Scaling & Automating Data Governance

April 23, 2020

Join Simon Pilkowski, Global Lead Digital Performance Management at Boehringer Ingelheim, and Andrew Geddes, Senior Manager, Consulting for ObservePoint, in this webinar, Scaling & Automating Data Governance, and discover how to:

  • Operate data governance at scale
  • Manage website mergers and updates
  • Govern multiple websites at once
  • Use data governance to comply with privacy regulations

Simon:

Thank you very much for having me. As Aunica said, my name is Simon Pilkowski, I'm the globally digital performance management within Boehringer Ingelheim's animal health business unit. And then in addition to that, I'm a lecturer in performance marketing in German UNIS for applied science. And as you might have, maybe a few if you joined the virtual analytics at the beginning of this year, you might have seen this quote by myself already. So that's why I tend to put it on every of those presentations. It's a fool with a tool is still a fool, but a fool with a process at least knows what to do next. And I think this is also very important when talking about governance of marketing tags at scale.  What we will be talking about is basically three major things. It's why you need a tag governance process and what it could look like. You will learn how ObservePoint can help you to manage a large scale of digital assets and you, and I think hopefully this is the most interesting part, and see how Boehringer Ingelheim used ObservePoint for over 550 websites and to fully automate our digital governance processes. And in regards to marketing technologies. 

So let's kick it off. Why you need a tag management process. And, I want to highlight one thing. So when you're talking about tags, you know, a lot of marketeers, they want to use new technologies. They find maybe even a free subscription of something they can try on their website and they tend to want to put every new technology on the website just to test out how it can help them. But the problem is if you don't have an appropriate process in place, you might end up actually with this huge trash heap you see on the left hand side. So, basically you need to ensure that you're not harming the performance of your websites. In terms of speed, for example, we all know a website which is slow has a huge correlation, and also causation, for click through rates and conversion rates. And even if you're loading everything asynchronously, you will never know, again, which tags are implemented on your website and how they should be working for you. So that's very important and why you should have a process. And, let's look into, okay, how could such a process look like and then it's kind of easy, I would say. So it's very simplified. Yeah. And I try to not make it like a super complicated process which you would be able to build within weeks. So basically what you need to do is have a plan. Yeah. You need to know which tags do you really need? What is nice to have, what is nice to explore, but what do you really need in order to achieve your marketing targets? That's number one. And if you don't have this, you might end up already putting everything into your website, although you're already counting certain success measures with other tags, but you maybe don't even know about it. 

The second phase is getting into your website. So implementing it and here, hopefully, everyone nowadays using a tag management system, doesn't matter which one, if you're using GTM from Google, if you're using Launch on Adobe, Tealium, there are others as well, but you should be using a tag management system. If you're not aware of what a tag management system is and you're not using one yet, my recommendation is definitely start using one. It's basically an abstraction layer in between your website and all your marketing technologies and analytics technologies. And the third phase could be also three to five, but I put it into bucket number three is test, deploy and test. And this is very important. So you need to ensure once you implement your tags, they are working as expected and also not interfering with others on your website. You should do this in a staging environment. You should deploy and also do it afterwards.  Because very often you can also see something worse on staging, but it doesn't work on production. If you don't check it again this might harm your performance as well. And the last one is a little bit what I mentioned with this trash heap, it's the review phase. So ensure the tags you have deployed are still valid on a frequent basis and if not, decommission them. So be it quarterly or biannually, set up really a review process for yourself to see which tags are still valid. Is this campaign still running? Do we need to still fire those conversion pixels and so on in order to clean up your house for this one. So that's an example of a process how it could look like.

And if you're thinking, okay, processes, this is really, I mean processes might be boring, right? But they help us to achieve business consistency and what a process can help you with as well is it can also help you to, for example, move into usage of certain tools which would then be able to automate a lot of processes we tend to do manually. So we all know as analysts for example, the ETL process is like 80% of our work, extracting data, loading data, cleaning it up. And here we're preventing something, we're preventing garbage collection as well. So really you can, with a tool in place, you can also automate all those processes. So a lot of those processes. 

A second point is also you can provide consistent governance performance measurement, which means you can set KPI for yourself and for your governance and really monitor it in a consistent way. Yes, you can do it every time manually as well. But let's be honest, it's like with a study, if you do it manually every time your methodology might change slightly, one employee does the assessment slightly differently than another one. And then to be honest, the data you're comparing might not be as correct as if you automate it. And I think one thing which is also very important is if you have a tool in place that can help you also to highlight issues proactively. So you don't need to search for them, they would come to you to your inbox. And I think that's one of the big advantages of having also a tool in place to help you to actually bring your governance to life.  

And talking about the tools. So I would like to share a little of our story with ObservePoint. And I wanted to start with a problem statement at the beginning. And basically when I joined we had this huge amount of domains we were looking at across regions, across business segments and the question was to due to the fact that we had a merger three years ago. Two large scale companies actually moved together. So you can only imagine it's a lot of different processes which put in place a lot of different technologies that were in place. And now with this very diverse ecosystem we are starting to harmonize a lot of this. And key questions were actually, which websites are still valid websites and that's the life and we need them? Which marketing technologies are we using on those websites? And also from, obviously, a cost saving perspective, also how compliant are our websites in terms of internal standards within our marketing technology landscape? And here we went for ObservePoint and started basically to set up so-called audits. And so we created a template audit within ObservePoint but I think we have Andrew on the call, so you might be better to explain what an audit is and then I'll try to say what we did about it. 

Andrew: 

Yeah. Great. Thanks Simon. An audit is going to help you to, you know, automate that manual process, right? If you're going to go in and see what tags are on all your pages, you have an option of opening up your tag debugger or network stack and clicking through hundreds or thousands of pages. ObservePoint audit is going to allow you to do that at a large scale. So you can put in the domain that you want to scan and it's basically an overall site scan. It's going to go through your site according to the pages that you want it to run to. And it's going to collect all the tags and the information that those tags are collecting and display that in ObservePoint. And that's what the screen you're looking at here is, that's the audit score and the audit summary page. 

Simon: 

Well, what we did is, basically we used the interface of ObservePoint and created this template for us with certain configurations. So, for example, what is the domain you want to crawl? Which page do we want to start at? And how many pages do we want to crawl, and how frequently is this audit going to be run. And with this template in place we did something interesting. We basically went a step further and said okay, but now we have a lot of websites, we have a lot of domains. So we basically programmatically took all our domains, you can do it however you want. You can use an Excel spreadsheet, you can use a CSV, you can have it in a database. And this is what we did. We have a database with a lot of information about our domains and we basically, for every website iterated through this, and said, okay, we're loading a template audit. We're creating based on the template audit, we're creating a copy of it and within this copy we're replacing all those values I just mentioned. So, obviously, for all our domains we want to set the domain we want to crawl, we want to set the starting URL, we want to also, for example, exclude certain sub parts of the domain. And all of this is being done automatically within the replacement of this template audit. And based on this replaced audit, within setting it in ObservePoint fully programmatically and scheduling it afterwards. So on a weekly basis we're getting all of this data for currently 575 audits we've created and via the API. And this looks like this, so this is the ObservePoint, interface and you can see we have 575 audits up and running currently. You see here, this is our biggest OTC brand. So you see a lot of cats and dogs in the imagery. So this is always nice I heard on the internet to show. And basically we achieved a full automation of all our governance in terms of, "Okay, do we know which tags are there, do we have Google Analytics, do we have Adobe Analytics. I've seen Puig, I've seen FullStory, Hotjar or those marketing technologies. 

And what we can do based on this information is we can on the one hand side and sure up our standard technologies, which for us, is Adobe. It's not a secret because you can look on our websites. Is the data correct we're collecting? So you can set up certain rules. You can also validate whether tags are being fired twice, three times on a certain page, which will also inflate your numbers. And plus in addition, and this is very important, is we can also start using this tool to get rid of technologies we don't want to use anymore because we all know, two analytics tools in place, for example, is more confusing than helpful for anyone. And for us the standard is Adobe. So if we found Google, for example, we can say, okay, why do we still need it, without starting a conversation about which tool is better, that's not the purpose. The company works with one tool, we should stick to this standardized tool. So this is basically the overview. And the beautiful thing is about, if you're capable of managing the API,  there will be actually something at the end of this call as well, is you can also not only create those audits because as I said, if you're managing a large scale of digital assets, you will maybe even not have the time to go into every audit and go into the interface of ObservePoint and click through every audit because there is that huge breadth of information in there, but sometimes it's not convenient if you're talking about 575 audits. 

So what we're also doing is we're using the API to pull all this data. And we're pulling all the audit data into our databases and then doing something with a data viz tool. So some of you might know which tool this is by just looking at the design of the interface. But basically it doesn't make a difference. Yeah. So what you can do then, and here it's a little bit anonymized, but what it enables us now is we can quickly see which marketing tags are used on which website, how frequently are they used across the websites, how many websites have a certain marketing tag, etc? So let's take the example of an analytical tool. We're talking Adobe or Google or Puig or Matomo, whatever, and we can see how frequently and where those tools are installed. And our objective is obviously to harmonize, which means we would only have one tool on all those websites in the future. 

And this can be enabled by such usage of the API as well, where you're not going manually through the individual audits, but pulling the data programmatically and then you can narrow down and filter and there's so much information that we could have spent most probably half an hour just digging through a dashboard and then another full blown day to really see what kind of information in depth you're getting. And if you're curious, and that's what I mentioned before, if you're curious to maybe expose for yourself because you're also operating in multiple assets, not only one, because if you're doing it for one, you can set it up manually right. If you're doing it for two, you can set it up manually. My tendency usually is, if it's more than two, I already am thinking of automation, but with 575, I think, it's understood that, we didn't have the time to sit in the interface for weeks. 

And what I would like to share as well as we have developed all this integration with the API in Python. And we've now created a Python notebook. And following this link you see here on this presentation, you can, after this webinar also download it. If you're curious, and want to start immediately, feel free to download the notebook. You would just have to have an API key which you can put in there as well, and then run some methods and see how this works for you. So that's basically, what I wanted to share with you. So mainly talking about the three takeaways we have and hopefully you have, and I hope for everyone and it was very interesting. And, even if you have questions afterwards, obviously we have a QA,  but feel free to reach out to me as well because obviously this is a very detailed topic and you cannot nail this in a mere  30 minutes. 

So, the first key takeaway I think you should have is you have to have a governance process in place. Otherwise you've seen this picture of the trash heap. So really this is the first takeaway and maybe it's not even on the slide, but if you don't have a tag management system, you should better have a tag management system, whichever. And second one is automate as many steps as possible within your governance processes because you don't want to waste valuable time to go through your websites every time to QA it manually and here ObservePoint can support and I think even more crucial is then also define a clear set of success measures and KPI for your governance in order to make it also visible for your peers and also for maybe for senior management so you can show also how you improve your performance. Thank you very much. 

Andrew: All right. Thanks so much Simon. That was awesome. So I just want to go over a few things and Simon's hit on a lot of this, but I have been working with clients for about five years here at ObservePoint and helping them develop their tag governance strategy. There's some recommendations that I would point out. And like I said, Simon hit on some of these so well, and the first one is to have that plan. You know, even if you're not using ObservePoint, you can still go ahead and get your tag governance strategy started. And that's first by identifying the tags that are approved, you know tags that you expect to be on your site, and really just knowing what's on your site that's with your tags, it's also with your cookies, and identify, you know, what is approved and what you would identify as rogue or you know not expected or not approved. 

Then I'd run a scan. Like I said, with an audit, you can do that easily at scale. In ObservePoint, you can do thousands of pages, hundreds of thousand pages with an audit. But either way you're going to need to take inventory of your site and you're going to need to understand what requests are on your side and what data those requests are collecting. And then finally, you want to catch the errors. You need to know if there's an issue and if there's a challenge and the automation is key. One of the big things you can do to automate is to build rules and ObservePoint. So if there is a data issue or a data leakage or a tag or vendor that's collecting data that you don't expect, you can have a rule that will proactively alert you if there's a data issue. 

So if you do have ObservePoint or if you're considering looking at a tool like ObservePoint or a data governance tool. This is what your setup would look like. First thing you're going to start with that large scale audit. You're going to go ahead and put a thousand page, ten thousand page, a hundred thousand page audit in place to identify the tags. And in just a fairly short amount of time you'll have that information in the platform. So you can download your tag presence export. It will show you all the tags that are present. In addition to that, you can get all the variables and all the data that those tags are collecting too in a downloadable format, or if you're more advanced like the Simons of the world,  you can go ahead and utilize the API and then compile a lot of that into data visualization tools, which is a fantastic approach to your automation process. 

And then I want to reiterate here. You know, building those rules out, getting those alerts set up to where you have a proactive approach and you have the early warning system. The other thing I wanted to call out here is when you're setting up audits, you can also account for things like opt-in and opted-out of cookies. So if GDPR or CCPA are important or significant to your organization and you've been tasked to make sure that you're compliant, you can use ObservePoint to set up scans to where you can see what tags are being collected when a user is opted in or a user is opted out. And so we have that capability in a user session, in one of your audits. 

Okay. Now, finally you want to make sure that your results matter, that people are getting the information and you are getting the information that people care about. One of the ways that I recommend doing that is to make sure you have identified all the teams and tag owners that are involved with your data governance strategy or just with your tagging in general. You know, if you have an issue with say your Adobe analytics tag, but you don't know who in the organization owns the Adobe analytics relationship or owns the technical aspect of updating or modifying those tags, it's very difficult to make changes or really get any action from this data that you're collecting. And Simon mentioned earlier too, using a tag management system also allows to manage those tags and a lot easier format. But really identifying the tag owners and that's internally and externally with your vendor context too. If you have an issue with an implementation, you should know who owns those tags. ObservePoint allows you to document all that information within the platform. If you don't have ObservePoint, I'd still recommend documenting this information and curating that and keeping that accurate. 

And then we also recommend setting up regular communication. So those individuals, those tag owners, they should meet regularly and you should have regular communication with them. We often refer to that as a tag governance council. And then finally you should make sure that you have your process in place. I love Simon's quote at the beginning. Right. You know, at least with a process you know the next step. So you should ensure that you have that, that process in place that if you do have an issue, if you do have data quality concerns that you have a process in place to remove or update the tags. You might even consider doing things like an integration with your ticketing system, like a JIRA to where an alert from ObservePoint would go right into the system and create a ticket automatically. So really, start with a plan and start building towards that automated approach that Simon has shared with us. And you'll have great results with your tag governance strategy. 

So thank you. And we'll take some questions.

Previous Video
Analytics Best Practices to Drive Value During Uncertain Times
Analytics Best Practices to Drive Value During Uncertain Times

Join John Pestana, Krista Seiden, and Eric Matisoff for a discussion on how to drive value with analytics d...

Next Video
Automating QA Testing for Failproof Analytics & Marketing Data
Automating QA Testing for Failproof Analytics & Marketing Data

Learn how to automate QA testing to benefit your organization with Chris Mavromatis and Mike Maziarz