Deep Dive Episode: Governance of Marketing Technology and AI

 
 
00:00 /
 
1X
 

In this episode of In-Ear Insights, listen as CEO Katie Robbert and Christopher Penn discuss the process of aligning marketing technology and artificial intelligence initiatives with best practices for governance and compliance as part of the run-up to the IBM Win with AI Summit. (register here for the summit in NYC!)

One of the reasons why marketing technology and artificial intelligence feel like the wild west is that very little has been done in the CMO’s office to implement proper governance practices. By contrast, IT has built decades of processes and procedures for governance; as a cost center to most organizations, IT brings rigor and discipline to the technology stack. Marketing has much to learn from IT on this front.

We review the BrainTrust Insights Marketing Technology Governance Framework, based on the well-proven ISO/IEC 38500 Standard (2015 edition) and how companies should use it to bring more discipline to their marketing technology and artificial intelligence processes, platforms, and teams.

The BrainTrust Insights MarTech / AI Governance Framework (ISO 38500-2015)

Grab a full-size PDF version of this framework from our Instant Insights Resource Center.

FTC Disclosure: BrainTrust Insights is a registered IBM Business Partner. We receive financial compensation for business referred to IBM.

Machine-Generated Transcript

What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for listening to the episode.

Christopher Penn
Alright, today we are chatting about p deployment and the operational ization of artificial intelligence. This is for IBM is when with AI summit, which I will be attending in September in New York City. So we’ll have a link in the show notes if you want to register. And I believe attendance is free. And, and IBM might even feed you something. So

in terms of governance stuff, one of the best frameworks out there for AI, I think I love your perspective on this AI really is in many ways, it’s just yet another form of it from an operational perspective. And so you want to talk about your experiences in IT governance, and how you know what things tend to go right and wrong. Really, really obviously, from the start.

Katie Robbert
Yeah, absolutely. Um, you know, one of the things that tends to go really wrong from the start is a lack of planning. And, you know, planning is the unfun, the unsexy part of any project, you know, where you have to sit down and do some documentation of what am I even trying to do, because people in my experience tend to get, like, so amped up and excited, which is great that they just want to start, you know, coding things and plugging things in and testing things out,

right, which is all well and good, like experimentation. The innovation has its place in time. But it can’t necessarily be used for production. Now, that’s not a hard and fast rule. But when it comes to it, you really want to be kind of rigid about it. Because there are so many things that could go wrong.

I used to work at a company where our IT department was responsible for servers that held protected health information. And so there was a lot of rules and regulations. And we would be audited at least yearly by a pharmaceutical company or by the FDA. And, you know, you had to have all of your disaster recovery and your business continuity and all of those plans in place. And, you know, if you’re a small company like us, those aren’t as important. But you need to have some semblance of if the power goes out tomorrow, and doesn’t come back on for a week, what do you do, because that’s the problem. So I think that the thing that goes wrong 10 times out of 10 is a lack of planning.

Christopher Penn
So in terms of planning, then one of the frameworks we should probably spend some time with is, well, everyone and say, I think with when people think about artificial intelligence, they think about this whole new frontier of computing and things. And there’s some validity to that in terms of conceptual application. But when we talk about the governance of it, it’s probably a better idea to not reinvent the wheel. And look at some of the existing governance frameworks for it. One of which, this, this whole bunch, there’s Corbett, there’s, there’s an I triple E standard, there’s an IEEE standard. And the one that I’ve looked at in the past is the ISO standard, the ISO, it’s ISO 38, 500

dash 2015, the most recent one, and this one basically says, There’s six things you need to look at. When you’re when you’re talking about governance, there is overall corporate responsibility, there is your overall IT strategy, there is the acquisition and management of assets, the performance of it, the conformance of it to rules and regulations, and then managing the human it interface to humanity. And in the framework, which will put on the brain trust insights website, under the resources section, you can actually download a copy of this will put a link in the show notes,

there’s the concept than this function, application and measurement. And so when you’re talking about planning, where does planning fall into into the broad framework? Is

Katie Robbert
that at a business strategy level? First should be Yeah, because the first question you should be asking or being able to answer is, why am I doing this? And so how does it fit into your business strategy? What’s the problem that you’re trying to solve by adopting something like AI, or updating your it or whatever the thing is, so it’s really, you know, your, your overall business strategy, it needs to run in parallel or fit nicely into it. And if it doesn’t, then you need to, you know, take a step back and figure out what am I even trying to do. And so one of the things that you know, is difficult to pin down, but is an absolute must, is one key decision maker. So, who at the end of the day is the one who has to flip the switch, who has to push the button. And it’s not necessarily the most fun job, because when things go, right, you’re the hero when things go wrong. You’re the first neck to choke, which is it’s, but it’s true, though, and I’ve been on both sides of it. And it’s not a fun position to have. But it’s one of the most and because at the end of the day, you can be getting fed all of the information from all of the different silos and stakeholders and everything. But ultimately, one person has to make a decision, you know, if you want to do decision by committee, that’s fine. I personally don’t like decision by committee, I find it gets very messy, and it tends to hold things up. But ultimately, your business strategy is the place where you start. And it has to include a key decision maker or someone who ultimately is responsible for this thing. And it needs to answer the question of how does it align with what your business is trying to what your business goals are? So that’s where you start? How does that does that fit really under strategy? Or is that also part of the human change management process of getting people to adopt something? It’s absolutely, so it’s absolutely part of change management. So when when you sort of talk about these buckets of where to start, it’s when you actually start to dig into it, it’s not as clean cut as the way do you have it outline because there’s going to be a lot of overlap, there’s going to be a lot of redundancies and you can’t necessarily do each piece in isolation. So you can’t do the business strategy and then figure out how to do change management, they have to be done together. Because most companies, 99% of companies still have humans working for them. So there has to be that level of, you know, how does that work? How does change management filter down. So it’s all sort of as this is going to be all try to articulate those. So basically, if you start with concept, and that sort of that top left, and down in the bottom right is pm, capacity capability, Maturity Model integration, if you think of all of these little boxes as dominoes, they all have to start to fall for us together to use, you end with the execution, okay, and so you can’t just do business strategy, put that away, then think about the martek strategy and then put that away and then think about the balance sheet, it all has to be done in parallel together, because it is all at the end of the day, it’s all one integrated government’s plan. Well, they’re just different pieces of the same plan.

Christopher Penn
Right. Okay. So that, does that mean that something like is accurate framework is better for this, then then, sort of the discrete things where the Zach, when framework is who, what, where, when, why, how, across the top, and then every single department on the left side of like a spreadsheet?

Katie Robbert
I don’t think it necessarily, at least in my opinion, it doesn’t matter which framework you’re using, if you’re using that one, if you’re using this one, because ultimately, the result should be the same. And the method the method of getting to the answer of what it is that you’re doing should include that who, what, why, when, how. So if you’re looking at this framework, for example, the martek a governance framework, those questions should be baked into each of these steps, okay,

Christopher Penn
for how important is this? Or how do you scale this down? So obviously, this framework, ISO 38, 500 is really intended for the enterprise especially where you have very strict compliance measures to hit you have maybe you have pH I in the mix or you are PII, certainly you are publicly traded companies, there’s sec compliance, how do you scale this down to more of like a mid sized business where maybe you’re privately held, maybe you’re not working with anything sensitive, but you still wanted to benefit from AI, and, and, and not spend enormous amounts of time and money on on doing something that might not work. So I

Katie Robbert
think you start with this, I think you start with the enterprise framework, and then you basically you can sort of x out the boxes that don’t apply to you. But make a note that says why they don’t apply so that if you do you know, grow from a midsize to an enterprise company, you can say, okay, at the time, this didn’t apply. So we don’t, we didn’t necessarily have patents at the time, or, you know, risk and compliance wouldn’t apply, because we’re not kept capturing any PII information,

you know, so just acknowledging that it doesn’t apply right now, but it could in the future. So I think it’s not necessarily getting rid of any of these things. But it’s just sort of saying, Okay, I’m gonna push this aside, because the thing that’s most important to us right now is the business strategy, the balance sheet, and how we’re going to implement these things

Christopher Penn
in your past IT governance work has much come up about the it balance sheet, in terms of sort of the data as a financial asset, no, really surprising given that you worked in healthcare where the data was literally what you were selling?

Katie Robbert
Yeah, well, it’s it was never afraid that way. So the data was what we were selling the de identified stripped out of any personal health information data, but it never, it was always run through the academia, ology department. It It was, it was an interesting structure, it was very much left out of the loop. And it was a constant source of frustration for it, because they were the ones managing the data, but the team at the time was really trying to keep it separate. And, you know, it all kind of came back to that fear of automation. And that fear of AI, because if they made it any easier than there was that concerned that their jobs wouldn’t be as valuable, or that they would be redundancy with the human side, and they would have to cut down the size of their team. So I think that that’s what happens a lot

Christopher Penn
is that, you know, people don’t get it involved as much as they should, even though it can make your life a lot easier, because there’s that concerned about, well, if I automate this, or if I make it more expedited, then I’m personally not going to be needed anymore. And that’s just not true. Okay, how do you value the data, because one of the things that in at least in on the artificial intelligence front is data is the training material. So in your average AI system, you have a training data set your test data set, and the training data set is the the known good stuff. And in our, our two by two matrix is the known knowns and you’re trying to use that to explore the known unknowns and the unknown knowns,

how do you value that? How do you put a, a, a financial

because when you if and when you are, you go to sell your company, those assets that are on your balance sheet should factor into the the value of your company, I know, certainly at the last company we worked at when they sold one of the selling points that allowed the the company owner to as they put it, get a premium for the company was the capabilities, but I don’t think they ever factored into the actual intellectual property?

Katie Robbert
Well, I think like, like any good question about data, it depends, I think it depends on the type of data it isn’t pens on how extensive the data set is, how complete it is, how clean it is, and really sort of what questions you can answer with it. So, you know, in the example of a previous company that I worked at, where we were collecting data about use and abuse of opiates, directly from people in recovery, that’s very valuable data to a pharmaceutical company who’s trying to understand have they created and abused determine formulation. So that in that example, is a lot more valuable than if somebody hit my website through Google Analytics. You know, it’s not to say it’s not important, but I think it depends on the question that the data is trying to answer. But to your point, it absolutely should be factored in on the balance sheet as an asset of value that’s unique to your company. Because you’re collecting it, you’re you’re putting the puzzle pieces together, you’re you’re turning it into an hour, if you’re trying to get into something actionable

Christopher Penn
stuff words in your mouth, and say, That’s not the data itself, I think it’s the data as a service has been something people bouncing around for a while. So it same for infrastructure as a service, or the raw foundational materials. And there are companies that want to sell you those things. But it almost sounds like the reality it’s more model as a service or model as the product, you have the data, but it’s your interpretation of the data, and it’s your algorithms around it, particularly, once we start talking about AI with that’s the value is the value is you’ve built a pre trained model, and that is constantly self tuning. But because of your unique combination of inputs, meaning your raw data, and your data scientists, presumably working behind the scenes, you’ll come up with a functional model that says, This is how we look at the world and how we think the world functions. And if you believe in, in, in the company that and and the product we deliver, then this is the benefit you’re buying. Does that make sense? Yeah,

Katie Robbert
absolutely does. I think that that’s one direction that it can go in. And I think the other direction is that, you know, if, you know, you get acquired, or if you sell the company, then it’s the data set itself that has some value, and then, you know, a company coming in and can do whatever they want to well, with the data and say, we’re going to develop our own algorithm. So at the core of it, you still have to have that data collected in a database somewhere. And that that is the value because in your example, the data scientists are valuable, the algorithms are valuable, but without a data set it there’s nothing to do, right? Well, it

Christopher Penn
kind of goes back to what you’re saying originally about the governance framework overall, which is no one pieces independent, you have to do all the things. So you have the pieces of that. So if you’re building data, as an asset as a thing that you want to sell, and then there’s, you know, this is part of governance, what are some of the critical things that you need to worry about when it comes to that, that data in order to prevent its value from decreasing and ideally, increase its value.

Unknown
So I think that goes back into,

Katie Robbert
you know, the business strategy of what it is you’re trying to do with it. But then that’s when you start to get into things such as risk and compliance. And that’s when you get into the architecture and the ownership because you need to be able to plan out how you’re going to structure these databases, how you’re going to protect the data, what you’ll do if there’s some sort of a disaster, and you can’t collect data for a period of time, that was something that we struggled with, where if everything, you know, we played out this scenario of, let’s say that there’s a natural disaster and power goes out for three weeks, well, we just lost three weeks worth of data, if not more, try to get the systems back up and running, what do we do in that sense? And so it’s being aware that those things can happen and what to do to either prevent against it, or have a contingency plan. If data is the product that you’re collecting and selling? What do you think about the impact of cloud and cloud computing on on things like that, um, you know, I think that it’s made it easier to collect data. But I also think that when I remember when cloud computing came about, it was sort of this buzzword and everybody kept saying, well, it’ll be in the cloud, it’ll, it’ll be in the cloud. And I saw so many PowerPoint presentations with literal pictures of clouds, because it was just such a misunderstood thing of what it meant, it’s like, well, I don’t know, the data went into the cloud, I think that it became a little bit of a headache for IT departments, because it wasn’t necessarily something that they were hands on managing anymore, they were relying on a third party company to manage this for them, and they sort of had to architect it out and just hope that they could get access to it. Um, I think that, you know, it really changed the rules of risk and compliance, it changed the way that people had to think about protecting their data,

you know, people were used to physical servers. And, you know, there was a i, there are still times where physical servers can’t cross, you know, country boundaries. So, we couldn’t build, we couldn’t have a customer in Canada, because they required a physical server there. And the cost to build a physical server was too much money, but they also wouldn’t allow us to host the data in in a cloud instance, and pass it back. So there’s a lot of things that people don’t think about in terms of security and compliance. And a cloud doesn’t necessarily make that easier. I would say, also, with cloud computing, at least, if data is a core part of your assets that your cloud computing infrastructure should be, as a, as a friend Tom Webster says reassuringly expensive,

Christopher Penn
it should but you should also you should also understand what all of those expenses are, you should be able to dig into where every single penny is going. And a good hosting company would be able to walk you through that and give you the information you’re looking for of, you know, how they’re protecting the data, what those you know, two factor identifications are, or who has access to it, because the thing you don’t want to see happen is you sign up with a cloud computing company, and everybody in their company now has access to your data, like, you know, you need to make sure that you’re understanding what’s actually happening with your data, I think, and this is will be a bit of a commercial plug. But one of the things that folks need to do as part of compliance is dig very carefully into the terms of service for the different companies and read to your point what your, your hosting provider or your cloud provider does, and does not have access to when you take a look at the terms of service. For example, on IBM Cloud, IBM Cloud is very, very rigidly saying they do not have any ownership rights or access to data other than what you explicitly grant. That means that the setup time takes a little bit longer in a lot of cases. But you’re it also means that you don’t have to worry about IBM turning your data into assets for sale or for themselves. Exactly. And

Katie Robbert
I think that that’s that’s a really important point is people need to understand what they’re signing up for. But on the flip side of that, if in that instance, and IBM isn’t responsible for your data, they don’t have access to it, then you still need to factor that into your strategy of they’re literally just holding on to it for me, I still need to make sure that I know all the other moving pieces that I’m still compliant, that I still have my disaster recovery plans in in concert with bears. What do you mean it’s not okay to have protected health information leptin and unencrypted table? I mean, if you really like the inside of a jail cell, yes.

Unknown
I don’t know if the punishment is severe, but it’s not good. It’s

Christopher Penn
not good. No. And depending on on where you’re located, it actually could be on one of the provisions under GDPR is that for significant data breaches, company executives may be held criminally responsible. Yeah,

Katie Robbert
well, and so this all kind of goes full circle back to the thing that tends to go wrong is the lack of planning and I think that there’s parts of it that you know, people are like, Oh, well, I don’t I don’t capture PII. So I don’t need to worry about risk and compliance. Well, you do because, you know, in the instance, that someone hacks into your cloud system and steals all of your data, well, that’s all of your IP and your assets, even if you can’t identify the people. So I think that you know, to your original question planning is absolutely the first thing you need to do. And whether you’re an enterprise a midsize or a small company, you absolutely need to factor in all of these different things and if they don’t apply that acknowledge them but don’t just dismiss them completely. So to wrap up the the AI governance framework which is based off of the ISO 38 500 framework will be available on the brain trust insights website please feel free to go grab a copy it’ll be in PDF form and and and make it part of your next governance and planning meeting to say are you have we checked the boxes quite literally on all these different things. As always, please subscribe to the YouTube channel and to the brain trust insights newsletter and we’ll talk to you next time.


Get unique data, analysis, and perspectives on analytics, insights, machine learning, marketing, and AI in the weekly BrainTrust Insights newsletter, Data in the Headlights. Subscribe now for free; new issues every Wednesday!

Click here to subscribe now »

Want to learn more about data, analytics, and insights? Subscribe to In-Ear Insights, the BrainTrust Insights podcast, with new 10-minute or less episodes every week.

Pin It on Pinterest

Share This