Subscribe: Google Podcasts | RSS
Subscribe: Google Podcasts | RSS
What’s the hot new technology of 2018? AI? Deep Learning? Pole-dancing robots? Maybe. Or, maybe it’s customer data platforms (CDPs) — a topic we actually covered way back in January 2017 on episode #053 with Todd Belcher, who, at the time, was with CDP provider BlueConic. Since then, Todd left BlueConic to start CDP Resource, which is, well, a resource for companies looking to select, implement, and maintain a CDP. We asked Todd to come back on the show to give us the rundown on how there is now — finally — clarity, consolidation, and maturity in the space, as all of the providers have aligned around a common definition of what a CDP is, what it does, and how it should do it. Alas! The space isn’t even remotely there yet! We have yet to even reach the peak of inflated expectations! Which was probably why it was such an informative discussion.
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.
00:26 Michael Helbling: Hi, everyone. Welcome to The Digital Analytics Power Hour. This is Episode 103. Listen, we all deal with marketers, and generally they’re great people. Some of you are marketers, and you’re the great ones I was talking about. But real talk for a minute; sometimes marketers aren’t good at the techy stuff. But one thing they seem to always be good at is really wanting whatever that latest category of the Scott Brinker MarTech map is currently promising to solve all their problems. Today’s landscape, it’s CDPs or customer data platforms. I can dig it. Tim, am I being too hard on marketers?
01:09 Tim Wilson: You couldn’t be too hard on anybody ever if you tried your hardest. And I don’t think I’m being too hard on you for saying that.
01:16 MH: Alright. Moe, am I not giving CDPs enough credit?
01:20 Moe Kiss: Well, let’s just have a bit of a chat about it, and then we can regroup.
01:26 MH: And then I have to ask myself a question: Isn’t giving marketers great tools to connect with their customers a good thing, Michael? Yeah, I guess so. Okay. I guess we’re gonna find out. Now, you probably remember way back on Episode 53, we actually discussed CDPs with Todd Belcher. But since then, Todd has stopped working for BlueConic, which is a CDP, and started working for himself. And now he’s dishing out all kinds of great information and knowledge on his new site cdpresource.com. We’re glad to welcome him back on the show. Welcome back, Todd.
02:04 Todd Belcher: Thank you. Thank you very much. Good to be back.
02:06 MH: Alright. So let’s kick this off this way. I think you probably defined CDP on Episode 53. Hopefully you haven’t gone back and listened, so could you define CDP again and then all of our listeners can compare and then comment on the Measure Slack about how we didn’t make the same definition twice?
02:26 TW: Or would that actually be expected, because it does seem like an area that is evolving with so many entrants, is the definition actually changing?
02:35 MH: True. Everything is a CDP right now.
02:39 TB: Right. I was very curious to see what I said, and I did listen to that episode a couple times over the last week or so.
02:45 TW: Well, thanks for boosting our numbers.
02:48 MH: That’s right.
02:50 TB: I would say my definition probably hasn’t changed a ton, except… Certainly at that point in time, every word that I said was under the big eye of software; right now I can define the category however I please, which is nice. But it doesn’t really make it any easier, unfortunately. So the way that I define customer data platforms today is to say that they facilitate the use of entire life cycle customer data, meaning just about any information that you could capture about a customer without a vendor-based preference for where data is coming from or going to. And I don’t know if that’s few enough buzz words, I know I got knocked on that one last time, but I’ll check in with you guys and see.
03:42 MH: Well, since we recorded last, it is exciting to announce that this podcast is actually a CDP [laughter] and leveraging blockchain technology. [chuckle] We’re also accepting sponsors for 2019, so line up.
04:00 TW: Oh my goodness.
04:00 MH: Oh, sorry, sorry.
04:02 TW: When you say customer entire life cycle, and this is… As I have talked to some specific CDPs, and it does seem like there are CDPs that say, “Entire life cycle from the point that they become known forward,” and others say, “When we say entire life cycle, we mean we will bring in from your DMP, these can be unknown just IDs that enables us to stitch them backwards.” So when we say entire life cycle, I think they go forward after they become known and then become a customer, and then over their entire life with the company is one thing, but how far upstream and does it vary? Does it depend on who you ask or what matters to you?
04:45 TB: Of course, yes. There are dozens and dozens of systems in the category, and they have taken many different shapes, and certainly that kind of anonymous data, or data coming from advertising I think you’re getting at, may or may not be something that’s included within the context of customer data. But almost always there is at least some sense of anonymous being converted to identified data.
05:15 TW: Okay. So that would be… They come to the website, they’re anonymous, the CDP may be collecting some stuff, then they convert, and it stitches them back, or even if it’s on a subsequent visit. I guess it’s more the DMP, the impressions, the audience. And I remember we had this discussion the last time of where’s the line between the DMP and a CDP? Do I need both? Are DMPs all trying to go down-market to say they’ll take in that first party data as well and become your CDP? I don’t know; it still makes my head hurt. Everybody I talk to seems to say, “Here’s the definitive definition,” and then it completely counters something that another vendor has said.
05:56 TB: Certainly, you have a number of problems where the categories have overlap and the vendors or providers may also portray there to be overlap in areas where there isn’t, or there’s similarities that are an overlap. So one example of that would be using a DMP to personalize email versus using a CDP to personalize email. The Data Management Platform personalizing email is very different from a customer data platform personalizing email. At least from what I’ve seen, a DMP is capable of changing offers out in an email, but it’s done based on anonymized IDs that are being matched to known identifiers in the email and there’s not necessarily an orchestrated email campaign here. This is what’s the next best offer for this person, let’s put that in their email in the spot that has an ad.
06:53 MH: Got it.
06:53 TB: And that Data Management Platform approach to personalizing email is very different, I think, from what we see with a customer data platform, and actually sending completely different email content to people.
07:05 MK: So… Sorry, you keep referring to emails, and we’ve just gone through the process of our first CDP, and a big discussion point is about deduping accounts. What’s the industry best standard? Is that something that most CDPs have to offer now, or how is that being managed by the industry?
07:28 TB: There’s a huge divide in the customer data platform category, where some percentage of the customer data platforms have data management capabilities around identity, and those go all the way to deduplicating and probabilistic matching on things like addresses, partial addresses, or phone numbers. Normalizing addresses and phone numbers and email addresses, and matching all that information is a data management function that some CDPs have and some do not have. And then you have other systems that are not CDPs at all, who also help with that type of information.
08:08 MH: I’d love to just quickly pause and do a quick little definition, if you don’t mind, Todd, for people around probabilistic and deterministic matching, ’cause I think that’s helpful for people thinking through the CDP concept.
08:20 TB: Sure. In this context, when I’m thinking of deterministic matching, I’m thinking of, you have one or multiple systems with identifiers and your deterministic matching is taking an identifier that says ABC and finding another identifier from another system that says ABC and saying, “Those are the same person.” Where probabilistic, there’s actually a couple different flavors of that, where probabilistic can mean still using PII or collected information about customers and trying to determine what the probability that this record is the same person as this record based on partial matching. And that’s a different probability to me than what DMPs have offered in the past with probabilistic matching, where they’re saying, “Well, this person’s on the same IP address and they also use most of the same browser plug-ins so let’s assume that it’s the same person.”
09:14 TW: Which gets… ‘Cause the whole nature of probabilistic matching is that there’s no guarantee. You’re not gonna set your probabilistic matching threshold at 99.99999 and say, “I’m gonna treat it as deterministic,” you might, but that’s part of the challenge with probabilistic matching is that it’s a probability, it’s not a guaranteed… This actually… I feel like every few weeks when I’m back in the CDP world, I have these flashbacks to 11 or 12 years ago when we had a whole customer data integration initiative at the company I was at, which was kind of something screaming out for a CDP ’cause we found that we were sending… To some customers, we were sending them 11 emails in a week, but that was because they were registered with us in different systems. And so we went through this whole exploration of saying, “How many of these fields have to be an exact match for us to say, ‘This is absolutely the same person’”?
10:10 TW: But Todd, are you saying… And again this just feels like one of those questions that I’ve asked a couple of vendors, there are some that say, “Yes, we will do probabilistic matching and we will take on all the overhead that comes with that because now you need to be very careful about not treating a probabilistic match as a deterministic match,” so there’s a whole world to that. And there are others that say, “No, no, no, we will let you join, we will let you do deterministic stuff, but that’s not really… We’re not gonna get into the probabilistic and the fuzzier world of matching.” And that, again, is… Some vendors do, some vendors don’t?
10:50 TB: Right. And it’s definitely where we’re at right now. And I think it’s okay because certainly there are organizations that already have some form of MDM system in place that’s doing this.
11:02 TW: That’s Master Data Management. Man, we are dropping the acronyms…
11:04 TB: Excuse me. Arrrgh. Yes, thank…
11:05 MH: Jargon patrol. [chuckle]
11:08 TB: Yeah, thank you for that, yes. But that’s right, a more established organization will already have something in place that’s doing records management, making sure that records are unified, that there’s a system of record and a customer profile, even if it’s not real-time. So it’s not a customer data platform, but it is helping to resolve identities, and that information can be used in a customer data platform that maybe doesn’t have those types of features.
11:35 TW: So Moe, what prompted… Do you know what was the driving force behind, we need to… We, THE ICONIC, need to add a CDP? What was the problem? How tightly was the problem defined that the hope that would address?
11:52 MK: I think it was just the common problem where we have all of our own different channel owners, and they want to be able to segment customers in a different way, and push messaging across different channels, and have it be really consistent because basically they were operating across 8-10 different platforms plus. And they were struggling, I think, sometimes to identify the specific groups that they wanted to do.
12:20 MK: The thing that’s been really interesting on the side, is that I probably didn’t anticipate that it would help me from an analytics point of view. I probably came at this thinking very much about CDP being a tool for a marketer, and it’s actually had this side benefit of me being able to say, “Look at a group of customers and then understand a whole bunch of different attributes about that group of customers,” which I could do in SQL. I can write it in SQL, but instead of writing 200 plus lines of code and spending probably three hours, I can now go in there and do that pretty quickly. Yeah, so I probably, from my perspective, underestimated how useful it would be to myself, so that’s been pretty interesting.
13:11 TW: So a platform has been rolled out and it actually exceeded expectations? [chuckle]
13:15 MK: Yeah, I know, it’s kinda nice.
13:17 TW: Maybe in the southern hemisphere, that sort of thing can happen.
13:19 S?: I didn’t realize that was ever a possibility.
13:22 MK: The thing that I wanted to chat to you about, Todd, is these things can also go horribly, horribly wrong. Anything that you implement, it’s all about how it’s implemented. Who in an organization should be part of this decision and which vendor you choose to go with and why, and what kind of attributes are put in there and what kind of, I guess, values are you concerned with? And who makes those decisions about do you do probabilistic matching or not?
13:54 TW: And who owns the whole… Who tends to own it, ’cause you’re inherently trying to break down, where does it best live, or where does it typically live, and where should it best live? [chuckle]
14:05 TB: So there is no one situation, there’s no one common path even for folks to decide, “We need a customer data platform.” Sometimes there is a very specific problem that you need to solve, and that has pointed you in that direction, and you end up then considering, “What are all the other things that we might be able to use a customer data platform for?” and backing into it that way versus having a holistic data strategy and coming at it from the angle that the customer data platform will be used by the entire organization, and it will need to have some sense of governance, analysts will need this. We’re going to deploy data from this platform across multiple groups within the company. Those are very different situations, and so it certainly depends on what the context is that the customer data platform is being purchased for. I think the best case scenario is certainly that the technology and the customer data are centralized and that there is organization and people have thought about a first party data program and not just how do we personalize messaging to our customers and how do we build our business through customer interactions and just working as better partners for our customers, but also what data do we have as an organization that we may be able to capitalize on next year or the year after?
15:35 TB: So really thinking about not just individual customer profiles, but the aggregate of those profiles and what the ownership of all that data actually means is something that I think a lot of organizations are starting to figure out. This is what we can be doing with a customer data platform. Personalization is absolutely a great thing, activation of the data in order so you can personalize any communication channel, whether it’s a services or support channel or an acquisition channel are all important things. But there’s also this other area where if you have a data lake or some massive amount of customer data combined with other business data and context or experiential data, however you wanna call it, if you put all that together, now you have first party data that you could assume no other organization has. And how can you capitalize on that and how can you use that, either to monetize it directly or to better build your business? I think those are really interesting questions now that should be considered as you’re evaluating a customer data platform.
16:38 TW: But does that mean that… As you’re describing that, it seems like there is this really fine line to walk. That if you say, “We’re gonna solve this one specific problem, we’re over-communicating to our customers, or we’re not doing effective personalization or whatever,” you can make a decision for a CDP that’s a little bit short-sighted and doesn’t have a long-term vision. If you have the long-term vision of saying, “We’re gonna pivot,” and there’s probably some organizational thinking and a shift that’s a two or three-year plan, and to do that we need this foundational enabling technology, but we have to be careful that we’re not promising that it’s gonna solve every problem anybody in the organization has ever had, and we have to have it on a deployment, on a rollout process that means it’s delivering value” which seems like it could be, “Hey, the first thing it’s gonna do is enable analytics to be more effective.” But a thing with a CDP, as I understand it, is it’s specifically meant to be able to support operationalized use of the data, so it’s not a data warehouse master customer record 360 degree view of the customer, it can only be used for analysis.
17:49 TW: So it does seem like it’s a big challenge, especially as the landscape is… Every week there’s another five vendors on the market. So even trying to figure out where to start considering… But somehow somebody in the organization with enough authority, enough vision, enough pragmatism, balancing pragmatism with vision, has to try to lay out what’s gonna happen. So it seems like it almost has to be part of a broader data strategy that you’re really setting yourself up just for vendor creep. If you say, “We’re gonna add another technology that’s gonna pull data from these half dozen systems and solve this one problem,” you’re almost certainly painting yourself into a corner that two years from now you’re gonna be like, “That was not the right way to go.”
18:36 TB: Yeah and it’s hard to say, because let’s suppose for a second, that setting up, either building up your own kind of platform of customer data or buying more infrastructural customer data platform that’s really helping consolidate your data, build out a data lake, manage the customer records; if that has a longer implementation cycle and you have needs next quarter, then you wanna talk about vendor creep. Why not look at two customer data platforms? The spectrum is so varied that I could actually see you going through an evaluation cycle for foundational or infrastructural customer data platform with the entire team, and at the same time, taking an innovation team or something within your organization saying, “That’s all great, but we also need to start learning about unifying customer data and activating it now. So we’re also going to add this other CDP that’s more experience-focused in nature, and maybe we could be a bit more agile with getting it up and running, because we know that work is really feeding into the more foundational work that we expect to get done later.”
19:48 MK: So how does this fit with I guess the movement that’s going on in analytics and data science at the moment, which is very much going the other way? It’s really moving away from vendors. Teams want to use open source, they wanna build themselves, and that’s really like a prominent trend in the industry that I’m seeing anyway. And going through the CDP process myself, it actually feels like that’s the opposite of that trend. That again, you’re giving data to a vendor who then gives you something back. So it feels like there’s a bit of tension there.
20:25 TB: I think it depends on the customer data platform, for sure. So, Heap Analytics is actually an interesting company because they put out somewhat of an anti-customer data platform white paper, I think it was, a month or two ago. It was saying, “Look, you can build your own customer data platform and here’s how you do it.” I thought that was really interesting for sure. But at the same time, looking at the infrastructure part of what you’re trying to do with customer data, you’re trying to collect a whole bunch of information from one place. That’s the data lake. Whether you use Heap to do that or you use a CDP to collect that information, you’re still doing it. And so the customer data platform doesn’t have to be the collection point. If you’re using another tool to collect all that information, that’s great. The customer data platform is there to help with everything after that.
21:19 TW: So the customer data platforms typically allow… I guess, maybe I’m gonna answer this myself. One of the challenges with DMPs, they save our attribution, management platforms. They’re like, “We will bring it all together.” And then on step two of the implementation is, “Deploy our data collection mechanism everywhere that you want to, include this.” Or some CDPs there are the tag management systems that are pivoting, and saying, “We’re CDP.” Well, they’re inherently out there in the, let’s drop a tag in a bunch of places. So I assume they’re saying, “Yes, unified data, even if you’re dropping a tag in some sort of server-side mechanism to pull data in.” But your CDP is typically, or is there a class of CDPs that say, “Hey if you can whiteboard what the keying is, let’s stick with deterministic for now, then you can bring data from Heap or from Adobe Analytics and you don’t have to change those at all as long as we’ve got a way to link it to your email platform data. We will bring both of those in”?
22:30 TB: To me, that’s basically the definition of a CDP is that they must be able to do that, whether they can collect their own data or not, they should also be able to ingest data that’s collected from some other system.
22:41 TW: Well but do they? I guess, I wanna say that… Martech, what’s the current count? What’s your ballpark on how many knowing that there’s a gap between when we’re recording this and when it rolls out? So with your predictive analytics, how many platforms will be out there calling themselves CDPs? It’s got to be over 80, ain’t it?
22:58 TB: It is, yes. I think you can go 90 to 100 right now. But they’re not necessarily all calling themselves a customer data platform but maybe Scott Brinker or David Robert or somebody out there has pointed at that and said, “It looks like or smells like a customer data platform.” And certainly a lot of them are saying, “Yes, we’re this other thing, but we have this customer data platform technology as part of that.”
23:21 TW: Well, maybe to put you on the spot because of the magic of time and the lack of time travel, your guide that recently came out, how many vendors does that have in it?
23:33 TB: 93.
23:34 TW: Okay, holy crap. Okay.
23:36 MH: Wow.
23:36 TW: And it’s probably already outdated even though I’m talking about it before, I’ve officially seen it…
23:42 TB: Well, now I have to make sure I delete enough vendors that it ends up being exactly…
23:46 TB: Exactly 93. [chuckle]
23:48 MH: It can be 93 or more. Yeah.
23:51 TW: Yeah. [chuckle] But.
23:53 MH: So it’s interesting Moe, that you perceive CDPs as kind of monolithic in that sense because I still view them very much as point solutions, and not part of, sort of, bigger ecosystems yet. Like…
24:05 TW: What do you mean by that?
24:06 MK: Yeah, I’m like, “What?”
24:08 MH: So Adobe doesn’t have a CDP. Google doesn’t have a CDP.
24:13 TW: What? Are you including Adobe? I mean, doesn’t Audience Manager kind of start to claim that they can…
24:19 MH: That’s a DMP. Historically, anyways. I’m not saying they’re not working on that stuff, I’m just saying they’re not part of the… If you look at the… What is it… The Hype Cycle or whatever.
24:33 TW: Gartner Hype Cycle.
24:34 MH: Yeah. Gartner Hype Cycle, we’re very much in this ascendant, early phase of all of this, where it’s like lots and lots of… And we haven’t even yet begun the mass consolidation phase, where all the big players swoop in and sweep up all of the CDP players.
24:51 TW: I operate in the trough of disillusionment on all topics, so I don’t care where you think CDPs are.
24:57 MH: Alright, that’s where I see them.
24:58 TW: I’m in the trough of disillusionment.
24:58 MH: I mean, what… Treasure Data got bought, but there’s still a lot of them.
25:06 TW: Salesforce buying Datorama, and all of a sudden they’re talking like Datorama is a CDP. We’re not really…
25:10 MH: But it’s not, is it?
25:12 TB: Well, so Datorama certainly had been included in customer data platform lists way prior to that acquisition. Certainly, it has some of the capabilities that customer data platforms have. I didn’t see it doing that much from a delivery standpoint, it seemed like much more of an analytics.
25:31 TW: Analytics platform.
25:32 MH: Yeah. Okay, good. ‘Cause I agree that it wasn’t doing… Okay. It’s hard…
25:36 TW: But that’s part of the… Part of the business.
25:40 MH: This is the thing that frustrates me about CDPs. I think, is that it’s just super hard to really drill down into what exactly each one of them is and then stick to it.
25:50 TW: It’s almost like you need a guide.
25:53 MH: Yeah, I feel like if there was some sort of guide that could…
25:57 TB: You know what’s funny too is I do have… My guide should hopefully be out by the time this podcast is aired. Right now…
26:03 TW: Frankly, at this point, it better be, Todd. I mean, really.
26:07 TB: Yes. So today, you can go to MarTech Advisor and get their guide, which is fabulous. It covers so much content in the area of customer data platforms…
26:18 TW: I can’t… Every time I go to their guide, I feel like I’m getting part nine. I even followed a link from your blog.
26:23 TB: I understand. Yes, because they’ve been doing this for a long time. I think they set out to do a nine-part series on CDPs way back in the spring, and I think part eight or something is this guide, which is super valuable because it covers a lot of use cases, not every use case but a lot of them. And that would really help any organization looking to evaluate a customer data platform think a little bit more broadly about what features all the platforms could potentially have, so they can figure out what features are important to them and start to pare down that list a little bit.
26:57 TW: But realistically, are there some that are… And I feel like we have run through in to Michael’s point, over the Hype Cycle, I feel like we’ve run through this with BI, we’ve run through this with Web Analytics, where you get three kids in a Bay Area, co-working space, and saying, “We drew it up on a whiteboard and we figured this out,” and you’re like, “Wow, between you, you have seven minutes of post-college work experience and you’re designing something and you’ve got slideware that talks about how you’ve got this licked,” as opposed to… You know, take a Tealium, they’ve been in the industry, whether it was the CDP industry or not, they actually have been talking with real marketers, real companies, talking about their stuff. So when… That’s part of it, I just wanna have the line to say, “What’s my risk aversion level?” because yes, you may have designed a better mousetrap, but I’m sorry you’ve literally seen a picture of a mouse in an Encyclopedia Britannica, and that’s it. I kinda rather have somebody who’s been exterminating mice for five or 10 years telling me they’ve got a better mousetrap. So is there a case of that? Are there upstarts saying, “We are… We’re CDP, we figured it out” and the reality is, it’s like it’s two or three smart people, but they maybe don’t have the breadth and depth of knowledge about the reality?
28:22 MK: You could also argue that they haven’t been biased by the industry and will actually build something for a specific use case rather than trying to tackle like 10,000 problems with one product.
28:32 TW: No, no. I mean, I think that’s…
28:32 MK: I mean, you could, I’m just a devil’s advocate.
28:33 TW: Totally legit. I’m the one that said, salesforce.com, cloud-based CRM would never, never work. Those were guys who came out of Oracle, which was a disaster and Salesforce was awesome. So I can’t predict, that’s a great point. Like, no, why don’t you just do a clean slate and actually get down to the fundamentals of the problem? Although to me, customer data, does feel like again, CDI was… I was in long meetings about customer data integration and there’s a lot that sounds the same.
29:03 TB: Yeah. One thing that I think the definition of customer data platform needs to include and it might sound flippant but it’s the word “software” because like you’ve described, there are companies out there that don’t have a whole lot of software. They have services and more services wrapped around a little bit of technology that they might have built and that could be a three-person start-up on the West Coast. It could be a couple dozen person start-up on the East Coast with VC backing, but they’re both in the same position where they’ve got great ideas and not a lot of screenshots or videos, or proof to kind of share at this point. They may have even evolved out of an agency and some of the stories that they have to share. If you really look at them. That’s a web development story. They developed a new website is what they did.
29:53 TW: Which is interesting. Tag management systems are interesting if you look at Ensighten coming out of Stratogent, basically. And DTM came out of [30:03] ____ Search Discovery and became… Satellite became DTM… I’m not really… I actually don’t know what the Tealium or the Signal back stories are. And some of those were. They actually were solving a very specific problem and figured out how to solve it very well. So they had enough of a near-term focus that they did establish an industry. Who was considered? Is there anybody considered to be the first CDP?
30:31 TB: There are… It’s hard to say because I think there are probably six or so systems that claim to be the first CDP. Maybe more, and I don’t have that as a… I should put that in my guidance like, “This system claims to be the first CDP.”
30:45 TW: Which would be awesome since it already came out.
30:51 TB: What’s interesting though is if you really look at all of the researchers or analysts out there, and you just try to figure out which customer data platforms have been researched the most, do kind of like a FiveThirtyEight view of the field, you get a pretty short list of less than 10 providers who just about everybody includes in their report, which I think is interesting.
31:16 TW: You’re saying… So that’s kind of doing a… What do they call that in academia where you do a study of the studies? If you look at the MarTech Advisor, you Google what David Raab says. If you look at anybody who’s…
31:26 TB: Gartner.
31:26 TW: Gartner.
31:27 TB: Yes.
31:27 TW: And then you say, “Which are the ones that bubble up?” That’s interesting.
31:32 TB: That’s right. Yes. I think Real Story Group, market and markets, or MarketsandMarkets, there are a handful of folks who have done some form of customer data platform guide and so each one includes a dozen or two dozen. In the case of MarTech Advisor, I think they actually included 60. So that’s been the most complete work to date. So definitely a different approach than I am or that I have taken of my release guide.
31:57 MK: Are there actually any open source products available?
32:02 TB: To my…
32:03 MK: I’m just… It’s not something I’ve ever investigated.
32:06 TB: Yes. So the one to my knowledge is, and I hope I pronounce this correctly, is Jahia? Jahia is an open-source customer data platform, if you will.
32:18 TW: Outside of an open source platform, are you finding that there are companies that have just been a little bit ahead of the curve and are basically developed? Again, tag management systems seems like a nice analog in that there were companies that said they basically built an in-house tag management system because somebody on the dev team said, “This is killing us. Let’s just do a project to… ” And they basically built a TMS. Never gonna commercialize it, just used it themselves. Are there organizations that have actually been able to look far enough into the future for their needs around customer where they’ve said, “We don’t need a CDP because we actually… We’ve already built something that’s good enough and probably is mature and tailored to our organization” and therefore they’re like, “We have a CDP. It’s just not a brand. We built it ourselves”?
33:09 TB: It’s possible but I think, at that point, they’re probably plugging in to a couple of solutions that do a very big part of what you might expect a customer data platform to do in another organization. For example, orchestration and the campaign management piece. They probably have pretty heavy-duty orchestration and campaign management. Where some organizations are looking to a customer data platform to do that, they may actually be handing off their data to a system that strictly handles the orchestration of campaigns based on your customer data.
33:44 TW: So, you’re saying they may have… They probably have built everything from scratch. They’ve got systems that they have integrated sufficiently to lessen the need. They probably still have some limitations. But does that mean they have an operational customer data store? Is that kind of something that almost has to exist?
34:09 TB: Well, I think it certainly means that they have… They’re bringing in the customer data together somewhere and if they’re not doing that themselves, then they have… They may have put themselves in a bad situation. And I don’t remember how many years out we’re looking but it can be hard to predict where data is going to be coming from and what different forms it might take. So I think the nice thing about a customer data platform is it lifts a little bit of the dependency on your own internal teams to take care of things like data ingestion, unification and the delivery, where certainly, if you have a really tight plan and if you don’t go off the road from that plan, you could look at the next couple of years and build out all the things that a customer data platform can do with other technologies, one way or another. But yeah, I think at that point, for me, obviously biased, sitting here and looking at nothing but customer data platforms for six months, I would just say, “Why? Why, at that point, not think about offloading some of that effort?” Because connecting to other systems and having people in-house to keep all that maintained, it doesn’t always seem to me like a valuable way to spend resources.
35:27 TW: But it’s when are you… If you have to… It’s a classic. IT runs into this all the time saying, “We’re on a trajectory where this is unsustainable. So where you need to invest, in a six-month project, that’s gonna take on an ongoing permanent licensing cost and after six months, we will be roughly exactly where we are but much better positioned for the future.” Which is a tough sell, unless you’ve got that story of the envisioned future of one, two, three years down the road and you’ve got the organizational fortitude to say, “We’re gonna shift to this. Things are gonna change.” I mean, you’re into organizational change and expectations management and then there’s a… You’re picking a vendor. I mean, although your point earlier saying, “Why not… Why not do the scrape and scratch and come up with one solution in the near term to solve a specific problem, but be very clear that the plan is to end-of-life that ’cause you’ve got a larger enterprise-wide vision? It seems like… Do CDPs, generally, I assume, bubble up to the executive suite, that is the CEO is buying into whatever the CIO or CTO is pitching?
36:42 MK: CMO, in our case.
36:43 TW: The what? The C…
36:46 MK: It’s the CMO in our case and he’s been quite involved in the decision. I do think with time, our CDP… We use one called Lexer, will be used by everyone in the building, obviously, with correct permissions, etcetera. But I do think it’s really difficult because the climate at the moment is that most companies don’t wanna commit to anything. Getting an exec to sign a contract for a year at the moment is tough, let alone two years. So having that foresight of, “We’re gonna build this and it’s gonna be with us for three to five years and we’re gonna keep adding to it” I actually think that’s pretty tough in the current climate.
37:21 TW: But it’s, I mean, if you’re thinking about it, you’re like… And that’s why there’s probably part of the reason the industry is exploding is because, I’m assuming, ridiculous switching costs, if you can get implemented. Web Analytics, we thought they had kind of high switching costs, and we said, “Oh, tag management systems, they had high switching costs,” and yet, shit, Adobe is making their entire customer base having to switch from one TMS to another over the next few years. CDP sounds like, you close the deal, you get it implemented and functioning, then it’s gonna be a really, really tough sell to displace them, unless I guess you go to a… Well, we’re bringing in a second CDP. We’re gonna feed that CDP from our first CDP, and slowly decommission. It’s actually kind of terrifying when you’re…
38:13 MK: That sounds really scary.
38:14 TW: When you’re making a big… You’re making a big bet, and people are trying to make an informed bet, which is probably why all of these guides are coming out, “Hey, can I read through this list of 93 and see the future? Which one of these are the most… Able to see the future and have built something that is gonna still be working in three years or five years, and which ones have made, have effectively built something on OS/2?”
38:41 MH: Well, and I think the other thing that I would think about is, sort of, “How much is CDP driving itself versus business requirements driving acquisition of CDP?” In other words, when companies sit down and like, “We need to think about our CDP situation”, do they really have a strong set or roadmap of what they’re trying to accomplish or is it sort of like, “Ooh, I really wanna get all my customer data together and do cool stuff. I don’t know what it is yet, but I heard I need a CDP to do it”? And I think probably everybody’s on some sort of gradient, but I think, I feel like it’s more of the latter than the former right now.
39:21 MK: Really? That scares the shit out of me. If someone said to me like, “Hey, let’s invest all this money and time, and by the way, data and analytics, you’re gonna be working closely with this team to help integrate all of the data. And we’re doing it ’cause it’s cool, and we think we might do some fun stuff with it.” I would probably get out my No button.
39:39 MK: Yes, I have a button that says No.
39:41 MH: Moe, we can’t all be as forward-looking as the company that you work with, so…
39:50 TW: How forward-looking are the CDP vendors, how much… It’s always, I think, a challenge for vendors or providers.
39:56 MH: That’s fair.
39:57 TW: That actually recognize the future that they’re envisioning and not seeing that as the only future but… I mean, any provider you can’t… They will always… “Yes, we can do… ” I was actually on a call with one and I was kind of impressed they said, “Yeah. We don’t really do that third party bit, unknown.” And I was like, “That’s impressive.” I wanna buy them just for actually calling out that, no, they’re not gonna bring in YouTube view customer level data, but that seems like a challenge as well. As you talk to these vendors, Todd, are they… Do they see the choices, the things, that they’re not trying to do clearly?
40:41 TB: Yes, and so some of them will flat out say that they’re not going to get into certain areas and others are super excited about some of the things that they have planned, so excited in fact, that it’s at an NDA level to just talk about what’s coming in the first quarter of next year which is, it’s… On one hand, it’s exciting, on the other hand, it’s frustrating because the… There’s so much differentiation. We haven’t even used… AI and machine learning haven’t even come up yet in this conversation, but obviously, the forward-looking CDPs are all talking about the AI and what that’s going to do for orchestration, so… And in fact, maybe the AI is even going to tell you what kind of data to collect, who knows, right? The… It’s getting very crazy with AI and machine learning. Certainly it can be hard to look and say, “Is this really visionary or is this a whole collection of buzzwords that’s being put on a product roadmap here?”
41:44 TW: There’s so many meta-analyses that I wanna… I wanna go to that list of 93 and say, “How many say AI or machine learning on their website?” And I’ll sympathize with 93 product managers, how the hell are they keeping up with what’s the reality? And that’s a problem when we’ve got a mature consolidated space, like web analytics. One vendor has gross misinformation about what another vendor can do. It would be… I hate to be a product… ’cause all you can do is you have to just talk about how awesome you are, ’cause there’s no way that you’re gonna be able to realistically say, “Here’s how we actually are different from the competitive set that you, customer X, are considering… The three providers you’re considering, here’s how we’re different.”
42:32 TB: Fortunately, there’s no shortage of providers who are willing to tell us how awesome they are. So, I think if you go to them with specific questions, that’s the best, the best that we can do right now is to really put some forethought into that and go to vendors with, either the same or very similar questions, to try to tease out the differences between them.
42:53 MH: It’s almost as if we really need someone familiar with the space to sorta create a consistent guide.
43:00 TB: Like a resource, almost.
43:01 MH: Yeah, like a resource.
43:02 TB: Yeah, yeah.
43:03 MH: So that people can start to separate fact from marketing as the iconoclastic Jim Caine would say, ” [43:12] ____ Hyped lies and shitty sales guys.”
43:16 TB: Well, to be fair, that’s definitely why I started down the path. I spent long enough time toeing the line for one particular provider and I’ve done that in the past as well, in the web analytics industry. And when you come out of that, you really wanna know, What is everybody else really doing? I wanna understand that. I wanna help other organizations figure that out so that they don’t… Hopefully, they don’t make the wrong decision or they make a better or more informed decision with a little extra information.
43:47 TW: So let’s talk about Coremetrics for a little bit.
43:52 MH: Wait. Don’t you mean… What’s it called now, Customer Experience Analytics with Watson?
43:57 TB: Watson Customer Experience Analytics. Yeah, you got it.
44:01 MH: Oh, okay.
44:02 TW: He stays in touch with the old…
44:04 MH: Yeah, you gotta stay… Alright, actually what we’re gonna do right now, Tim, is something we like to do on the show called “The Last Call.” So we just go around the horn, talk about something we’ve seen that’s interesting, what might be of interest to our listeners. Todd, you’re our guest, what is your last call?
44:23 TB: Alright. So I guess it’s a kind of holiday, gift-giving season. So I wanted to pull out one particular gift that was a big win in my household, which is Adafruit’s Technologies Circuit Express, I think it’s what it’s called. So it’s like a little device that you can use to either teach yourself or your kids some basic programming and kind of engineering skills. Very cool stuff. Adafruit Technologies. They have Circuit Playground Express is what it’s called.
44:53 TW: Like, what age range?
44:55 TB: Five and up.
44:56 TW: Okay.
44:56 MH: Nice.
44:58 TW: You’ve been working with it? Is it helping you out?
45:00 TB: So yeah. Obviously, you get much different use out of it as a five-year-old as a 10-year-old let’s say. But there’s lots of add-ons and things that you can pile on top of it. But it has a bunch of built in sensors, like accelerometers and things like that, that you can kinda play with.
45:17 MH: Ah, awesome.
45:17 TB: You program it with a really easy to use web interface which is just a lot of fun.
45:24 MH: Nice.
45:24 TB: Yeah.
45:24 TW: So you learn kinda loops and variables and that sort of stuff too?
45:26 TB: Exactly, yes.
45:27 TW: That is fantastic.
45:28 MH: Alright. Tim, What’s your last call?
45:32 TW: I’m gonna do two different articles from Medium.com because they’re both cool and I don’t want them to languish too long. One was by Alexandros Papageorgiou which was called, “Choosing between R and Python: For the Digital Analyst.” So it is a… For anybody who has been like, “I’m about convinced I need to learn one of these. The holidays are coming up. I’ll have some down time. Which should I learn?” He does a really, really thorough job of kind of comparing the two from the background of a digital analyst. He comes down slightly on the side of R but it’s a really well-written article where he says, “Look, these are the things that might actually tip you to Python instead. And yes, in a perfect world, learn both.” The other post was by… Picked it up on the Measure Slack, Rich Page wrote it which said… It was called “The Best Visitor Feedback Questions For Improving Any Website”. So, he basically has a huge list, kind of broken up in different groups of questions, that if you were going to, especially if you’re heading into kind of the CRO world, of ways of survey questions you could or should ask, which just seems like a great resource when you’d get a client or get your company to actually say, “Sure, we’ll do a survey. What should we ask?” It’s a great list.
46:48 MH: Awesome.
46:48 TW: So, those were two and they were quick.
46:50 MH: That was really quick.
46:52 TW: But I’m not done yet.
46:54 TW: No.
46:55 MH: Okay. Hey, Moe? What is your last call?
46:58 MK: I have two but they’re intertwined, so I feel like it’s really one. So I’ve been reading up a lot at the moment. I’m really passionate about understanding what makes a good leader and what are good leadership qualities, and it was actually something we were talking about in Measure Slack a few months ago, which has started me down this journey and I read this article which had… The title has nothing to do with the content. It’s called, “Work for someone who values your talents, hard work, and loyalty. Life is too short for anything else!” And I’m gonna butch this name, it’s Oleg Vishnepolsky. But he talks about the job of a leader is summed up as PRIDE, where P is promote, R is reward, I is involve, D is defend and E is empower, which I really love that concept. But then when I read Laszlo Bock’s book, Work Rules! , I actually found myself kind of in conflict because one of the things that Laszlo talks about is the fact that promoting and rewarding should not be something that a manager actually has within their tool belt. So now I’m kind of thinking along the lines of like the role of a leader is IDE, so like involve, defend and empower. So I’m just doing lots of thinking on this, I really recommend Laszlo’s book, Work Rules! If nothing else, he talks about all the really cool experiments that their people in Culture Team do on their own staff which I was find really interesting, so.
48:24 TW: Once you figure that out, if you could maybe spend a little time with Michael kinda get him squared away… [chuckle]
48:29 MH: I could definitely use the help, absolutely could use the help. But it’s mostly ’cause there’s some real tough cases.
48:41 TW: [laughter] Well funny. So, boss, what’s your last call? [laughter]
48:45 MH: Okay easy. So, yes, I also am carrying a twofer today and so first ran across this recently and it’s both scary and exciting, it is draft legislation in the United States being introduced by Senator Ron Wyden, the Consumer Data Protection Act, and so things like GDPR used to be over there, now they’re over here. So it’ll be very interesting over the next little while to see whether or not this legislation gets legs and gets picked up and makes any progress. In any case, if we were thinking that maybe we wouldn’t have to grapple with this on US shores, I think…
49:32 TW: So that’s federal instead of just the California… Okay.
49:34 MH: Federal. That’s a Federal Trade Commission Act as opposed to the California legislation that’s pending. And then the other one is a little bit more fun and/or scary…
49:47 TW: Literally anything would be more fun than federal legislation.
49:50 MH: Draft legislation.
49:52 TW: Okay [49:54] ____ Yeah.
49:54 MH: To be clear, I’ve not read the full document. [chuckle] So I ran across this because I think I got it from Ian Thomas and he posted this. So there’s a guy who is curating a list of Awful AI, David Dow has a list of Awful AI and its uses. And so it gives you kind of like a list of all these different AIs that have been used for terrible things. So don’t jump in that AI boat too quick, all you CDP vendors.
50:31 TW: That sounds like the Cathy O’Neil, The Weapons of Math Destruction?
50:34 MH: Yeah, probably somewhat similar vein, for those of you who like to take a skeptical view of all things AI, that’s a good little primer, it gives you some talking points and some tools to talk about.
50:47 TW: All right, we’re so happy to announce that the podcast is now an AI-powered customer data platform.
50:53 MH: Yeah. Right.
50:54 TW: Instead of the Digital Analyst Power podcast.
50:56 MH: Yeah. The Blockchain Analytics Power Hour. [chuckle]
51:00 TW: Yeah. [laughter]
51:01 MH: It’s too funny. All right. Well obviously, as you’ve been listening, you’ve probably been thinking, “I have a very specific question. Or where can I get this guide? Or how do I get in touch with Todd Belcher to ask more questions? Or any of us.” And the correct answer is on the Measure Slack or on Twitter or on our Facebook page. We’d love to hear from you. CDPs are not going away and there’s much more work to do to really illuminate what their role is in the grander scheme of marketing and we really appreciate, Todd, you coming on and talking to us again about it and helping kinda take that next step for us and our audience. So maybe the future is not totally foreseen but one thing I know that for my two co-hosts and I, we can fully recommend is that you keep analyzing.
52:00 (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.
52:18 S?: So smart guys want to fit in so they made up a term called analytics. Analytics don’t work.
52:29 MH: We’re a health check friendly facility. It has been zero days since our last incident.
52:37 TW: And I guess Todd, by the time this comes out, your CDP guide will be out.
52:42 TB: It ought to be. [chuckle] Yes.
52:43 MH: Good, ’cause it’s my last call obviously. [laughter]
52:48 TW: I can make them in different [52:48] ____ ray form… Woah…
52:48 MK: Are you off ’til today? How much have you drunk?
53:02 TW: Do you differentiate yourself from him in some way?
53:06 S?: Yeah, that dude’s a fucking pretender. [chuckle] Yeah. Yeah.
53:10 S?: Oh we’re good. Look, I think now you’ve just challenged me. [laughter]
53:17 TW: Now it’s a drinking game.
53:19 S?: Rock flag and 93 vendors.
Subscribe: Google Podcasts | RSS
This site uses Akismet to reduce spam. Learn how your comment data is processed.
https://media.blubrry.com/the_digital_analytics_power/traffic.libsyn.com/analyticshour/APH_-_Episode_220_-_Product_Management_for_Data_Products_and_Data_Platforms_with_Austin_Byrne.mp3Podcast: Download | EmbedSubscribe: Google Podcasts | RSSTweetShareShareEmail0 Shares
Subscribe: Google Podcasts | RSS
[…] #103: Customer Data Platforms Revisited with Todd Belcher from CDP Resource […]
[…] was elated to be invited back to the Digital Analytics Power Hour to be interviewed on the topic of customer data platforms for episode #…. At some point during the podcast, I let it slip that I maintain some "538-style" data points on […]