reaching out to other people, reading blog posts. It's hard to start in digital analytics as there's not really an education course preparing you for it. It's really your own motivation and your drive to really understand things and go look them up yourself.

Peter: I've invited you to the podcast because you had a very interesting presentation at the Superweek conference in Hungary. That's a conference at top of a hill, how was that?

Jente: Well, it was a really nice experience. I've been to Superweek before also as a participant and I really love that conference. It's one of the leading conferences within web analytics or digital analytics in the world at the moment mainly because many of the thought leaders are there for the entire week and you have, of course, great presentations being given. The most valuable part is that everyone is there in the hotel for an entire week. There's nothing in the neighborhood around, so it's indeed on top of a hill, more than an hour drive away from Budapest.

It's in the middle of nowhere and all you have is the hotel, the lobby bar, there is a big campfire every night outside, you have a hot tub, a swimming pool. There's a lot of room for exchanging ideas with your peers, really going into discussions about analytics and that's what makes the experience really nice. I would recommend it to everyone active in the digital analytics sphere.

Peter: I've seen the pictures of bonfires at night at front of the hotel, that looks really, really interesting. Your presentation was called a vision for sustainable analytics implementation. We've chatted enoughI hink, let's go directly to your presentation. Jente, here are your five minutes.

Jente: What we've done with the team of Stitched with one of our clients [unintelligible 00:06:43] in the Netherlands. It's energy supplier, they're a market leader in the Netherlands and over two years ago we were asked by them to implement a new data layer because they were switching from hardcoded [unintelligible 00:06:54] implementation to a Tag Manager implementation and they also [unintelligible 00:06:59] a new data layer. They have a really complex landscape, they have different departments, multiple brands, so many platforms. There were like eight different platforms with all different CMSs being managed by different development teams, different marketing teams.

It's your typical enterprise environment where there's a lot of complex things and not everything is aligned. As a business they require to have numbers across those brands, across those platforms, and they want to compare those numbers only one dashboard, those kinds of things. We start thinking from there what is the best approach to implement a data layer here so one unified data layer across all those platforms.

Also taking into account the challenges within the web analytics that we saw, where one of the biggest challenges was, of course, that's normal, the original web analytics is page-based, so you track every time your route changes. That is not really sufficient anymore because more and more development frameworks are modular, like Angular, for instance, you have single-page applications. It's not enough anymore to know that the page has changed but you want to know what was on the page at the moment.

Same when you look at different devices being used, the screen size of people coming into your website is always different. What do they see actually, instead of which page has been loaded? Same when you look at personalization, we show different things to different people on our homepage, so just having a report where you know that your homepage has been seen 10,000 times doesn't tell you what was on that page at the moment people visited. Those challenges we also try to solve them with our approach that we're looking for. There was also the fact that the implementation of this new data layer would be really quite a heavy investment from the organization because of the scale of the platform.

This was also something that they were willing to do, but of course they don't want to do this every two years for instance. What is the case in many companies that you see today is that there are new implementations happening every two or three years because all too often development implementation is based on the specific vision of one person. The person that's in charge of the implementation at the moment [unintelligible 00:09:13] for instance. Once that person switches roles or goes to another company, someone else comes in and he has his own vision and they must go through an entire implementation again.

We want to prevent those kinds of situations and just make sure that the investment was worth it for doing it once and you don't have to do it every couple of years again. That's when we came up with a framework that we've called the Generic Digital Data Layer Framework, where we changed the vision of [unintelligible 00:09:40] starting from page-based tracking to event-based. Everything that happens on our webpage can be considered as an event because already it's all the user interactions happening that are already seen as browser events. For people who are familiar with a bit of customization policies they already work with those events probably.

Again, we want to track things like someone clicks on a button, someone submits a form, a specific piece of content has been seen by the user, those are all events happening in the browser. What we've done is we've made those events abstract as possible. We start thinking, "Don't think on a specific page level." Don't look at what is on that page and what do we want to track now, just think on the component level of a CMS. Within your CMS, your developers, they build components which can then be used to create pages by Content Manager. It's based on those components, that level that you will start thinking about your tracking.

Every time a button component, for instance, is being used we want to know if that button has been within the view of the visitor. Has the user seen that button, and we want to know if someone has clicked on it? Those are two events that you want to know for every button. We'll tell the developer start implementing those events on the component of the button and once [unintelligible 00:11:01] on the page, the track is already included and we don't need to edit them.

This has some advantages. That one, it's clear where the responsibility is for implementing tracking, it cannot be forgotten because it's already present in the CMS. Also, as an analyst, you know that that tracking is available and you don't need to create specific briefings every time a new page is created by someone. There's a lot of time saved there for the web [unintelligible 00:11:27] that you would normally be spending on creating briefings you can now spend on analyzing data. That's how we start our vision, really abstract events happening on the page.

We ended up with a list of I believe 15 components that are typically used within CMSs to build pages and on those 15 components, we had 20 or 25 different events happening. That's our entire list. We have a [unintelligible 00:11:54] with 25 events and then for every of those events, we just add in the variables that you need to know because as an analyst it's great to understand when something's happening, when is an event occurring. To make sense of it, to really be able to give advice based on those events you need to understand the context of events, that's when the variables come in.

For every event, you define a couple of variables that need to be present to be able to make your analysis. For instance, again, back to the example of our button, for every button component we want to know the name and the placements for where is the button placed on the page and maybe also the text of the button. This can be different variables being used for every button that is in place on the page. That's the idea of the framework.

The other challenge is you want to make it sustainable, you want to be able to be sure that you only do implement it once and not again every time someone comes in again, new people enter organization or when new tools are being used by the organization the organization switches from Adobe Analytics to Google Analytics. Those situations also would need [unintelligible 00:13:00]. What we've done, we want to make our framework completely [unintelligible 00:13:04] agnostic and we started there with not building a real data layer because the data layer is always agnostic, it uses specific syntax based on the tag manager you're using.

Google tag manager has their own syntax, Adobe doesn't even really have a syntax that they prescribe, they refer to the W3C guidelines created almost three years ago, Tealium have their own syntax for data layer. All those vendors have their own syntax. What we've done is we decided to just all the implementation stuff, the implication of the event itself, we decide to stick with simple vanilla JavaScript and HTML data attributes for the variables. We've created these JavaScript that just listens to those events happening on the page. It puts them in an array, just like an event queue.

We have an invent queue that builds up when those events happen on the page while the user goes through the websites and then we have a translator script that translates the entire event queue to the syntax that is expected by your tag manager. When you use Google Tag Manager, those events will be translated to a data layer low push syntax, if you have a Tealium the data layer will be translated in another way.

What we have is JavaScript in HTML data attributes being implemented on the platform. That's the responsibility of your developer and he does that on the CMS template level, not on the page level. Next to that, you have one script that runs on the website which will listen to those events, which are our event subscriber and then you have a translator script that translates those events to the syntax as expected by your marketing tools. From then on, it's the responsibility of your web analyst that's in charge of the tag manager to decides what events need to be sent to where.

He can decide we need the tag manager, I want these events to be sent to Google Analytics, to Adobe Analytics, to our marketing platforms, to our Facebook pixels, those kind of things. That's all in the tag manager. Again, you don't implement different codes for Facebook for Google, it's just one implementation, one event, and it can be sent to multiple tools but not by implementing the two specific code on your platform itself.

The benefits are, of course, the mutation is much clearer hence all the vendor-specific or the tool-specific things those are-- you expect that the web analyst [unintelligible 00:15:23] people work with those tools, that they understand how those tools expect the data coming in. That's a bit the framework that we implemented and what we've done is-- [unintelligible 00:15:33] I present this framework I got a lot of nice feedback on it from the people present. We decide to make the framework open source so it's available for everyone who wants to use it. It's not something that we claim so the open-source codes can be shared in the notes of the podcast afterwards for those interested.

Peter: We'll do that. That's excellent and good to opening your code up to people. Your framework sounds great, the question here is how big of a company should it be for it to be a good idea for them to switch to your framework and not go specific to one of the tools that they can implement themselves?

Jente: Some of the ideas of the framework in there just best practices which I would recommend to everyone. Using event-driven approach is something that everyone should start using. It's also what you've seen with Google Analytics switching to the new prescribed measurements protocol, they are switching to more event-driven approach. Data analytics is something very specific so for the size of the company to make this useful, the benefits are larger for large corporations. To give you an idea, some of our clients that are using it at the moment is actually in the Netherlands with over 3,000 employees, multiple brands.

We have Bose from the sounds systems, they have implemented it on their platforms worldwide, we have an insurance company in Belgium who has implemented also 2000 employees. It's the large corporations who are using it at the moment. The framework enables on any platform the thing is you need to do an entire new implementation of your data layer.

What we mostly recommend to our customers is when you will do a new implementation anyway do it in this way. You know it's future proof, it doesn't matter if you will be changing tools somewhere in the future or not but if there's no need at the moment to redo your implementation or existing platform then just stick with what you have and you can continue building on that. Because it's still quite an investment to just implement your entire data layer over again. That's really useful when you go to a new platform or are switching tools at the moment or something like that. That's a good situation to implement this one.

Peter: Will be a good idea when you're redoing your web page to also redo your whole data layer or would you first say that first to do all of the technical stuff for your new web page and then go do the analytics?

Jente: No. If you would redo your website just take the data and the analytics part with it from the start, just make it one of the requirements that needs to be included. Because also in the organization we work with, within the definition of done for an organization working in a [unintelligible 00:18:28] way, the definition of done includes analytics components as well. Tracking needs to be present and needs to be verified by an analyst before something can be released.

Peter: If people would like to talk to you about analytics implementation, where can they find you?

Jente: They can always reach out to me via my LinkedIn profile or on the Measure Slack community. For people active professionally in analytics it's called Measure Slack, go look it up if you're not part of it yet.

Peter: Add the link to the show notes to that?

Jente: All right. We go into MeasureCamp Bratislava within a month, at the end of March. Probably I'm also at MeasureCamp Amsterdam but I don't have a ticket yet, I'm on

the waiting list there, and also MeasureCamp Brussels later this year. I'm quite a fan of MeasureCamp.

Peter: [chuckles] I wanted to ask you what conferences would you recommend to people to go to but it seems that Superweek and MeasureCamps are the places for analysts to go?

Jente: Yes. Those are really community-driven events and I myself get the most value out of those events where you have a lot of time for networking and discussing with your peers. I often value those more than just really good keynotes but that's my point of view.

Peter: All right, Jente, thank you very much for being the guest on the podcast number 32. It was great pleasure hearing about the framework for analytics that you've

developed. I think that's it, you can say goodbye now.

Jente: All right. Bye, and thanks for having me, Peter.

Peter: Bye-bye.

[music]

[00:20:30] [END OF AUDIO]