It’s 1:00 AM, and you can’t sleep. The paid search manager needs to know whether brand keywords can be turned off without impacting revenue. The product team needs the latest A/B test results analyzed before they can start on their next sprint. The display media intern urgently needs your help figuring out why the campaign tracking parameters he added for the campaign that launches in two days are breaking the site (you’re pretty sure he’s confusing “&” and “?” again). And the team running the site redesign needs to know YESTERDAY what fields they need to include in the new headless CMS to support analytics. You’re pulled in a million directions, and every request is valid. How do you manage your world without losing your sanity? On this episode, analytics philosopher Astrid Illum from DFDS joins the gang to discuss those challenges.
Items Referenced in the Show
- Astrid Illum
- Digital Analytics Power Hour LinkedIn Group
- DAA OneConference
- Wardley Maps: Topographical Intelligence in Business Strategy
- (Book) Everybody Lies: Big Data, New Data, and What the Internet Can Tell Us About Who We Really Are
- (Recipe) Strawberry Poptarts
- (Podcast) Radiolab – Unnatural Selection
- Tim’s Shiny App: Test Simulator
- Contentful Headless CMS
00:04 Announcer: Welcome to the Digital Analytics Power Hour. Tim, Michael, Moe and the occasional guest, discussing digital analytics issues of the day. Find them on Facebook at facebook.com/analyticshour and their website analyticshour.io. And now, the Digital Analytics Power Hour.
00:27 Michael Helbling: Hi, everyone. Welcome to the Digital Analytics Power Hour. This is episode 122. Not to make matters more complicated but it appears that being a great analyst is after all just a little more complex than knowing R, writing great SQL, mastering data visualization and the relationship between digital data and the organization you support.
00:52 MH: It turns out there are different kinds of people who you will be working with, both in type and in style and sure but we have alluded to it over the years, what to tell an executive versus a product manager or how to handle situations such as GDPR or if the analysis has no clear conversion metric in some part of the business and if you’re working in a large enterprise organization, that could span multiple divisions, departments and even continents. Such a large diaspora of disparate and sometimes desperate stakeholders is truly a complex and challenging aspect of an analyst’s job but never fear, we will not shy away from discussing it here. Tim, you’re known for your diplomacy.
01:40 Tim Wilson: Let’s cheer.
01:45 MH: Okay yeah, sure. I mean it’s just one aspect of being the quintessential analyst and hey Moe, do you ever deal with this while also simultaneously recording a podcast?
01:56 Moe Kiss: It has been known to happen from time to time with the different people and pulling me in 50,000 million directions.
02:05 MH: [chuckle] But we do need a guest and someone who could round on our discussion while providing relevant experiences. Astrid Illum is a digital analytics at DFDS in Copenhagen and prior to that, spent years as both a digital and marketing analyst at Kristeligt Dagblad, I can’t say that word right but I’ll be trying my best and other roles in marketing communications that operate AS and today, she is our guest. Welcome to the show, Astrid.
02:37 Astrid Illum: Thank you, Michael. I’m happy to be here.
02:39 MH: Yes and we’re happy to have you. So maybe just to provide a little bit of context, could you give us all a little bit of understanding of DFDS and some of it’s background ’cause it’s a very interesting organization. I think it’s why we were so excited to have you on the show because I think your role inside that organization really embodies this topic.
03:01 AI: Well yeah, DFDS is an interesting company. It’s a very old company, it was established in 1866 and it’s quite large in Europe, at least it spans a lot of different countries, 20 countries and there’s around 7,000 employees. We’re headquartered in Copenhagen but we’ve got a lot of the operational centers also located in other European countries. So not all the steering goes on from a central location but what we do is that we’re a ferry shipping service and transport solutions so that means it’s both passengers going on ferries, either cruises or just big or short trips from, for instance, France to UK and then there’s a shipping service where people, logistics companies and end customers ship stuff with us on our ferries and we also have our own logistics company.
03:52 TW: That’s a lot.
03:53 MK: That really is a lot.
03:56 AI: That’s a lot. Yes.
03:58 MK: I suppose so, even just the complexity of, you have your own passengers but then you would also have businesses that you work with, even just that tiny portion of that problem sounds complex.
04:10 AI: Yeah but also just the fact that our logistics company is a customer of our ferry shipping solutions and books spots on our ships alongside other logistics companies. There are some shutters between that part of business and the other part of the business. Then there are also ferries that are both freight ferries and passenger ferries at the same time. So there’s a lot of complexity in that but at the end of the day, it’s all about ferries, terminals, trucks, an A to B process in some way. So there is an underlying process that can be captured in a description that would fit us all, at least that’s what I want for us in the future.
04:53 TW: But if the business is inherently moving things but when it comes to the marketing and the digital side, it does sound like it’s fundamentally… There’s B2C for passengers, although I assume there’s B2B for passengers as well.
05:10 AI: Yeah.
05:10 TW: And then there’s primarily B2B for freight, which is a completely different procurement process, right? It’s not like I’m an online retailer and most of the transactions are happening is purchases online, right? Aren’t you dealing with groups that are fundamentally thinking, “This is how we sell” is fundamentally a different way in the role that digital plays within that or no?
05:35 AI: Oh, yes. For the B2B area, there are still the relevant activity of seeing a new company in the neighborhood and going to them and doing a cold call in person to say, “Hey, if you want to ship some stuff with us… ” so there’s also the very, very basic type of old-fashioned business sales happening and then the digitization of all those processes is happening just slowly. So I would say that that’s one of the major dilemmas that we span across a mature B2C area when it comes to digital analytics and digital offerings generally but then also less mature, at least when it comes to digital marketing B2B area and we’re doing the first baby steps or maybe it’s more than that I would say but there is at least quite a big journey ahead of us before we reach a similar level of maturity across the two different areas.
06:26 TW: So in a high level, who of those 7,000 employees… Which groups are the ones… You’re in a position where you are trying to serve broad masters. You could have somebody saying, “We absolutely need to make this update… We’re making this update to the site and need analytics for this part of the B2B lead management process or something and you could have somebody else coming at exactly the same time with the exact same urgency from their perspective for some other work. It seems like the majority of the company gets to live somewhat in their silo, not that everybody is not in some way working across departments but do you feel like analytics falls as one of these areas that inherently has to span across those competing priorities which is something that the C-suite has to do or am I mis-imagining that analytics falls in this kind of unique aspect on that front?
07:30 AI: No, no, that’s completely right. To address the first part of your question with regards to the groups of stakeholders, I think there are two overall different groups. One is the business drivers, the business people who are responsible for selling stuff or giving experiences to people and then there are the internal teams, the experience teams who develop all the different functionalities and features and those groups have different needs for data and they use me for different things and my stake in what they do is also quite different.
08:04 AI: So if we look at the business drivers, as you say, they are allowed and they still live in silos, when it comes to the B2B and the B2C area but beyond that, we also actually have a legacy of compartmentalization or a localized thinking when it comes to the B2C area because we’ve had country offices doing their own marketing for quite a lot of time, many, many years, that’s been the standard operating procedure. So a marketing director for each country and marketing specialist and agencies and whatever you could imagine within the B2C area duplicated across many, many countries.
08:42 TW: And in that case was there analytics? Did they get to manage those locally, on their own, set up their own property, their own everything, they just… They had an entire ecosystem within themselves?
08:56 AI: Yes.
08:57 TW: Okay.
08:57 AI: But they share the back-end system, the booking engine and that fostered some limitations. But it was a set up where particularly one market was more advanced than the other so it was the birthing ground of new stuff and then those were slowly disseminated but the speed of dissemination wasn’t quite at the level of the ambition so you had… The further away you got from the core markets, you would have less advanced setups, less options, less data, less activities so that would be like a call and then circles of complexity going out to the simplest setups and the agencies wouldn’t necessarily talk to each other either.
09:36 AI: So what it means is that now that we have done away with or in the process of doing away with the separate web entities that we’ve had, we had a separate website for each country and we’re moving into one website and one Google Analytics property and one Google Tag Manager container. What it means is that we’re trying to create this shared picture of what is it that we’re tracking, what is it that we’re supporting in terms of integrations? But it’s an ongoing process.
10:03 AI: But the second part you asked about, with regards to the prioritization and if you could see digital analytics as a space where the silos meet up when they come with a requirement and I think that is totally the case. That is the case that the desires of the “here and now” setup that people have, it has gone on for many, many years. It has been a culture of the business drivers showing their prowess and activity and results through that type of continuous and agile marketing working and there hasn’t been any central pushback from that because there’s been de-central handling of all these setups.
10:42 AI: So the desire to halt the agile marketing in favor of cleaning up, purifying, doing foundational digital analytics work is not that high and that has also become part of the history between the process between the business drivers and the experience teams, right? So the experience teams are used to a way of thinking about analytics as a continuous stream of never-ending requests that they don’t really know what the value of it is, they’re not in the feedback loop of all these things what they’re really supposed to be doing.
11:21 MK: So I’m just curious from a business perspective, that is a huge transition to go from. From having disparate teams for each country with different marketing efforts and analytics. I guess, what drove the business to really decide that that was important. That you needed to pull it back and centralize it?
11:38 AI: I think it was born out of the rising awareness of the role of digital as an area where shared services are handled, are grown, developed, innovated and that there is much more value to be had through leveraging the scale of DFDS rather than harnessing the competition of individual countries against each other and I think that’s the digital maturation of the whole market, not just ferries but any general market right now. The tipping point of digital solutions coming to a level where they give us benefits beyond just a simple utility, I think that’s a tipping point we’ve realized that DFDS said, “Yeah, this is the future.”
12:24 AI: So we are not served by having 10 different solutions for this one process. The value added in getting one, two for all is just so huge and then I think there’s another huge driver, which is to say that the idea of one DFDS is also… It’s something that has weighed a lot of people at DFDS and it informs our strategy to begin thinking of DFDS as one thing and not as separate entities.
12:53 TW: So getting to the point where everything is unified and centralized and in one big happy family is great. It seems like the realities of the people, the complexity, the processes, all of the challenges that come in, that now there’s centralized prioritization across very different groups that has to be worked through, my sense is that always winds up being wildly more hard and complicated than when the decision was being made and it was being drawn up on the whiteboard. Everybody thinks about, once we get there it’s gonna be awesome but the process of getting to that point seems like it’s always just 10 times more complicated than anybody really thought, is that fair?
13:51 AI: That’s fully true but I believe that people knew how complicated it would be to some degree, in advance but I think that maybe there’s also an issue with the way that you frame it because I don’t believe that it’s something that will be done because when you look at us, 21 countries and all these employees, different business areas, it’s not a process that is ever going to end so I would think of it more like an ambition, right?
14:08 AI: So now we’ve made this decision to have one website or one web presence and that’s an ambition which is not yet answered in the organizational part of that equation, on the processes below. So it’s just the starting point and it’s a output facing point and I think that’s a good place to start and then that forces the process of unification across all the different underlying processes but it’s never going to be done. We’re such a big organization. We’ll always have multiple people coming up with new stuff at the same time and suddenly you have 10 different intelligent agents, machine learning stuff happening that may be interfering with each other because they didn’t know they were doing it. It’s never gonna stop. So this is an ambition and a way to also work on getting all the infrastructure in place of communication about this.
14:55 AI: What’s the governance of being on the same website? Well, how do we handle policies? How do we communicate? And I think that that process is much larger than just the sheer presence on the website. We can see that there are effort, it flows into also how does digital work together with IT and how does these two parts of the organization work together with the business? That’s also a continuous question that we’re working on and improving I hope. So yes, it’s hugely complicated.
15:25 MK: I also think though, it’s about how do you get all of those stakeholders to relinquish control because, at the moment, my guesstimate is that they would also be doing a lot of their own reporting and telling their business how they’re performing and if that’s coming back to be centralized, they’re gonna have less control over how they tell the story of their performance and the numbers and that in and of itself is a… Even just when you, as a team, mature from the marketers reporting on their own performance, to having a digital analyst that’s reporting on performance, even that can be tough, let alone when you have 21 different country directors who now suddenly are gonna be held to a different level of accountability, they might make different calculations and so on and so forth.
16:09 AI: Agreed. I think that just to be clear, the different decentralized organizational compartments, they were broken up for the passenger business and the central marketing agency was created internally. So there’s been a new line of business that is specialized around the different parts of the funnel and so they’ve done away with all the de-central directors, just to be clear. But I think that its worth here to mention that my role was new, it was newly invented with me, essential analytics person spanning all the different business areas and there is a full scope of reporting happening within the passenger area, the B2C area and I’m not going to replace that, that’s not realistic.
16:55 AI: So my role is more to ensure that we have standards and ensure that we have a shared understanding of our data source and to achieve that goal, what we’re going to do is go away from the direct reporting through Google Analytics, since we have 360 to actually report on the data through BigQuery and in that settin g, agree on the definitions of data that we think are useful and then everybody builds their reporting through that source, in [17:22] __ but I wont…
17:24 AI: They’ve got an army of people doing reporting and consuming numbers. My humble ambition is to be able to, at some point, giving people a fuller scope of attribution analysis through the integration that we’re doing in Google Analytics because what we used to have with all the separate entities and properties and installations meant that they didn’t have necessary old external data sources integrated either and that’s what we’re also moving into. Ensuring that, yeah, we did include Search Ads 360 and we did include all the different parts of the Google Marketing Platform or Facebook or… To achieve those integrations that allow the business to do more advanced stuff, that’s my ambition, that that’s the capability that I will provide and I think that’s, for them, at least for now, it’s a good payoff for giving away a bit of control but it’s an ongoing exploration. Mutual exploration, I would say.
18:23 MH: Okay, and I’ve got so much to say.
18:27 MH: So first off, it just reminds me why I hold you in such high esteem, Astrid. The way you think about some of these problems is amazing. So I don’t even know if people are really connecting the dots on some of this stuff but a couple of things really stood out to me and I want to see if maybe you could just expand on them a little bit. So this is almost going all the way back to the first description you made of DFDS and how you integrated that and I’m really glad that you just mentioned that this was a brand new role that was created in the context of centralizing analytics within that organization ’cause I think that helps.
19:07 MH: But you said something about understanding that underneath all these different business areas and lines of business, there is this common thematic element of the transportation or moving ships or things like that and I was wondering, how did that emerge for you or how did you approach that in terms of doing that? Because I think that’s really crucial for an analyst who’s operating in an environment like the one you are, to find that thing that you specified and I wonder if you could talk a little bit about whether there was a process there or just an emerging understanding as you grappled with the data and the business itself.
19:49 AI: Yeah. Thank you for that good question. I think that it’s been an ongoing process for me to begin having that ambition even and I think, initially, just grasping what is our business, what are we doing and what are the differences and likenesses between the different parts of it. I think that initial integration effort, in my understanding, was what sparked my interest but then, when we did this huge unification process, we also did a switch up in CMS and moved to the CMS called Contentful. I don’t know if you’ve heard about it but it’s a headless CMS and it’s very much based on an idea of underlying, meaningful information architecture and they really sold me on that and I began thinking a lot about how much sense it makes that we actually have many different assets that are similar across our businesses.
20:42 AI: So we have terminals, right? And from a passenger perspective, that’s one thing and from a B2B perspective, that’s another thing. So inside the CMS, there could be a content piece that contained all of these things but then had each of these elements, descriptions, photos, whatever it could be, pitches, tags, either B2B or B2C and then CMS could pick up that content and display it in a relevant context with the right tag, selecting which part of it to expose and that we’re thinking about it, began making me aware that you could actually conceptualize this also when it comes to our e-commerce because we’re all of us, a bit stuck in the Google Analytics e-commerce if we’re in that area, thinking of products and we’re forced into that and that’s fine but if we are able to take the different parts of our business and then unify that in our thinking, then afterwards, mapping it into the e-commerce settings, I think, is a much more fruitful endeavor and I think it’s…
21:46 AI: For us, it ties into the architecture journey we’re also undergoing because our strategy within architecture is one of composable architecture and it’s an idea of building components that can be used by all the different systems that require them and that way, you can much more agilely include a new component and decide whether you should build it yourself or purchase it.
22:11 AI: So the speed to change is quite different and that, for me, fits together because if we have a shared understanding of our whole business, of what are the individual parts of moving something, person, goods from A to B and then there are added services while they’re on the ferry, it might be a person who wants the upgraded cabin or booking a dinner reservation and all of those things are added services then. If we can conceptualize that in the same way, then maybe we can look at it at the endpoint afterwards inside of digital analytics and see new patterns emerging is my hope.
22:52 MH: Yeah, well that’s… I love that.
22:55 MK: You actually sound like the ultimate diplomat, which it sounds like that’s the skill that you really need to pull this off, is the ability to zoom out, see the big picture and then, I guess, pull everything together.
23:07 TW: But I was thinking it was… I was thinking… Literally, you’re thinking diplomacy and I’m thinking, what’s going through my head is you’ve gotta have somebody who has got a deep enough understanding of data and taxonomies to come up with a overall… I feel like I’ve watched some people go through this and I’ve seen it done horribly, where somebody just starts with a spreadsheet and lists stuff out and inside of like two weeks, they’ve built this horrible Frankenstein monster, as compared to somebody who conceptualized something and 10 years later, there’s new stuff coming out that still continues to fit right into that taxonomy. So maybe that’s the question. Is this a brilliant data architecture question or a diplomacy question? Is it both and can you find those in the same person?
23:54 MK: I think it’s both and I definitely think the data accuracy… Sorry, the data architecture pace is huge but I think in order to get that moving and get buy-in, then you need to have the diplomacy.
24:07 AI: Yup, I would just like to add also that I think that one of the key ingredients for this is that there should be some value to it and I think that that’s one of the things that it’s on me at least to ensure that that’s part of the equation. Because right now, we’re talking about something like information architecture for a website and we’re talking about assets that are fairly fixed but on top of that, we have the specific campaigns that are running. On top of that, we’ve got personalization and for us specifically, in the new unified setup, we also have locales which is countries and languages taken together and we have just as many as almost the countries that we’re in, so 20 plus locales and all of it is translated content, everything is translated.
24:54 AI: The complexity and the difficulty of assessing the marginal value of the value offerings inherent in differentiating like this, I think, is very difficult for us to, at some point, ensure that we know, is it worth it?
25:08 TW: But are you having to link together the… As you started, there’s a content taxonomy, there’s a customer taxonomy, there’s a campaign taxonomy. Do you run the risk of struggling to move forward at all with a centralized, unified framework because everything touches everything else? How do you keep it from all grinding to a halt because everything’s interconnected to everything else and is there a need to say we’ve got to sufficiently solve everything at once or at least a base-level minimum viable product before we can really move forward?
25:45 AI: Yeah, I follow you but I think that, in my head at least, it’s the same challenge just considered from different perspectives. It is a challenge for us as analysts going forward to prove the value of differentiation because we’re all of us doing it, right? We have the mobile perception or the idea that something should look different if it’s on mobile and it should be looking different if it’s on that platform or that platform and utilizing content across different areas when it’s the same asset that you have, it makes sense but it adds complexity and that added complexity, keeping it alive and updating it and ensuring that it’s continuously correct, I think it’s so, so complicated but I don’t think we have any choice.
26:29 AI: I think that our company is a big one but many companies are moving across borders becoming global companies and creating versions of their site across those different countries and for all of them, there is the same question of, do we just copy? Do we customize it? If we customize it, how do we ensure that it’s updated without hiring a whole country of people to actually overlook it? So I don’t know exactly what the solution is but I think the solution might be the same for all these types of vastly complex fragmentations of what we have as companies present on the different channels we offer our content.
27:07 AI: But just to add a final thing, we can make perhaps different compartments of stability, right? So we will have things that are very stable and I think those, of course, are the ones we should, a no-brainer, the ferries and the terminals, they’re there for 20 plus years, 30, 40, whatever, right? So let’s just get the info about them because we’re gonna be reusing that for a long time. So I think that those things are a no-brainer to actually ensure we have a stable taxonomy for and then that’s all the fluid stuff.
27:37 TW: That’s interesting. I like that thought of what is most fixed and not gonna change. Got it.
27:42 MH: Yeah. So this is perfect because this actually ties into the other thematic thing I wanted to go back and pick on because you’re talking about all these different things and actually, I think maybe one of the things that actually is part of the solution is something you described as a common ambition of pursuing this unified approach and that I would maybe go and say you described that also as vision, right? There’s a vision for us or an ambition and then there was a couple things I wanted to pick out and then get you to just react to.
28:15 MH: One was you described that ambition corporately, as something that will never really have a “done this” or a completion to it. They’re always gonna be new and emerging things that will just expand that and so I guess the two-part piece of that is first, that common ambition or vision that has been established, how are you evangelizing that within the organization? And then the second part is, how do you find people reacting to that concept of, here’s something we’re pursuing that will never actually truly get to in any specific time.
28:52 AI: Well, I think that the latter part is more like a pragmatic reality that each of us have to grapple with in the big organizations. So I’m not sure you can actually quite evangelize it but it’s more like ensuring that there are therapeutic sessions where people can moan a bit about the fact that, but I was just doing this and then now you’ve done it as well and what are we gonna do? I think that’s, at least, how I think about that part of it but I think the other part, I think that the way I try to evangelize the need for us to do this work of unifying is to say that contrary to the European Union, where harmonization means that you try to arrive at a mean that won’t piss people off. So it means somebody has to lower their standards and others have to raise theirs significantly. That’s not the way you do it in the business.
29:42 AI: You find the best standards out there and those are the ones you ensure are used everywhere and to take just a very concrete example, as I’ve mentioned, we have many different agencies involved in our business marketing, digital marketing and they have different practices and now that we’ve added a lot of them to our unified setup, we can see that it’s just… It’s not gonna function if we have 10, 20 different Google Ads tags in our Google Tag Manager container. It’s gonna be horrible to update and it’s gonna be useless in many different ways and then what we did was to say that I have created this Digital Analytics Board with representatives from marketing for both B2B and B2C, BI, the digital VP and the Digital Solution Architect and then I’ve proceeded to explain to them the issues of what is uploaded to Google Tag Manager container, what is the cost of that for us and what will happen if we let people just go in and add stuff themselves and that’s not very difficult because we’ve had to import a lot of the legacy stuff so we’re already bloated, right? So it can only get better from here.
30:50 AI: And then I showed them then that what we can do is that we can take one of the best practices in our old setup and then we can make that data available to everybody. So ensuring, for instance that five, six different surplus parameters are served in the Facebook tag to anybody who wants to do Facebook business for one of our locals, right? They might not have requested that information and they might not be able to use it but why shouldn’t they have it? So there’s… We create a best standard for the tag and then we make that available to everybody who’s working together with us.
31:23 TW: Can you talk about how that board came together? Who’s on it? How you get them engaged? How often you communicate? It seems like that’d be great but it requires that there are engaged members who then go back and have the right roles within their organization. Can you talk about the structure and frequency of communication and how that works a little bit?
31:46 AI: Yes. The whole idea behind the Digital Analytics Board, from my perspective, was to ensure that participation was happening across the different parts of the business and including the different considerations and that it was a transparent deal and not something that just happened willy-nilly in my office, of the corner of the office because as I mentioned, my role is new so there’s no existing standards for how this is done. I didn’t inherit big policy paper about our Google Analytics or Google Tag Manager set up and so what I did was sit together with my manager and talk about who should be a part of this panel or board and from my perspective, I wanted both people who had a technical stake in it so I thought that BI was a necessary representative because even though we don’t have a huge overlap, we are moving into a future where we will have a huge overlap.
32:41 AI: Data Lakes are springing up all around us and I will be a corner of those, our data will be a part of that and so we will be moving towards each other. Then the data solution architect is the one who is tying together all the different systems and teams from a technological perspective and I felt very happy when he was added to the architecture team because to have somebody overseeing those parts of the equation is crucial for us actually getting the most out of all the things that we can do with digital analytics, so he was a no-brainer.
33:11 AI: And then lastly I wanted also to ensure that the VP of digital was there and that was also to get the grounding in place because he’s much closer to the overall strategy than many others at a lower level and to ensure that her perspective was heard and that she was able also to, once we reach some kind of agreement upon a specific thing, communicated back to the next level up and verify the decision high up when necessary.
33:40 AI: And finally, I felt that marketing for passenger or the B2C and marketing for B2B are quite disparate both in numbers and in what they’re doing because of the maturity, we’ve talked about the difference but having them both represented with one person on the board each I think was an important element that also sent a signal about the ambition of the board. If we were to put in people on the board in terms of how much activity or the spend they have, then it would look quite different because there’s definitely a higher spend on the B2C area but I felt that having both of those paths equally represented sent a signal that our goal here is to utilize the breadth and width of our business and to harmonize and see things from a shared perspective and not just continue with a very strong focus on the places where we’re already digitally mature.
34:33 AI: And so I just sent a description of the purpose and the content to the people I thought would be good candidates and said, “Let’s try this out, I’ve got an agenda for the first meeting.” and then we did the meeting and they all thought it was relevant for them to be a part of it. The way I structured the meetings is that I ensured that there’s information points and decision points and I tried to describe the decision points in a way where I ensured there’s a background and there’s a why and a how and a when so that it’s quite clear what I would like to get out of that decision and what the purpose of it is and to name a few examples of what we’ve had up in the board, for instance, is ITP 2.0, 2.1, 2.2 and how to mitigate that.
35:16 AI: Yeah, yeah, that was one of those presentations where, going in, I knew that most likely, the majority would not understand the final technical details of my presentation but at least the people with the biggest stake, in terms of the digital spend they have and that value they would get out of it, understood enough to actually say, “Yes, this is important. Yes, we should pay for some service to ensure that we can fix this.” So that’s one example.
35:41 AI: Another is the Google Tag Manager complexity as an argument for saying that what we provide centrally is a standard of excellence for the different integrations you can do on our website. Another was that we wanted to get a developer resource allocated to work cross-experience teams with a responsibility for analytics because we saw that we have the traditional setup of having data layer pushes to find experienced teams on the website and we’re not doing the scraping to Google Tag Manager. Everything is quite above board. So there’s a lot of work they have to do and there’s a lot of knowledge they have to have in advance and not after the fact and it’s difficult when it’s five different teams working to, for me, to span across that area. So that’s another example and then we meet every other month and I bring the documents.
36:32 TW: And how many people… It’s every other month and roughly how many people are on the board?
36:36 AI: I think we are six in total. So the digital VP, the BI, the digital solution architect and the marketing B2C and marketing B2B and then me.
36:45 MK: It sounds completely incredible and almost like any company that’s of a big enough size and I’m literally running this through my head being like, this is genius because it’s a way of getting all of the stakeholders involved in the decision-making processes around what you’re doing with your data and it sounds also like it’s successful because of the amount of preparation that you do. I really like that split between what’s information that they need to know and then where is the decision points that you need to make as a group and I think a lot of companies could benefit from something like that because, like you said, it really is always still evolving. There’s no company in the world I think that’s like, “Yeah, we’ve got our data sorted, we don’t need to make any decisions moving forward.” It’s always a work in progress.
37:34 AI: Yeah and as I’ve mentioned a few times, the unification process and the growth of our digital part of the business, it also means that we’re developing a lot of new stuff and so there’s a lot of foundational work going on in my department at least, just tagging stuff and to be able to show the roadmap of stuff we have to do and for everybody to look at it and see, “Yeah, right.” That’s all foundational stuff. We have to do that.
38:00 AI: That is a huge relief for me to communicate to that level because it means that the people who are further down, who are struggling to meet their goals and who desperately want some specific type of tracking installed, at least their boss knows that yeah but we are quite full on the things we have to do. So we have to consider it in the bigger perspective of what else is going on.
38:20 MK: So we’ve talked about diplomacy, we’ve talked about being a data architect, it sounds like you also… The third skill is you essentially need to be a project manager.
38:29 AI: Yeah and I must say it’s not my strongest capability in the stack, I would say. It’s not what I enjoy the most but yes, there is a lot of project management in terms of just keeping track of all the different disparate parts because even though I’m the only one working full-time, hired for my role, I do have people working with me from implementation partners and external consultants and so on and managing what they’re doing and the communication and the progress is… It’s a huge part of what I do and it’s… Sometimes, it’s difficult also to bridge the very operational hands-on work that has to be done in the parts of the business where they’re not necessarily as knowledgeable in just pulling out numbers themselves and then at the other end of the line, setting up the agenda for the board and finding that right strategic level of decision that is appropriate for the knowledge and the specific goal.
39:28 TW: Do you find on that front that you… You’ve got the board, they’re aligned, they’re looped in. Presumably, a lot of the stakeholders you’re working with on a regular basis have… You’ve brought them along with, this is how the work you’re doing aligns with the overall corporate vision. Do you find… I feel like, as an analyst, there’s a lot of having to repeat yourself. It may not be a new agency but a new person at an agency, do you find yourself having to give the elevator pitch for, “They just asked to have this pixel dropped in place X on such and such page.” And do you have this response of, “Oh crap, this person has no idea that what they just asked is the old way of thinking.” and you have to bring them back up to speed? Does that make sense?
40:24 AI: Yes and just, that would be sometimes the people internally and it would also sometimes be people externally. I had a recent agency where they wanted access to our Google Tag Manager container to put in some stuff and I was just having to say, “I’m sorry, I can’t do that. You can send me the tags and I will ensure that it gets in there.” And they just didn’t understand, right? They were just, “Yeah but we always have had access to the container, so yes, we should.” And then I, in the end, I had to go over in detail that we’ve got a specific setup where we’ve got the custom tags that actually do lookups for the specific market and locale to ensure what pixel is deployed for which parts of the website and we can’t let you go into that tag and mess around with it because it’s a property of multiple agencies, it’s serving many, many people. So no, you just can’t touch it. We have to be the ones who does that because we understand the setup and once I had got through to that point, they were like, “Oh okay.” and then they told me what they needed but yes, definitely because it’s for many, both internal and external it’s difficult to understand why not just be able to do it quickly.
41:36 MK: So I was gonna actually ask that, about what’s been the communication with a lot of the agencies that you’ve been working with because I imagine there’d be a heap of pushback.
41:48 AI: Well, I think that… I don’t think there has been a lot of pushback yet but we also… We’ve gotten a lot of the small old websites into a unified set up but all the big ones are still waiting to go and that means that I’m fairly confident that that role of migrations will be a bigger headache and that’s also why I’m trying to, at this point, ramp up so that we actually have the processes in place and it’s weird because things pop up all the time that you need governance for when we live in one property and one container and are serving that many different parts of the business and locales and so on.
42:27 AI: Just something like some simple stuff, like having a Google Audience set up. It seems so trivial but then an agency goes and adds 50 different audiences and it’s just one or two locales and it’s… The naming, they don’t have any naming convention for it and just imagining going into that at some point because yeah, there is a limit at some point, even if it’s 360 of how many audiences you can actually have, so it’s… Yeah, we can’t do without that.
42:55 MK: But it also sounds really interesting and in… On reflection like the right strategy that you have migrated some of the smaller sites over first because I’m sure there were lessons in that process that you can now apply when you do the bigger ones. Yeah.
43:12 AI: Yes.
43:12 TW: I just… I feel like I need to chime in. I’ve been on both sides. Well, on the agency side, there are times we are like, “I just need to do this little thing and you’ll get pushback.” and it can take forever, you can’t figure out the way to get through and it’s really hard to know, on the agency or consultancy side, is that because they have a really core unified vision and well-defined processes that they’re aligning with or is it because they really don’t have a clue how to do anything and nobody really knows how to get into GTM? And I think the flip side, if you’re in-house and you’re working with an agency, initially you don’t know, “Is this somebody who really knows GTM and GA best practices or is this somebody who is flying by nine?” For this to work well, you have to feel out, “Who am I dealing? What’s the sophistication level on both parties and figure out where to meet in the middle?”
44:06 AI: I think I’m actually going to be able… I’m going to… Creating briefs, I’m going to be creating a Facebook brief so any agency saying, “Yeah, we’d like to do.” and I’m just gonna send them that brief that shows you, “This is our strategy, this is our tags and this is what… And you can use it like this.” because we know the business, they might not know it, right? It’s new to them so we’re going to be, I think, producing stuff similar to if we were an agency to detail what is in our offerings for you.
44:35 MH: Yeah.
44:35 TW: Well, then you could have the agency that says, “We’ve got a Facebook brief where we explain to the clients how Facebook pixels work.” You’re gonna have competing briefs, like, “No.”
44:46 MH: That’s great. So I think you should probably charge the agencies for that as well.
44:52 TW: Well, but it’s not necessarily gonna be portable ’cause they’re gonna have lots of clients that aren’t at the level, that really are still flying by night and that’s one of the real challenges we have.
45:01 AI: No, there’s this one example, the pushback. So we have agencies that, for instance, have a history with the audience they’ve created for Facebook and stuff like that and they want to keep that history because it’s a bargaining chip for them and their work with us and we understand that and of course we support that but at the same time, centrally, we have an ambition to create our own shared pixel in a set up that might suit us in a future world where we have it enriched already now even though nobody is using it just so that it’s ready for when the maturation arrives so that we have a centralized, maybe Facebook department that will do that work and then we have the data for them, they can use the remarketing stuff off the bat.
45:45 AI: So I think that, at least that’s how I see that yes, the agency will just sometimes want to have some quick little thing and sometimes we can allow that but most of the times, they will have to fit into some kind of mold that we’ve set up but I don’t think it’s my role to limit the business. I think it’s the same way as in architecture, that there is operational freedom if you follow this and that method, right? Yeah, you can build a system but you have to do it in this manner and has to fit with these specific requirements.
46:16 MH: Yeah, I think for your agencies too, you could even try a model where you do something like your Digital Analytics Board, where you even have them come together in some kind of central way. I’ve participated in things like that in the past and it’s been helpful.
46:30 AI: I think that will only happen if we move closer to the business than we are right now because I think the business is quite protective of the specific relationships and the way they steer and ensure that the agencies deliver.
46:45 MH: Well, if anyone can do it, you can.
46:47 AI: [chuckle] Thank you.
46:49 MH: Anyway, we do have to start to wrap up, Astrid, this is really, really good. It just reiterates, I’m like, “Oh yeah, that’s why I’m such a huge fan of you.” ’cause of the way that you describe these things, it’s just so refreshing, so thoughtful, so well thought through and it’s so applicable. I think so many people who listen to our show will get a lot from it.
47:12 TW: She hasn’t even brought up how blockchain is gonna actually be the answer to all of those unification codes.
47:18 MH: It’s a whole other… That’s a whole other episode, whole other episode and hey, we haven’t gotten to last calls yet either, so… And actually, great segue. We do like to do a last call on the show and Astrid, you’re our guest. Do you have a last call you wanna share?
47:35 AI: Yes, I do. I don’t know exactly how well known it is in the digital analytics world but it’s not like it’s news. It was created in 2005, I don’t know if you’ve heard about it but it’s a Wardley Mapping. I thought I’d mention. It’s a guy called Simon Wardley, former CEO and the inventor of the the Wardley Map, he’s created a book or written a book that’s available on Medium about it. There’s also some great webinars you can look at and it’s about creating a map of the structure of a business or service and mapping the components needed to serve the customer or user and to understand that a strategy is something that is contextual, so the why is relative to the exact context of where your company is and he’s also relating that to different stages from innovation or invention to commoditization so that whenever something new is created, it’s slowly moving towards becoming a product and then being a commodity and once it’s a commodity, it can then be utilized to create a new invention that can then be harnessed for interesting special value and which will then also move towards commoditization.
48:45 AI: And to have that concept of your business and what you’re doing to see where should you attempt to commoditize and what will be garnered from that, I think it’s a valuable exercise and I think also just within digital analytics, at least if it’s a business, it spans many different areas. What you can do is create maps for each of the different elements of your business and try to see if there are shared systems and integrations and components that would be utilized better if you just bought something off the shelf that people didn’t have to customize and then you could build something on top of that that would garner more value for you. At least, that’s how I’m thinking about it right now. I’m sure that will evolve. It’s a complicated and interesting topic.
49:28 TW: That’s great, I’ve never seen it, it’s literally 19 chapters posted as 19 posts on Medium.com, that’s topographical intelligence and business strategy.
49:40 AI: Yeah.
49:41 TW: Very cool.
49:43 MH: But is there an infographic that could just compress it down for us? I’m just kidding.
49:48 MH: Okay, Moe. Moe, what’s your last call?
49:53 MK: So I’ve been reading a book that was recommended to me called ‘Everyone Lies’ and I’m totally gonna butcher the author’s name but it’s Seth Stephens-Davidowitz. Dave… Anyway… It’s really interesting because he works in data analytics but using Google Search data. So basically, he started doing analysis that way and then eventually went and worked with Google and now has moved on and it’s super fascinating, I suppose the realities of trying to understand someone’s intent by using search data and I think it’s not a data source that I’d ever typically really thought about that much and yeah, it’s pretty insightful.
50:32 MK: I do have one section that I highly disagree with though, where he goes on a bit of a tangent about how he’s in the prediction business and not the explanation business and we don’t need to be able to explain the why. So one example is that apparently during hurricanes, people eat more strawberry pop tarts and he’s like, “the business doesn’t need to know why, they just need to know that if there’s a hurricane, order more strawberry pop tarts.” and he goes into the same with when you’re working with machine learning models about, “it doesn’t matter as long as you get the numbers right.” and I’m like, “No, I think that understanding the “why” is hugely important.” So yeah, the book’s definitely an interesting read so I recommend it.
51:09 MH: Very nice, strawberry pop tarts all around.
51:12 MH: Alright. Tim, what about you?
51:15 TW: I’m gonna do a twofer but they’re related and I’ve been thinking about… For the last three years, I’ve been thinking about uncertainty. So one is a podcast, an interesting Radiolab. You guys know Radiolab? A month or so ago, they did a series that was all around genius and investigating IQ. So there was an episode on unnatural selection that talked about genomics and what was… What struck me, the topic’s interesting. Initially hooked me that they’re saying we have a million plus genetic mappings of humans and we can use machine learning to then figure out… Predict different traits and outcomes but then within it, they started talking about correlation and 0.3 or 0.4 and you just… I was struck by how much the media really struggles to try to explain what “correlation” means.
52:11 TW: So I went back and forth of saying, “Well,” it’s a little more nuanced than that and you’re kind of on target and you’re kind of not” but it wound up getting a few… If you replace the predicting the individual versus predicting the population, it applies directly to doing propensity modeling, except with the marketing world so that was interesting and so combined with that, on the uncertainty front, a log roll for my test results simulator that’s now been out for a while but it was a little something I built with shiny that it’s a Bit.ly/test-simulator and we’ll link to it but it’s basically trying to help user… Help stakeholders understand what’s really happening when you observe the results of an AB test, what that really means and it’s trying to help with not over-interpreting test results and getting a little bit of intuition around confidence in statistical significance as to what’s going on there and so those two seem related to me. One very tangible, a little toy you can play with, one that’s just a fun, fun listen.
53:20 MK: And the toy is very, very cool to play with. Let’s be real.
53:26 TW: I agree but I built it so I’m biased. What about you, Michael?
53:31 MH: Well, I’ve probably mentioned this before but an organization that I choose to spend some of my free time on is the Digital Analytics Association, which in this year is celebrating their 15th anniversary as an industry association supporting the community and the development of analysts and analytics professionals, specifically in digital analytics and as luck would have it, in a little less than two months, we’re having our first ever Digital Analytics Association Conference, the DAA OneConference in Chicago and as I am on the board of this association, I am legally required… No I’m not. I’m actually really excited for this and I would encourage you, if you’re in the vicinity or have the opportunity to join us at the conference, I think it will be…
54:25 MH: It’s the first of its kind and so I’m very excited to see how it takes shape. Our industry is so young still, that we have these firsts that are happening all the time and it would be great for you if you could say you were one of the first to join this conference, which I think will be a huge success. Alright, so that’s my last comment.
54:43 MK: And the Quanties will also be on then, won’t they?
54:46 MH: That is correct. So the Digital Analytics Association has an award… Set of awards, called the Quanties, which I feel like doesn’t really take into account the qualitative work that so many digital analysts do but nevertheless, it’s a name and we market it.
55:04 MH: Okay. So you’ve probably been listening and like me, been thinking, “Man, how do I become as smart as Astrid about all the complexity associated with this?” And there is a way to do that. One is to read that set of Medium articles that you mentioned. The other is to join in the conversation ’cause that’s where we all met, was at Super Week a few years back. So just being part of the community is a great way to amp up your intelligence and we’d love to hear from you.
55:39 MH: If you have questions or thoughts about this episode, please reach out to us on the Measure Slack or on our Twitter or on the LinkedIn group. We would love to hear from you. Astrid, it is amazing to have you on the show and thank you so much for taking the time away from your family to join us today and to record a podcast about some topics that I don’t think many people give as much thought to as maybe you have and I’m really excited that you get to share some of it.
56:08 AI: Well, thank you. It’s been a pleasure.
56:08 MH: Alright, well I know that I speak for my two co-hosts, Tim Wilson and Moe Kiss, when I urge all of you, no matter how many different kinds of stakeholders you’re dealing with and the different structures of your organization, the one thing you never wanna do is stop analyzing.
56:30 Announcer: Thanks for listening and don’t forget to join the conversation on Facebook, Twitter or Measure Slack Group. We welcome your comments and questions. Visit us on the web at analyticshour.io, facebook.com/analyticshour or @AnalyticsHour on Twitter.
56:50 Charles Barkley: So smart guys want to fit in. So they’ve made up a term called analytics. Analytics don’t work.
56:57 Tom Hammerschmidt: Analytics. Oh my god! What the fuck does that even mean?
57:06 MH: That’s great. It seems like the realities of getting from here to there are always…
57:21 AI: [chuckle] I’m glad of that interruption, I think you can do better than that.
57:27 S6: Fucking Doug, yeah that kid, he’s nixing me. Oh geez, I’m just looking at that company name again and being like, “Oh, I’m gonna screw this up so bad.” Okay.
57:41 AI: Tim doesn’t look happy. He looks worried. That’s his worried face.
57:45 MK: He has that face a lot, let’s be real.
57:47 MH: What’s the prognosis, Dr. Wilson?
57:56 TW: And Moe, yeah, I’m thoroughly… Oh, geez!
58:07 MH: Alright, no one ever did anything I told them to, anyways.
58:11 TW: Yeah, I don’t have to listen to you anymore than I did listen to you when I was supposed to be listening to you.
58:16 MH: Well, I never expected you to, Tim. I look to you as the quintessential analyst…
58:23 TW: Good lord!
58:24 MH: To provide the guidance that our company needs.
58:27 AI: Michael, do you have a KPI that you need to get quintessential analysts into every single… Every single episode?
58:35 TW: Oh it’s beyond that. He’s got other people conditioned to do it too.
58:38 MH: Yeah, well, when Tim spoke at Super Week this year, I made sure to reach out to all of the people who would introduce him and make sure that in their introduction they included quintessential analyst so it’s something that I think it’s important that everyone’s started… Catches on to and it is catching on. It’s both to make Tim uncomfortable and the thought experiment about proliferating a concept and watching it for the tipping point at the right time.
59:10 TW: That’s what you picked? You couldn’t have done something that would bring good to this world? You thought…
59:17 MH: Well, I feel a lot better now knowing that you’re on holiday and we’re not wasting your valuable time at work so…
59:22 TW: [chuckle] We’re just taking you away from your family.
59:29 MH: Here in the United States, we care about corporations as individuals, I don’t know if you’ve heard that.
59:33 TW: Rock flag and organizational complexity.