Matt Maddox - Tag Governance Tips and Best Practices

October 24, 2018

Tag Governance Tips and Best Practices

By Matt Maddox of ObservePoint

Slide 1:

Thanks Brian. I hope you all have been enjoying the Virtual Analytics Summit. I know there are a great number of helpful and timely sessions going on today. This presentation is titled, “Tag Governance Tips and Best Practices,” and it’s for people who know they need to do something to make sense of the dozens of tags on their marketing technology stack that are on their web pages and apps.

Slide 2:

But first let me introduce myself. I’m Matt Maddox, I’ve been with ObservePoint for three years now helping customers learn how to make sure their marketing decisions are based on valid web data.

Slide 3:

So today’s session is all about tag governance. What it is. Why every enterprise needs it. And how it fits into the validation framework. It’s all about wrangling control of your own data so you can use it effectively to help your business get the greatest return on your digital marketing investment. I’ll sprinkle some tips and best practices throughout, but I’ll call out a few specific ones at the end that I think you’ll find very helpful, especially if you’re the one who is struggling to keep up with the demands of the moving target we call digital analytics.

Slide 4:

Okay, let’s first figure out what tag governance is and how it relates to what you do.

Slide 5:

One of the previous presentations today talked about the six phases of the tag governance framework. This turns a complicated process into a well defined process that helps you ensure accurate data and accurate data collection. This model helps you visualize and execute on what needs to be done all along the lifecycle of your digital properties.

We start with plan. This becomes your blueprint for how you want to execute on your digital marketing strategy. We’ll come back and talk about this more in depth. Next, the comply phase. This helps you make sure you implement to meet various standards. Now these standards may include legal or internal or vendor requirements. Part of this includes planning to mitigate the risk of being exposed to the liability of such things as rogue piggyback tags or third party cookies collecting non compliant data. Now this can be used as a real competitive differentiator as you build your trust with your users. Then the deploy phase. This is when you take the approved plans and put them into action across all your digital assets whether those be websites or apps and devices. And you automate the testing of your plans with the quality assurance phase before it goes public. And you do this as part of your phase over and over again. But once your technology has been released into the wild, we don’t stop there because then you have to make sure that it’s right and it’s right in production where people are seeing it and you validate it as you have it in production. Then there’s the ongoing phase to proactively monitor your website and your digital properties. You want to make sure it continues to produce that accurate data and reliable data that your organization depends on.

So, with that quick introduction to the tag governance framework, we’ll focus our discussion on the plan phase as the foundation of a successful digital analytic practice that has a proven return on your investment.

Slide 6:

So, what is tag governance? Most websites have at least twenty marketing technologies. And those are implemented across all of their pages. Some have many many more. And so combine the hundred of data points these tag represent, multiplied by the thousands or tens of thousands of pages on your site, and you have the potential to get buried under an avalanche of data.

So tag governance is proactively taking control of the data management. It requires you to determine what you need to collect and define the need for it. It allows you to cut through the daunting mountain of what may seem like unrelated data points in order to organize, manage, and validate it.

Slide 7:

A Harvard Business Review study found that less than half of all of an organizations structured data is actively used in making decisions. Wow. And less than 1% of it’s unstructured data is analyzed or used at all. Now this includes more than web analytics data, but you get an idea of the scope of the issues that we’re talking about. Now there could be many reasons for so little data being used including, maybe there’s too much data. This is the overload point, that the organization reaches and they get into a paralysis and they can’t really use the data so the decision makers just revert to their gut instincts. Perhaps you’ve seen that before. Or perhaps the data is not trusted. Too many times data has seemed to be wrong, and then they send you back for further validation and you come back and you go back and forth and finally, they just say, “We’re going to rely on our guy instincts.”

Either way, the digital marketing data systems become a very expensive paper weight, so to speak, if we’re not using it for making our decisions. So, looking at that unstructured data as it relates to digital analytics, this might be too much of a data problem too. Is a very very hard to take unstructured example and make something useful out of it. Things like trying to quantify brand sentiment form social media posts or identifying uploaded photo content. Or ascertaining reader reaction in the user comments. All of those are examples of unstructured data that we may be actually collecting and not using because it’s so hard to deal with. It’s a difficult problem, and it’s difficult to use it well. And it’s expensive to troubleshoot and maintain. So, in all of this with structured and unstructured data, if you’re going to all the trouble to collect and store it, why not get your money’s worth and use it? Or looking at it from a different perspective, if we can’t justify data with a business value, we don’t collect it.

Now, another thing that Harvard Business Review found is that more than 70% of employees have access to data that they should not have access to. So you have to remember and treat it like proprietary data. It is your data. You’ve collected it, and you should guard it as any other asset that you have though remember, there is liabilities to your organization for sloppy data handling and giving access to sensitive data to the wrong people.

So one of the things you might do, and this is a tip that we often talk about, is look at the last time your users have accessed their analytics account. If it’s been a while, then you might look at removing their access or suspending their access until they say, “I really need it.,” and they have to justify it to you. Those organizations, when I was a trainer, the organizations that I went out to that I could tell had good data governance and tag governance procedures in place, got back to me after I left and had maybe left my account there after two or three months, noticed that I hadn’t used it, and they just had a pattern of going in and making sure that I still didn’t need that account. And if I didn’t need it, they would just turn it off for me. You might be surprised at the length of time some of your users have taken to actually login. It may be years since some of your users have actually logged in. Clean those up.

And then, the study also found something about analysts’ time. 80% of analysts’ time is spent simply discovering and preparing data. So I have a couple of things to say on this. If the data is not used in decision making, why are we even paying somebody to discover and report on it? Right? And if we had a tag strategy that captured only what we needed, we would have more time to make decisions on that data because we wouldn’t have to be worrying about the time and resources it takes for unused data.

Slide 8:

The sheer amount of data collected sitting in unused repositories reminded me of Fibber McGee’s famous closet. Right? But instead of junk, we have data repositories that just keep collecting data that never actually gets used. The shelf life of data is quite short, so why would we want to store something that we’re not ever going to use? Because we wouldn’t suddenly use it a year later would we? So if we haven’t used the data in a year, let’s not rack up costs and liability by storing it. Let’s purge it to make way for something that will show a return.

Slide 9:

Now, in addition to those numbers, ObservePoint conducted a survey of analytics and marketing professionals among our own customers and others interested in data quality. We found that 39% don’t even document their analytics strategy. Over a fifth didn’t even know if they had an analytics strategy which is as bad really as not having one in the first place because you can’t get any use out of it. It seems likely to me that those whom we didn’t survey would have a higher rate of not documenting their analytics strategy simply because, I would argue, their analytics programs aren’t as mature because their not using an advance validation tool.

What’s also interesting was that those with a documented analytics strategy were 83% more likely to trust their data. So this leads me back to the issue a moment ago. Your spending a lot of money and resources capturing data from your website, but if the organization isn’t using that information to make better decisions and drive your business value, what is the purpose of your digital marketing data? My hunch is that if we looked deeper at the reasons for not documenting the analytics strategy, there pretend some larger problems such as analytics strategies are hard to document. That’s a problem right? So it’s difficult to get the time or resources to do it. And analytics strategies are difficult to maintain. And so they might get started, but then they might get lost very easily.

Slide 10:

So it’s much easier to manage your multitudinous tags by planning them out in the first place then by forging ahead by putting tags on without thought as to how the data will be used or who actually needs the data or what questions should the data even answer.

Slide 11:

So, documenting our data is important for many reasons including, we’ll be adding content to our website and need to keep it consistent. After all, who doesn’t add content to their website?

Employee turnover leaves us all vulnerable to losing institutional knowledge. I know of an instance where this happened. The main developer left the company and they were in charge of the data strategy, the data mapping. But when that person left, they didn’t leave behind anything to document it and so that organization was in a world of hurt because they didn’t have a plan and they didn’t know how to maintain going forward. So, especially in this economy, we need to be aware that employee turnover is just going to happen.

So we take steps to make sure that the data is credible and trustworthy by validating it. Remember, much of the data is collected but never used in decision making. So I would argue that proof of validation is a key to data adoption throughout the organization.

And then, another reason for documenting our data collection, we have decentralization and democratization of data in the modern organization. That means there are more people than even throughout the enterprise collecting data on our website. So that makes it even for a larger problem than we might think of to begin with.

Slide 12:

So, let’s go back to the planning phase. This is where your tagging management or strategy originates. Everybody has heard of key business requirements. These are the reasons why you have a website in the first place. These are the goals for your online presences. KBRs are generally about how we want our visitors to convert. A KBR might be visitors must be able to search and find our content quickly. Or use traffic to drive advertising revenues. Or we want to sell more stuff.

Now, that selling more stuff, that’s what most everything really boils down to. Ask the bottom line whether that be selling more merchandise, increasing subscriptions, boosting our bookings, getting more loan applications, selling more advertising space, or persuading more people to a point of view. It’s all about selling, right? And that’s what our websites are geared for.

So the key performance indicators then come out of the KBRs. They measure how we’re doing with the key business requirements. So how do we know if visitors can find our content? How do we know if we sold more merchandise or if we got more loan applications? We make a KPI that’s measurable and say x number of searches leading to content reads. Or x number of widgets sold. Or x ratio of visitors to applications submitted. X percent increase in click traffic. All of those are measurable things. And so if we translate the KBRs to the KPIs then it makes it a whole lot easier to create what we call some sort of tagging plan.

Slide 13:

What does that look like? Some people call it a solution design reference or SDR. Other people call it a variable map or a tag strategy or a tagging plan. And as I’ve taught hundred of analysts over the years about these, it doesn’t matter what you call it. You just need to have it because it’s your blueprint for the tags you have and the data you’re collecting.

Speaking of tags, with all the marketing technologies on your digital properties, don’t try to manage all of that without a written strategy. You’ll go nuts, and you’ll fail.

Slide 14:

It’s usually a spreadsheet. A spreadsheet that lists every variable for a tag and other critical information for each of that tag’s variables. Now, I’ve seen very simple ones and very complex ones with all sorts of functionality in them. You don’t have to be complex, though you’re welcome to if that helps you out. So, what I would suggest is you keep it very simple.

Here’s a simple example of one that I whipped up. It’s from what you might see on a media site. So it’s simple enough to look at and understand how then you can take this and pattern your own. And you might make this in many ways, you might have different columns, but the important thing that it contains whatever is critical to your organization. That’s why it’s based off the key business requirements and the performance indicators.

Slide 15:

So let’s take it column by column. First, you want to list every single variable and show what their friendly name is as it appears in your analytics tool. It makes it so much easier to find them if you can say, “I know I need my image call type,” or “the writer of the story,” or “the story ID.” Rather than saying, “Was that cd5 or cd4 or eVar 3 or prop 2?” or whatever it is.

Slide 16:

So then you want to document the reason for having the variable in the first place. This column should be related to the KPI for the variable.

Slide 17:

A few examples of the way the data should look then be there. Here’s an example. Home or sports:scores:nba:suns or politics:election:prop-2-losing-support. Whatever your pattern is, give a few examples so that whoever is coming back to using this and either implementing it on the site or whoever’s checking and validating can see whether the patterns are correct.

Slide 18:

Of course, we need to know on which pages the variable should be set. Not every variable and not every tag is going to be set on every page. So you can describe it here or you can put maybe a regular expression here or another type of pattern syntax to show where it should be set. Maybe it should be set in certain directories or on certain conditions on a page with that all there.

Slide 19:

And then the data element. This is actually what captures the data. This maybe goes back to a rule in your tag management system. Or maybe it’s the data layer element that you’re using. Either way, you’ll need to know what to look for when you’re troubleshooting.

Slide 20:

And there needs to be a place to write notes or write exceptions. Maybe another explanations. Maybe these are “gotchas.” You’ll thank yourself later for this because it’ll come back and it’ll help you as you’re troubleshooting later on.

And maybe you have additional columns so this is a good starting point. But other columns might include the type of variable whether it’s traffic or conversion variable. Or maybe different analytics setting like if it has an expiration for attribution or if it’s set up for merchandise. All of these are important to set up.

Slide 21:

And then if it were me, I’d have either a separate sheet or maybe a different workbook for each of the marketing tags that apply to me or anybody who shares this. Ideally, you’d have one for each marketing technology on your site. Now some of those aren’t as detailed as analytics, but some of them are fairly detailed, right?

Slide 22:

So, let’s give a word of caution. A tagging plan can get so detailed and granular that it becomes difficult to sustain the effort you’re required to make. You want to make sure that the data here like the data you capture is no more or no less than what you need because it takes such a great deal of effort to maintain, it would be self defeating if you put too much in here. So keep that in mind.

Remember, setting up will take some concentrated time and effort. It may be a team effort. And you probably can’t do it all in one sitting, but be strategic about it. Plan it to be a tool that you and your organization can use to manage the marketing tags and the data they collect. And use it as a basis for validation.

Slide 23:

Now that I’ve mentioned validation, that is a primary purpose of documenting your marketing technology stack. You use your tagging document to make sure that the execution matches the plan. This helps your organization gain confidence in your data as you’re able to demonstrate how will the actual website data match the plan that you’ve built from your key business requirements and the correlating KPIs?

Slide 24:

And another advantage, is that by matching up the KBRS and the KPIS, you show how the data you track is critical to the success of the enterprise. Hence, has greater adoption in decision making.

Slide 25:

So, now let’s call out a few more tips. I’ve sprinkled some throughout here, some best practice tips, but there’s some I want to make mention of.

Slide 26:

First, let’s identify some of the challenges to validating your data even after you’ve built a clear effective tagging plan.

Slide 27:

You’ve got employee turnover, which we’ve already mentioned. You might put somebody in charge of it who then leaves the company with unfinished or with a lot of institutional knowledge.

Slide 28:

So, you’ve got silo teams. You’ve got a lot of departments within the organization trying to collect data, and they may not be talking to each other so there may be a great deal of duplicated effort going on. Missed opportunities, put best practices in to use.

Slide 29:

You’ve got technology limitations. Problem with technology is once we’ve got what we asked for, you find another use case that the technology doesn’t address.

Slide 30:

Also, in the past, there really hasn’t been a good way to create and maintain a tagging plan. A part of this has to do with the speed of deployment. Everybody is rushing to keep pace with the demands of the business and the customers. So we don’t want to fall behind serving our customers. So sometimes, we just say, it’s good enough and leave the task largely unfinished as we dash onto the next sprint.

Slide 31:

So, here’s some tips.

Slide 32:

I’ve got four tips. One is, and it’s most important I can leave you with today, is document your tagging plan. It’s not just a tip, it’s critical. The plan becomes the foundation for producing credible and trustworthy data analyses. And you rely on it to validate your digital marketing data.

Slide 33:

Next, keep the document fresh. One of the ways this can be done is by distributing the work and establishing global areas and team areas of the plan. The global areas reserve for those variables that are the same across the enterprise no matter the domain or the app or the digital property. These are standard and required variables, so they must get the buy-in of the entire tag governance team in order to modify them.

Slide 34:

But if you’re feeling a little inundated see that the load is daunting, you know to create and update tagging plan regularly. Consider that maybe you don’t have to do all of the work. Are there areas of the tagging plan that can be assigned to other individuals or teams to control? For example, think of domains that need somewhat different KPIs. Or maybe a business website has a hardware website that also might have a help portal. These are two very different even though they’re supporting sites, they’re critical to the business. But their conversion paths are going to be completely different. Their KPIs are unified with the company’s business requirements. But they’re going to be different.

So if you’re not directly responsible for all these digital properties, allow other teams or individuals to have ownership over their parts. The parts that they have the most invested in. that reduces pressure on a single individual to create and maintain it and to of course share the joy and responsibility of ownership.  

Slide 35:

So, this leads to a fourth tip, and that is share the document where everybody who needs access to it can refer to it, change it, and make suggestions on it. It should be a public document. Use Google Sheets or shareport or whatever enterprise level document sharing repository you have. Then put permissions on it so people can only change what they need to change or what they have oversight over. And make sure that your quality assurance engineers and your web developers can use it to build, update and test the site.  

Slide 36:

Now, there’s a couple more things. I just want to touch on this one thing, and that is don’t depend on the spreadsheet. We’re going away from spreadsheets. Just the tedium of creating and maintaining it is so daunting. And it’s even more time intensive and less valuable to validate it manually. So, after all I’ve said here, I’m going to tell you, there is a need to go away from spreadsheets and do something a little bit different.

So, even though everything I’ve said here applies to spreadsheet, it will also apply to what ObservePoint has been working on.

Slide 37:

We’ve known this has been a problem for quite some time, and we put a prototype together that our customers have used to automate the process and upload data validation rules right into ObservePoint. So, this prototype has helped create your own customizable online tagging plan that you can run an audit against and check each page for all the right tags and all the right variables and all of that without the manual labor and the effort involved in that.

So, we’re taking that prototype and turning it into a feature inside of ObservePoint. So, if you’re already using ObservePoint, you already have the basis for a tagging plan because we’ll take this tagging plan capability which has already been in development for quite some time. And reverse engineer your site based on the results of any audit you choose. Your site is already coded with tags and variables, we just take that, turn around, and use an audit that you have already been using, and it discovers what data is there. You can say what data is wrong and on what pages, and you can update your plan very easily.

Slide 38:

So that tagging plan manager creates hundreds of rules from the audit, letting you simply approve or unapprove patterns it’s found. So you don’t have to have the tedium of documenting every possibility, and you can track all of your marketing tags in a single place.

Slide 39:

Then after you’ve built your tagging plan online, you can run any audit against it automatically validating against your tagging plan. You don’t need to sit for hours navigating through your website, looking at the variables until you go crazy. It will free you up to be the analyst you want to be instead of a data investigator having to constantly go back to confirm your data accuracy. You may never need another spreadsheet again, at least not for tagging plans. So watch for that to come out.

Slide 40:

So let's conclude by reiterating a few things. First, whether you’re using a manual tagging strategy or you choose to use an advanced validation tool like ObservePoint’s upcoming tag planning manager, principles are the same.

Slide 41:

Planning a tagging strategy is the foundation for successful tag governance.

Slide 42:

Maintaining that plan to keep it relevant with the changing business demands is required.

Slide 43:

Use that plan to prove your data is reliable and trustworthy.

Slide 44:

And share the plan with the community, with the company, to overcome challenges of silo teams and data decentralization.

Slide 45:

So, I’ll be glad to take any questions, maybe you’ve been putting in questions as we’ve been going along here and if so, that’s great. But thank you very much for being a part of this session.

Slide 46:

So Brian, we’ll give it back to you. 

Previous Flipbook
ObservePoint Complete Solution Product PDF
ObservePoint Complete Solution Product PDF

Learn about ObservePoint's complete suite of tag governance solutions to help you validate your analytics a...

Next Video
Reports
Reports

Stop digging through code and network requests—use ObservePoint's aggregated reports to diagnose tagging er...