#012: The Governance of the Data

Most people find the concept of governance about as interesting as an afternoon of quality control work at the beige paint factory. If you agree with this sentiment and are listening to this week’s podcast, we hope to change your mind! With special guest John Lovett, Senior Partner at Web Analytics Demystified, Tim, Michael and Jim talk about what governance is for a digital analytics practice, why it’s so darned important, and how anyone can get started. All of this AND a little poetry (really!) for the low, low price of 45 minutes of your time, in the Digital Analytics Power Hour.

Episode Transcript

The following is a straight-up machine translation. It has not been human-reviewed or human-corrected. However, we did replace the original transcription, produced in 2017, with an updated one produced using OpenAI’s WhisperX in 2025, which, trust us, is much, much better than the original. Still, we apologize on behalf of the machines for any text that winds up being incorrect, nonsensical, or offensive. We have asked the machine to do better, but it simply responds with, “I’m sorry, Dave. I’m afraid I can’t do that.”

00:00:03.74 [Announcer]: Welcome to the Digital Analytics Power Hour. Three analytics pros and the occasional guest discussing digital analytics issues of the day. Find them on Facebook at facebook.com forward slash analytics hour. And now the Digital Analytics Power Hour.

00:00:25.31 [Michael Helbling]: Hi everyone. Welcome to the Digital Analytics Power Hour. This is episode 12. Tonight we’re talking about data governance. Data governance and digital analytics is a pretty important topic. And one that probably doesn’t get as much attention and love as it probably deserves. Joining me are our two other hosts, Tim Wilson, and Jim Cain. And to add some additional spice to the mix, a man who needs no introduction, but I’ll give a small one anyways, John Leavitt. He’s a senior partner at WebAnalytics Domestified. Yeah, kind of a leader in thinking about this topic of data governance. He also has the small distinction of being the only person in the analytics community I’ve ever written a poem about. Maybe I should quote it for you right now. Oh, you must. There once was a man, Lord Love It, which goes to his nautical past, which every analytics conference did covet. His jolly demeanor as social gathering convener and the lobby bar will do. He’ll love it. Thank you. Welcome, John.

00:01:35.26 [John Lovett]: Oh, Michael, you see, in the event that you and I disagree, be warned that I might just tell you to shove it.

00:01:44.30 [Tim Wilson]: And that is why he is more loved.

00:01:48.27 [John Lovett]: He was ready for you.

00:01:50.13 [Michael Helbling]: That is outstanding. Welcome. We’re glad to have you.

00:01:53.73 [John Lovett]: Thanks. Pleasure to be here.

00:01:55.79 [Michael Helbling]: Let’s kick off and start in on this topic of data governance and start in on, you know, specifically in the realm of digital analytics. There’s some special aspects of data governance that probably are a little unique to our industry. And actually, I’ll throw it to the you three first. What are the first things you think of when you think data governance and how do you start up addressing that problem in an organization? Maybe we’ll be nice and let John start.

00:02:23.24 [John Lovett]: Well, it’s a tough one because data governance means so many different things to so many different people. The way I like to explain it, and I’ll take a little bit of a liberty here being a guest on this podcast, imagine if you will, I’m the leader of our organization. So I’m the CEO and the three of you guys are my three chiefs. So you guys are all VPs. Governance to me is if I make a decision that the three of you are going to be able to reiterate and repeat that decision and communicate it out to the rest of the enterprise. So if we can all get on the same page and act the same way and act according to the same standards, that’s governance.

00:03:03.81 [Michael Helbling]: So you obviously don’t listen to this show because the idea of us all being on the same page and acting as it now, I’m just kidding.

00:03:11.04 [John Lovett]: But in reality, that’s like most organizations, right? You’re not always going to have total harmony and total agreement, which in turn is why you need some protocol for governance.

00:03:21.33 [Michael Helbling]: No, I think that’s a good starting point and actually a good frame of reference. So where do you go from there?

00:03:26.40 [John Lovett]: Well, for me, I do believe that every different type of company, and we’re going to keep our conversation focused here to digital analytics governance, because we can talk about so many different kinds of governance. But let’s stick to the topic at hand here. Whether you’re a big enterprise or a small shop, you need governance. And your culture dictates in certain terms how much governance you need. Now if you are a small organization that’s nimble, you can probably get away with some light governance. Just simply having common definitions, understanding where your data comes from, and understanding the right ways to be able to utilize your data. As you get to become a bigger organization or as you know if you are a big enterprise you need more governance. So specifics about who gets access, how do you manage data quality, how do you ensure the ongoing longevity of that quality. There’s lots of different things that come into play here but to me there’s you know whether it’s a formalized approach or something more casual. I think governance is, you know, and companies probably have governance to some extent, and they may not even know it. They may not call it governance, but they’re governing their data because they have common definitions and they’re using it in the same way across the organization. And again, this is all kind of blue sky stuff because, you know, many companies don’t have these things, but that would be the perfect world as if at least they had common definitions and were able to talk about data in a common way.

00:04:59.17 [Michael Helbling]: No, I think that’s actually a great intro. Jim, up there in Canada, where are you seeing governance happen? Where are you intersecting with that?

00:05:08.26 [Tim Wilson]: Socialist data governance, I think? Is that the form of governance you have?

00:05:12.59 [Michael Helbling]: I think his perspective is unique in that he’s seeing both sides of the border, maybe. Really? I was trying to be nice.

00:05:21.24 [Jim Cain]: In Canada, governance isn’t required because whoever owns the largest seal tusk just gets to be in charge. So we don’t have these problems here. I think I was one of the ones who originally asked for this topic. It’s one of my new favorite things to talk about. And I’ve come at it from a really different level because of the kinds of work that we do, like long-term engagements with big companies. And I’m sure we’re all going to touch on this at some point, but there’s little to no concept of governance of any kind. If you don’t call it governance, you just call it documentation and remembering what the hell you did last month. very, very little of it in the digital practice. And we’ve really come at it from the perspective of what I call, I call governance if shit. If shit happens, how do we have some kind of mechanism to deal with it immediately and not spend 300 hours trying to do root cause? So how do we appropriately document what we’ve done, what things are called, even basic things like let’s all agree on what our marketing campaigns should be called and audit them once a quarter. We’re spending a lot more time on it. And it’s almost like, you know, if shit insurance, What I thought John said that was really neat, and it’s one of the things I’m hoping to learn from you guys on today, is beyond the practical, tactical, again, basic governance that I’m talking about, and we’ve seen a lot of wins in it, but John, you didn’t lead with write down everything that you did and make sure it stays that way. You talked about kind of C-suite alignment, which I think is pretty badass, but I wouldn’t know how to, other than having lots of conversations, build that into a methodology. So I’m looking forward to hearing some more about that. For me, again, it’s more like insurance. I don’t know if that’s a good way to describe it.

00:06:58.35 [Tim Wilson]: I mean, I think that’s definitely, there’s kind of the data, and I’ll say I am the sort of person that I feel like the phrase, it’s like it’s really important, but it also triggers narcolepsy in me. And I wind up when I think about data governance, I’m often thinking of the data collection and data quality, like do we know Where it’s coming? Do we have a current solution design document? Do we have mechanism in place to know that the data is somewhat complete, that every order that gets placed on the website gets logged as an order? And it seems like when the people who really get jazzed about it, Michael and John, It winds up getting, you know, broader than that. And to me, that’s the stuff that the hard stuff is that the data quality and the data completeness because we’ve got multiple systems. And so when you say something like campaign, I had an email late last week where it was paid search agency and we were going back and forth about campaign tagging. It was having. And she said, oh, you know what? I realized we’re talking about two different things when we say campaign. And she was kind of thinking AdWords campaign. And I was thinking campaign in the web analytics tool. And they’re like related and similar. So we got a couple of iterations into the thread before we realized that I thought she was talking about exactly what I was talking about and vice versa. So even those little things when you’ve got exact target and Adobe Analytics, heaven forbid you’ve got Adobe Analytics and Google Analytics and just trying to get agreement on for these core measures, what’s the system of record and what can we and can we not do with the other systems and how are we making sure that the data isn’t breaking and we’ve got it documented more on the data capture front, less so on the data usage front, the reporting. And maybe that’s because I feel like I have a really good system in my world for consistently repeatability for, you know, generating reports or analyses. But I’ll admit, it just feels like it goes all over the place as to what it even is.

00:09:19.00 [Michael Helbling]: Well, and that’s a really great point is mapping a metric into the organization. There’s governance that needs to live around even that process, just because, hey, campaign and campaign, two different definitions. A lot of organizations struggle with definitions even around what’s a customer or what’s a new customer. different departments, different groups have different definitions of what those things are and understanding how, hey, here’s my digital data and here’s how it maps into these definitions. Really important stuff because again, using the data, making decisions with it, interpreting it without that governance kind of puts you in a bad spot sometimes or could put you in a bad spot. I think we’re jumping around some topical areas of data governance and maybe you would

00:10:05.10 [John Lovett]: Let me pull a thread here that I think Tim started and you just picked up on. So Tim talked about the collection and the quality, which are very much analytics implementation type of data governance topics. And then what you just mentioned was what I considered to be presentation. So the two are that’s kind of the other side of the coin, right? So if as an analyst, you’re responsible for data collection and data quality, there’s an inherent responsibility on the business user on how they use that data that a lot of times organizations have no control over. The analytics group has no control over it. Let me give you an example. So let’s say somebody is doing campaign analysis. They built the campaign, they have self-service access to their data, and they go in and they call it out, and they look at the information, and they don’t really misrepresent it, but they look at it in a way that’s going to tell the story that they want it to tell. When in reality that might be, you know, that’s a bad presentation technique because either they were looking at a segment that was you know, so narrow that it just told the story they wanted it to tell, or, you know, missed something entirely, I think that governance around how data is presented is a very important thing that rarely gets addressed.

00:11:22.10 [Michael Helbling]: I completely agree with that and saw that need emerge in practice as an analyst. You know, because you would see different groups dive into your very well curated adobe analytics garden and come out with a bouquet that was ridiculous from the perspective of the analyst. And there’s almost a need to be able to certify the interpretation ahead of presentation in some way so that people don’t misuse that.

00:11:51.72 [Tim Wilson]: I just made metrics into a rose garden. a little bit of stuff that kicked around beforehand. I do think there’s a challenge. When you have four EVARS with different allocation and expirations and they’re all labeled cryptically, shoved into the name as to which ones those are, I mean, I think it’s much, much less common for somebody to say, I’m going to go pull the data and find the story that I want to tell. And much, much more well-intended of, I am in a tool that just lets me just walk right off a cliff and not be aware of it. and pull something that says, hey, this is, you know, I thought this campaign went well and I pulled the numbers. And it seemed like they were the right numbers. And it backed up, it was actually even better than I expected. So there’s no warning bells going off for that person. And definitely I’ve been in cases where you say, that’s why you need, it behooves you if something looks too good to be true. Or if you’re doing anything beyond the basics, you should have somebody who is much deeper on the collection and the quality front review it. But that does introduce a delay. I’ve seen it to the point of, Yeah, we’re having enough of a problem with that that we’re just going to not provide access to the tool at all, which is kind of comical when you think about how much the platform vendors are talking about the democratization of the data. And you’ve got at times analysts who are so sick and tired of getting hauled into the CMO’s office to gently explain that this marketing manager just presented something that was totally false and the analyst gets blamed for it. It gets ugly. But it’s not a simple thing. You say, well, everything’s going to be reviewed by somebody who really understands the data, then your throughput goes down. And so I don’t know the answer. Although, as I want to do, I will point the finger at the tools a little bit.

00:13:47.36 [Michael Helbling]: Ben Gates always appreciates that.

00:13:49.42 [Tim Wilson]: I tried to not list tools by name, but I guess if I say 40 bars with different allocations, maybe that’s a different name.

00:13:57.57 [Jim Cain]: I’m not going to tell you who it is, but it used to rhyme with slomnature.

00:14:01.87 [Tim Wilson]: I mean, it’s a tool with great power. I guess that’s the thing is you don’t walk out. I mean, I will totally legitimate case that you don’t take. I’m not going to go out and drive in the Daytona 500, right? I mean, that’s when you have more, you’re asking for more sophisticated analysis. The data, I think, actually, it doesn’t, if you want to do analysis If you make your data 50% more sophisticated, you’ve somehow exponentially increased the opportunities to misuse or, you know, misinterpret that data. So it’s, it really is not a straight up, you know, blaming the tools that if you have four E bars, that’s because somebody had business requirements that said, we need to be able to look at it our channels this way, this way, this way, or this way. And somebody diligently documented that and implemented it, but then that’s exposed to all the world who has access to that. implementation.

00:14:55.17 [Michael Helbling]: And there’s a part of me though that wonders is this a realm for the tools or is this the proper look you know can the tools really solve this problem and I mean I’ll just throw it out there as a question to us like are there tools that help manage or govern interpretation and delivery or presentation of data and insight you know there’s there’s some things you can do with a tableau to create views and things like that but I mean is that really the solution or is it something else beyond that

00:15:24.00 [Jim Cain]: My thought is that, again, we’re coming into it from maybe a different perspective, but good governance is like good project management. There’s project management software out there, and there are some tools that we use that assist in governance, and some of the data capture vendors have components that make a governance program smoother, but if you don’t have someone who knows what the heck they’re doing, who’s moving everything forward and achieving consensus and making sure things are done at a particular period of time, then all you’re doing is buying more shit because four lines of code will fix your problems.

00:15:57.73 [Michael Helbling]: And you know, we’re all not saying people process and technology is kind of at the center of all this, but we all probably agree that’s true. But there are vendors out there that help with the problem of data quality from a collection and completeness perspective. The auditing systems like ObservePoint and HubScan.

00:16:16.84 [Tim Wilson]: I do think there’s an opportunity for the tools to Google almost as it accidentally because a lot of their if you if you’re capturing advanced data you can’t get to it easily without writing it you know setting up a custom report so you’re kind of in a sense you kind of have to there’s a little bit of a wall from you can’t It’s harder to wander into the trouble spot because you have to actually know, oh, how do I go make a custom report? But I definitely, we’ve had threads actually, there was a thread a while back and I think Eric Matisoff might be the one who put it into the, what is it, idea exchange, that the actual true documentation of implementations is never built into any of the web analytics platforms. Why can’t I get my SDR literally the three or four things that I need to store. All that’s kind of stored is what the actual settings were, but I don’t get the definition. I don’t get any hierarchy of what’s it used for. I don’t get the ability to say, hide this for anybody who doesn’t at least, isn’t at least kind of an advanced user. So that’s, I realize that’s heading down into the weeds into one very kind of narrow specific component, I guess.

00:17:28.54 [John Lovett]: Yeah, my take on it is that I don’t think that the analytics vendors should be responsible for governance. I mean, we’re talking about things that they could do to help with that. So, you know, putting those user access or hiding data from some users, you know, those are kind of feature switches that you should be able to flip to protect people. But I don’t think the vendors should be responsible. they can aid, you know, what Adobe tried to do with their anomaly detection is an example of, oh, how can we start to govern this so that people will pick up on things more quickly? But again, it comes down to the individual.

00:18:02.23 [Tim Wilson]: That’s going to get edited out. We can’t head down that disaster, the way that feature actually works.

00:18:09.90 [Jim Cain]: I think what John was trying to say is that it’s not about technology, it’s about people, like what I said. before Michael disagreed with me. So John, I think you’re going down exactly the right road.

00:18:21.36 [Tim Wilson]: If somebody’s writing JavaScript, you can put comments in JavaScript. So I am going to stop. It’s kind of like the vendors aren’t responsible for privacy. No, they’re not responsible for actually putting it in, but when they’re literally things that make the most sense to have a home in concert and companies have to go and kind of invent how they’re going to link the things together with kind of a, I mean absolutely it’s not a, to Jim’s point, it’s like project management. You can give somebody a great project management tool and if they’re not good project managers they’re still going to kind of screw it up. So I’m not saying it’s up to the vendors to solve governance, but for the vendors to actually say, given the complexity of our system, this is kind of a natural way to do governance and manage a spreadsheet and hope that it’s sitting in your file share, your network share somewhere. that somebody is maintaining it when changes are made. And we’re going to have log access systems that are damn near impossible to use unless you’re in there every day. That is actually hindering the people who are saying, I can lay out a governance program.

00:19:32.91 [Michael Helbling]: I’m gonna sound like such a… Let me just clarify a couple things here, Jim Cain. First, I don’t disagree that the way that I’ve seen governance work, especially around interpretation of data, has been focused on people. But I further wonder how do we incorporate technology at that point in the process? And the reason I wonder that is because could that be a bridge to not having to restart this process as you lose those people? So like, do we have to just come to grips with the term like, Every three years, you’re going to need to restart this whole thing because that’s what’s happening across the analytics organizations is they’re turning through and people are leaving and the person who owns this process.

00:20:13.06 [John Lovett]: That’s exactly right. That’s why it’s a people thing, right? I mean, you need to continually train your people on what governance or how you’re governing your data and how data should be utilized.

00:20:24.76 [Michael Helbling]: And I don’t disagree with that, but I’m saying what is the opportunity for technology to help with that?

00:20:31.21 [Tim Wilson]: On the presentation side, and this is something I preach all the time, which is definitely, so this is a training of analysts or anybody pulling the data, because it’s been me and the ass enough earlier in my career that when I came back a month later, two months later, and somebody said, remember that analysis you did? Can you go and do it again, or can you dig into it? And I didn’t embed the documentation. It’s easy. If you’re pulling data in an automated fashion, whether it’s from Google Analytics and one of the platforms that’s middleware there, or if it’s Adobe Analytics and it’s Report Builder, it’s pretty easy to just get in the habit of saying I’m going to record, I’m not just going to do a data export, I’m going to make sure I include what segments, what dimensions, what time frame, and I’m going to kind of build that into my process. And to be fair, a lot of analysts don’t do that. Like I’ve had that happen, an analyst has done something, and the first thing I want to do is replicate it, and I can’t. And then I ask them, how did you get this number? And I’m like, gee, I don’t know. I’m like, well, why the hell did you not use three extra cells on this Excel spreadsheet in the background to record that so it could be replicated? So I guess that gets to the governance on the presentation front. What’s your best practice that when you pull data, as soon as you transform it, you better say what your source was and how you got it and what you did to it. So maybe I’m starting to understand that part of data governance.

00:21:54.10 [Jim Cain]: Can I ask a question to you guys? I saw a year and a half ago, John, the presentation where you talked about governance. I really quite liked it. It was a little bit more of a kiss principle approach. I’ve seen several presentations on governance over my career that frankly scared the shit out of me. It looks like the Manhattan Project. You just see these gigantic death by visio approaches to these committees feeding these things. And really the takeaway to me at that time was this isn’t something my organization is capable of being a part of because it’s just too complicated. And then what’s interesting today is everybody’s giving kind of tactical, practical. If analysts think to do these things, that’s governance too. So when you’re approached or everybody, because we’ve all talked about it with customers, what are the best practices for, I’m an organization that’s culturally not there yet, want to get started?

00:22:49.94 [John Lovett]: Well, I think it comes into, I think you, for lack of a better way to say it, you sneak it in. And by that I mean a lot of governance is process and workflow. So how are we going to add new tags for your feature? How are we going to provide the ability to access that data and report out on it? how we’re going to protect you from yourself by not giving you access to some certain some of the data that you may not be able to utilize. I think you you don’t necessarily say here’s our governance you say you know this is the workflow these are the things that you know here’s our intake process for analytics requests and what you’re going to get is if you’re not a super user you’re going to get a pre-formatted report that’s going to and here’s what this report is going to tell you. So a little bit of hand-holding to get them to the point of accepting data and being able to understand what data they have without putting a bunch of restrictions around it. You’re kind of spoon feeding them, if you will.

00:23:56.41 [Michael Helbling]: In our practice, we’ve seen tag management be this catalyst for data governance in so many situations, because it first exposes you to, I didn’t even realize those tags were on my website. They’re buried inside of an old double-click tagoon. I was on a call the other day and I was like, hey, do you know you’ve got Yahoo Analytics still on here? They’re like, no, we don’t. I was like, well, yeah, you kind of do. Ghostry says you do. And so we dug in and we found out where it was coming from. But so that’s the first thing. And the second thing is, is once you start running it all through one place, you’ve now given yourself one platform to own and manage. And all processes can flow through that one place inside of that tag management tool. And so that forces you to be like, whoa, let’s back up a step and talk about who’s allowed to put tags on our website? Who’s allowed to update our analytics tagging? when should we, now you’re going down a good path? Sometimes that’s the entry point, but the other place where you see it is when you start going good and you do some insights that actually are making them business money, somebody at some point is gonna be like, how do I protect the golden goose here from things like lost data or somebody else coming along and really messing this up? And that’s when governance comes in and says, okay, how do I create a system we can keep going on this path? So that’s the kind of two entry points I’ve seen.

00:25:15.89 [John Lovett]: So I want to ask a question related to that. And curious to know opinions of this group, I’ve got some of my own. But the idea of a chief data officer, or more recently I work with an organization, they had a data czar. Is there an individual that can be accountable for stewarding that data as a whole for an organization?

00:25:35.79 [Tim Wilson]: They are, but I’m going to jump on this one. What’s interesting, because a part of this is heading on the path of putting the data Governance as a core process or what I have seen at large organizations and then what I’ve also seen I guess when analytics is a little more on the periphery is that a big part of Data governance or the person who owns it is actually going and understanding the existing processes and saying what we’re going to do is take the core processes if we have that have nothing to do we’re not built for Data, but it’s our web development process and And anytime something new goes into production, we need to add one more step in that process to say, have you pinged the analyst? Have you done whatever? There may be, are there data ramifications? And so you don’t wind up, there’s a lot of data governance to me that is not in the data. What are we doing on whatever schedule for data governance? It’s what are the processes that are currently feeding the data, assessing those processes, which of these are most likely to change, which ones are most likely to break data, and then how are we affecting change on those processes, which are likely owned by somebody else entirely. They’re owned by the PMO, they’re owned by IT, so that if you have that person, and I worked at a large insurance company, and it was around customer data, But they actually said, OK, this guy is going to do customer data, data governance as part of his job. And it took him about three months of digging in and understanding what it was. And he went and made a business case and said, not only is this a full-time job, I need a team of two people because I have to touch all of these other processes. And I can’t touch them. I have to go and negotiate and cajole and check in on and make sure those processes have been And that was a very legitimate and necessary role with the size of the organization and the criticality of the data. But he actually owned some documentation. He owned very little end-in processes. He was integrating with other processes.

00:27:38.55 [Jim Cain]: I was just going to say, I think that the role of chief data officer is an important role. I think that I wouldn’t be surprised if in 10 years there were a lot of them and there’s like chief data officer magazine or whatever, but I have checked the dot com on that one. Wait, the swimsuit issue on that baby is going to be… The Tim Wilson centerfold was pretty weird last month. But I think it’s a very aspirational gig. I think the kind of organization that could take advantage of that person right now would be hiring a senior vice president of hope and faith. It’s an aspirational, I’d like this goal. Tim, did you like that? You want to steal that for your next presentation? You can have that.

00:28:20.35 [Michael Helbling]: Yeah, I’m not very far away from where you’re landing, Jim, which is I agree that this conceptually is really important. But what I’ve noticed is that when those roles exist and we’ve interacted with them in our practice, typically digital data has somehow just skipped right over that office. And they don’t understand it. They’re not involved with it. It’s not a thing that they’re doing stuff with. And to the extent they are, they’re more a hindrance than a help sometimes. That being said, the path leads to that. I feel like each organization has to organically get to that point, and so many of them are still at this fundamental step of, I have analytics, but I don’t believe anything it tells me, because it’s all wrong. And it’s a sad statement that that’s where 90% of our industry still is today, but that’s kind of where, and that’s we’re not gonna have chief data officers until

00:29:16.07 [John Lovett]: Well, they’re going to hate their jobs if we don’t have the case.

00:29:19.28 [Michael Helbling]: Basically, their job is to run the audit scan to make sure all the tags and evas are in place. That’s not a role of a chief data officer. I think a chief data officer has to own the other aspects of governance as well across the organization, and to the extent that there’s not value coming from those, boards and CEOs and executive suites are not going to position someone in that role until there’s actual value or value protection coming out of that in a very big way.

00:29:45.77 [Tim Wilson]: So the people I’ve seen who bond up in not necessarily a C level title, but have had that sort of title, the pitfalls, the pitfalls are kind of twofold and they’re related. One is that the role of actually prioritizing the data doesn’t happen. I have a spreadsheet with a list of variables. I have a spreadsheet with a list of data sources. I have a spreadsheet with a list of data elements. I can’t remember. There was a case where somebody asked for, you know, they wanted a mapping and a description of every data element coming out of You know some system and we’re like well 25% of these are like crap and random like like we can tell you which ones matter and which ones don’t and they were kind of going through wanting to have their complete catalog, so they’re trying to eat the whole elephant in one bite. And then related to that, if you put somebody in that role and you’ve got the executive saying, my expectation is that any chink in the data is that person’s responsibility, they have an expectation that you can kind of eat the whole elephant once as well, as opposed to taking any sort of an incremental approach and saying, this is the data that is gold and matters the most. We must capture orders, and we need to be measuring Multiple times a day, we need to be confirming that orders are being accurately captured in our system. Social sharing clicks that happen 12 times a month on random obscure pages on our site, we don’t need to check that multiple times a day. We don’t need to be over monitoring it. But I’ve watched that happen where somebody says, I’m going to put monitoring in place, or I’m going to put processes in place, and I’m going to cover everything. And you’re like, wow, it took exactly as much time and investment and ongoing investment to do something that almost never matters as it took to measure something that’s super critical and you wound up with in the average. So that’s a people and a judgment of not taking a simplistic approach and putting some nuance into what about this matters the most and let’s put some priorities in there. I feel like I keep ending with rants that you guys are like, shut the hell up. I’m sorry, I wasn’t listening.

00:31:56.61 [Michael Helbling]: I mean, it’s really a problem that hits organizations at many different levels, because there’s a technology level, there’s a business process level, and there’s even sort of an action or, I don’t know, an action level. I don’t know. So as analysts, we kind of are struggling to kind of like grab both ends of this and stick it together. Are we even the right people to make this solve?

00:32:20.80 [Jim Cain]: And I think I want to ask John a question because it might put the rest of it in perspective for me and it’s kind of a two-parter. But the concept of governance or process governance or whatever you want to call it for anything at the enterprise, I’m guessing is not a five-year-old concept. It’s been around for decades, right? Of course. How many companies are actually doing it properly for like anything?

00:32:40.57 [John Lovett]: I would say it’s probably a very low percentage.

00:32:43.03 [Jim Cain]: So, you know, we’re talking about the EVP of hope and faith, right? We should be doing it. And I think the cool thing about digital is it’s… Well, wait a minute, wait a minute.

00:32:53.43 [John Lovett]: Let me retract that statement for one second. So when we talk about governance for financial accounting… As you say, there’s one case. You know, when we talk about digital analytics, that’s where it gets much more slippery. That’s the big distinction there. Yes, there’s been governance forever, and I’m not as close to it, but maybe the governance over business intelligence is much stronger than what we have because they say, okay, here’s the data that you have access to. You can cut it in this way or this way or whatever way you deem fit, but it has to pass through the data quality checks before it gets distributed to the organization. And web analytics, partly because we’re still maturing as an organization or as an industry, hey, go ahead, grab Google Analytics data and do whatever the hell you want with it. That’s sort of the wild west where our industry hasn’t had that governance. But I do think it exists in enterprises. I mean, there’s lots of big companies out there that will help you do, help an enterprise do data governance in the big sense of the data world. But us in the analytics world, I think it’s still a little bit of a maverick society where you can kind of get away with shooting from the hip and doing whatever you want with data, which is wrong. But that’s the reality of what happened.

00:34:11.99 [Michael Helbling]: Yee-haw, right? Because here we all are. But in 25 years, maybe there’s going to be generally accepted analytics rules, like there are for, you know, other… Like FASB? Yeah, exactly.

00:34:23.44 [Tim Wilson]: We’ll have our own little Starbucks actually for… GAAP.

00:34:27.15 [Jim Cain]: Yeah, generally accepted analytics rules. I like to help in Wilson principles. It’s got a very patrician ring to it.

00:34:34.19 [John Lovett]: But let me just touch on something that we addressed earlier. Maybe tag management is a way to oversee that in some ways because since they have a perspective on tags on pages, maybe that would be a good check to say, OK, well, here’s the standards by which we may manage our governance. Because I don’t think there’s too many different technologies at play. You can’t say that you’re Adobe. You’re Google’s in charge of governance. Because what about all the other tools you’re using? I mean, you guys talk about tools all the time on this podcast. So we know that it’s not a singular tool world. It’s the Scott Brinker slide where there’s eight or nine hundred different tools that we’re talking about.

00:35:18.53 [Michael Helbling]: Yeah, that’s a good point.

00:35:21.70 [Tim Wilson]: If you take all those tools that are generating and collecting data, if you took tag management and you actually, if that converged with the auditing tools just to cover your tags, all those technologies that can be detected on a site or in the wild, you could see having a process in place where you hooked them up, so we’re going to track We’re going to make sure we’re logging when somebody’s explicitly making a change, but we’re also going to monitor for variations in that tag, and we’re going to kind of link those. together, but somebody still has to actually set that up and manage that process.

00:35:56.17 [Michael Helbling]: Well, this has been a good conversation. So let’s maybe go around round Robin, maybe do a wrap up, maybe throw out an idea. Hey, if I’m thinking about governance for the first time, like what can I do to take a proactive step for my organization? What’s a good first thing for me to do?

00:36:14.20 [Tim Wilson]: So I’ll take one because I kind of wanted actually on the presentation of the data governance and we sort of danced around this, but I did this years ago at an organization and it worked. really, really well, which was we actually developed for the analyst organization. We were touching CRM data, web analytics data, stuff in the warehouse, even some ERP data. And we had an organization where anybody could get in and access data. And so what we did was said, we are going to kill two birds with one stone. We’re going to come up with a style guide for what we produce from a presentation and visualization perspective. And by doing that, it’s going to be very clear to the organization. Our stuff will look better and we’ll learn best practices for presentation and communication. But the real reason, and we had this explicitly stated, was that we want to have the stuff that’s being produced from the BI organization that this stuff has a higher level of scrutiny because we have internal reviews. We’re more knowledgeable about the data. So if somebody sees a default Excel chart that shows something that some product marketing manager is showing, It’s going to be kind of implicit and clear to everyone that that didn’t come through kind of the seal of approval process, and it will encourage people to run their stuff by us or use our resources to kind of check their work because they knew it would kind of come out. It would look better, would be kind of their motivation, but the fact is it would be clear and have more credibility in the organization because it had been kind of vetted. by us. So it was a way that we kind of slip in a highly desirable workflow without trying to enforce a workflow or a lockdown access to the data. That’s all I got.

00:38:02.72 [Jim Cain]: That was a gooder. So we’re not doing a wrap up today, Michael. We’re just sharing our like, get your toes in the governance water.

00:38:09.65 [Michael Helbling]: No, I wanted to try to find a way for us to give somebody a leg up on this. You know, as people listen, maybe they want to actually do something about what we say. I don’t know.

00:38:22.65 [John Lovett]: I got one. So I’d answer your question about what, you know, if somebody’s thinking about governance and how do they start, what do they do? To me, it breaks down into two different buckets. And you need to ask yourself, where is the bigger issue for your company, the company that you work with? The first one you need to look at is, do we have a data quality problem? Do we have problem with the way that we’re collecting data, the way that we’re configured, the way that our tools collect the data? Is that our issue? Or on the other side, the other bucket would be, do we have a presentation problem? Do our business users misinterpret the data or mispeak about the data or simply out of ignorance not know how to utilize the data? And once you can kind of pivot on Is it a quality issue or is it a presentation issue? That will tell you what to do. If it’s a quality issue, you need to start auditing your data. It’s a technical problem. Get out there, get one of the tools, audit your data, look at your solution, look at the way that you’re implemented, make some changes to build back the confidence in your data so that you don’t have underlying data problems. because that in turn will lead to presentation issues as well. But look at that first. If it’s quality, you need to start auditing your data. If you have a presentation problem, if that’s your burning issue or your burning problem, then you need to educate. You need to train your business users what the data means, how to use the data, what are the guardrails in which they can use information and data that’s being collected really I think that’s the path forward is to you know which one is it which is your biggest problem and then the solution is there in front of you what to do about it.

00:39:59.91 [Jim Cain]: I think whenever we get into an organization that needs to start going down the path to even some basic governance we do one of three things first. The first one I mentioned earlier is to literally get out a spreadsheet and create a common naming convention set for all marketing campaigns whether it’s Adobe or Google or Coremetrics or what have you. All the agencies, all the marketing users always have to call the same things the same ways and it’s really easy. I mean it’s a lot of phone calls and dirty work but it in a lot of cases helps to create alignment for the first time on anything and it could be a quick and dirty win. The second one when you guys talked earlier about Tag management systems, I actually don’t think that they’re as valuable from a technology perspective as they are from a workflow perspective for getting organizations into governance because that first time you go to a business user and do appropriate requirement solicitation and you properly translate it into IT and then you manage that workflow. In a lot of digital marketing organizations, that’s the first time anybody’s ever done that. And it’s, again, a very nice way to start getting people down the path to thinking about proper structure, proper governance. Third one that I would mention is, again, quick and dirty, but do some data normalization between the primary kind of goal completion metrics in the digital analytics tool and whatever the actual business tool of record is, because it’s never the analytics tool. So if you’re e-commerce, what does the CFO say revenue is in transactions are by day from digital? Are they close? And it’s a very, again, a good way to get people talking about why these things are important. They’re kind of quick and dirty.

00:41:32.68 [Michael Helbling]: So yeah, no, these are great wrap-ups and great commentary on the topic at large. I find it fascinating, and I guess I shouldn’t at this point, given that we always seem to have leftovers at the end of these discussions, but we didn’t even touch on the concept of data collection and the kinds of data collected in terms of privacy and things like that.

00:41:54.60 [John Lovett]: I was going to say we can’t close without mentioning privacy at least.

00:41:58.71 [Michael Helbling]: It’s another aspect of governance that is probably a part and parcel of this whole discussion. We just didn’t have the time. I loved the conversation and I think I’m not actually going to throw anything else on top of that because I think some of the takeaways and first steps were ones that I would heartily agree with from all parties. That being said, I will say, Jim Cain, that one aspect of tag management that you get from a technology perspective is data efficiency or quickness data, which from a governance perspective can help you to put new analysis or measurement capability in place.

00:42:31.87 [Jim Cain]: You’re totally right. It’s just as you violently agreed with me before, people trump technology programs like this. I think the points you made were dead on, Michael.

00:42:40.82 [Tim Wilson]: I mean if you can imagine a case if you’re not careful you’d have an analyst say a enterprising consultant analyst who says sure I can track that button for you and they can hop right into their tag management tool and add a deployment and click click enable and commit and publish and then get the call from. The actual guy is supposed to be working in tag management saying, you know, I had six other things committed. And when you hit publish, you just pushed a bunch of stuff out there a little prematurely. I’m scrambling to fix those now. That might be a personal anecdote.

00:43:10.98 [Michael Helbling]: Maybe if your tool allowed selective publishing, you would run into that prior.

00:43:16.13 [Tim Wilson]: No, maybe if the analysts actually, you know, these are all the things that are going to get pushed out there.

00:43:22.34 [Michael Helbling]: But this loops right back to the very beginning of the conversation where John was talking about, hey, if you’re an organization with a lighter need, that analyst basically is of the governance, right? Their ability to maintain the analytics stack, that’s the governance, quote unquote, for that organization, whereas in a more refined or structured environment or bigger multiple teams are working on this, we need a better structure, not just one guy sitting there with the keys to the whole thing. Anyway, great conversation. There’s so much more to add and that’s where you all come in. We’d love to hear your comments and questions. either on Facebook, or on Twitter, or on the Measure Slack. We love hearing from you. Once again, a huge thank you to John Lovett for being our guest on this show. It’s a rare opportunity to get your insight and wisdom on this topic, John, so thank you. Thanks for having me. Obviously, for Tim Wilson and Jim Cain, I am Michael Helblink, and thanks for listening.

00:44:21.86 [Jim Cain]: See you next time. And thanks for not bringing up that thing I said to John at the bar a couple of years ago.

00:44:31.49 [Announcer]: Thanks for listening and don’t forget to join the conversation on Facebook or Twitter. We welcome your comments and questions, facebook.com forward slash analytics hour or at analytics hour on Twitter.

One Response

  1. […] The Data Governance Episode of the Digital Analytics Power Hour (Episode #012) […]

Leave a Reply



This site uses Akismet to reduce spam. Learn how your comment data is processed.

Have an Idea for an Upcoming Episode?

Recent Episodes

#274: Real Talk About Synthetic Data with Winston Li

#274: Real Talk About Synthetic Data with Winston Li

https://media.blubrry.com/the_digital_analytics_power/traffic.libsyn.com/analyticshour/APH_-_Episode_274_-_Real_Talk_About_Synthetic_Data_with_Winston_Li.mp3Podcast: Download | EmbedSubscribe: RSSTweetShareShareEmail0 Shares