David Raab explores how CDPs, Customer Data Platforms, differ from their better-known cousins, CRMs. 

Listen to the full conversation here.

We’re talking about customer data platforms, or CDP’s. And I think we should probably start with the basics. What is a customer data platform, and what makes it different from CRM’s, customer relationship management systems that listeners are probably more familiar with?

 

Okay, so the definition of CDP is packaged software that builds a unified, persistent customer database that’s accessible to other systems. Like a CRM, a CDP is packaged software, something you buy, you don’t build it like you would build a customer data warehouse or data lake. Unified persistent customer data is really where it’s different. Again, a CRM stores data persistently, in the same way that a CDP does. But the unified customer database bit is really where they’re going to differ from each other because the CRM really works with data that’s been created within the CRM. So someone calls on the phone or someone talks to a salesperson or possibly makes a transaction depending on what your definition of CRM includes. And that all gets captured in the system. And it’s stored when it’s accessed. But the CRM typically doesn’t know about what happened on the website, because it’s not a CRM activity. It’s a website, or a personalization system activity. It doesn’t necessarily know what happened in the retail store, because that’s not managed by the CRM system.

Whereas the CDP pulls in all of that data from all of those sources and is specifically designed to pull in data from different sources and create those unified profiles. And if you want a system to do that, you have to design in a certain way, to store the data in a certain way and access to make the data accessible, and use a different technical design than you do in a CRM, which is really designed at the end of the day to put up a single person’s record in front of the salesperson or of a call centre agent. And not be able to deal with bulk masses of data, analogue masses of data in the way that the CDP does.

And then the third component of that definition, “accessible in other systems”, again, CRMs are primarily designed to work with the data that they themselves capture. They’re not really designed to share their data out with other systems. And some of them are pretty good at data sharing. Many of them, of course, naturally are not very good at data sharing and all of that, because that simply wasn’t sort of what the designers had in mind when they built. So looks like a pretty specific set of differences between the two classes of systems.

 

Is it fair to say that CDP is a kind of built for marketing activity, whereas CRM is more built with sales in mind? Or is that getting too simplistic about the essential differences?

It’s a little over-simplistic because the analogy would be, or the analogue would be, more marketing automation, which is to marketing what CRM is to sales. Marketing automation is specifically built to send out those marketing messages to people. And technically, they sometimes look a bit like a CDP, a bit more than a CRM looks like a CDP, because they are dealing with large bulk masses of data, people as a group, rather than one person at a time. But again, dealing with external data sources and sharing the data out, that’s not so much a sales versus marketing thing, and indeed, a lot of CDP’s support, sometimes they support sales, they’re often synchronized with the sales system, they’ll often support call centres, they’ll often support Customer Success teams, they’ll often support actually other operational teams outside of sales and marketing both. So a CDP ideally, is really an enterprise resource because everybody throughout the enterprise has good use for unified customer data.

 

One of the things I came across when I was reading through, preparing for this  nterview is an Econsultancy piece that was talking about how CDP is going to allow you to do more with real-time interactions and real-time data. Particularly for our B2B customer base, do you think real-time data works for B2B? Or do people see that as too pushy? Or, the other way of looking at is, is the B2B sales cycle too long for real-time interactions to have a significant impact, for a single moment to push things either way?

Well, real-time definitely has a large role to play in B2B in particular. Doesn’t really matter the fact of, yes, of course, there’s a longer sales cycle, but you’re still on the web, you’re interacting, you want to get a reaction in real time, you want to say send out the right copy or suggest the right content to someone based on what they’ve just done. That takes real-time processing, even though that particular web visit might be part of a much longer sales cycle. Similarly, in the call centre, you want to make sure that you’re recommending the agent that they tell the right person the right information to the right to that person at that moment, knowing the context in the larger sales cycle that they’re a part of, or you want to give this call centre agent access to the full customer history or in lead history, interaction history, which is real-time transactional, so you have to look up that data in some other system and typically actually display it alongside the CRM and others will put that into the CRM, just expose the data profile that’s built in the CDP. And that wants to be pretty close to up to date. So yeah, there are as many real-time use cases in B2B as there are in B2C.

 

Who are normally the stakeholders for a CDP? Obviously, a marketing automation system is mainly the marketing team depending on the size of the company, you may want to get the IT team and sales that are interested from a CDP point of view. Who would normally be the instigator? You mentioned enterprise companies is it mainly enterprise companies you see it in? Mid-market, and small companies, it wouldn’t be feasible for them.

There are a couple of CDP vendors who are attacking the Small Business segment. But generally speaking, small businesses are not going to be all that relevant to a CDP, or CDP is not all that relevant to the small business, because they don’t really have that many systems. A big company could have 90 different marketing systems in place, that’s not uncommon. You A small company might have a few systems, a handful, but not that many. And a lot of smaller businesses would have some sort of all-in-one system that does combines CRM, marketing automation and possibly e-commerce all in one place, in which case their data is not as fragmented, it’s all stored whatever system that is.

In terms of who buys it, still mostly marketing departments, maybe 50%, if you had to pull a number out of the air. It used to be higher. The other groups you will occasionally see are sales groups, you will see customer success teams. And part of that is definitional. Those products that are, customer success products, actually are CDPs from a technical standpoint, they’re part of the visual group of packages that we identified when we first defined the CDP category. Now you’re seeing a little more activity at the IT level, or often the Chief Data Officer or chief analytics officer will say,  “We already have my data lake, it is a lot of work getting that data from the data lake into the hands of my data scientists, who are spending an awful lot of time doing data cleaning and prep and all that.” And a CDP can act as an intermediate layer, pulling its data not from the original source system but from the data lake that itself has pulled it up from the source systems. But then so it’s a convenient source for the CDP but then doing still that profile creation. So the unification and the matching and all the fancy stuff that a CDP does to creating my customer profile. Because most of the time the data analysts or data scientists who were working with the data really do want a customer profile, if they’re working with customer data. So either they do it by hand, create their own for each project manually, which is what happens today in most cases, or they let a CDP do that and then they have a head start on getting to the good stuff, which is the real work they want to do with that data.

 

When you allude to the good stuff as such as having your hands on that data in one place for you to be able to make those decisions. So you mentioned marketeers generally there and you mentioned data analysts. So in my world, it’s about having that data so you can make those decisions, whether it’s, you want to run this campaign for longer or you want to enter this new market, maybe depending on what the data says, or you’ve got a new product you’re going to launch. I guess that’s the main benefit of having that CDP overlaying all that data?

Yeah, that’s exactly right. You know, there are lots of studies that show the data scientists and data analysts spend 40 to 60% of their time, just doing data preparation work, and the CDP automates that for them. So they don’t have to spend quite so much time, they still have to spend a little time. So this is that just pure labour savings for those guys, that makes it helpful.

 

Once you’ve got a CDP in place and you’ve got the data coming in, in the way that you need it to, how would the customer experience your marketing and your business process differently? Is the idea a more streamlined process? Whatever stage of the funnel you’re at? Or is there some other main benefit to it we need to talk about?

Well, the benefit from the customer perspective will be probably fairly subtle., What you’ve done primarily is make your marketers more productive. So your marketers will be able to draw more and more precisely targeted campaigns. So the customer might still get the same number of messages. But messages will be a little better targeted, will be a little more accurate, they’ll be a little more appropriate. They’ll be drawn on more data so they could so they are making better recommendations to that customer because they can see things that the old recommendation engine couldn’t. The old recommendation engine, could only see web transactions, the new recommendation now can see in-store transactions as well, then make better recommendations because it knows more. So the customer says “It’s  a recommendation, it doesn’t look any different to me.” But you know that it’s a better recommendation they’re getting, they’re responding a little better, maybe they’re a little happier because they see that the company is doing a better job supporting them or whatever way it supports them. So, it’s not going to change their day to day experience.

 

Have you seen any customers using a Customer Data Platform in a really good way? Because for me, I’ve not really come across many of them in the SME market. So Spotler, our main platform is a CRM, which is connected to quite a few different platforms. So in essence, a little bit like a CDP in the point that it is the central repository for our data where we can make some decisions, but it doesn’t pull in the ERP, which a CDP might pull in. We’ve got the marketing automation software going into it, we’ve got the sales using it. So we kind of use it quite well for sales and marketing and customer support. But we don’t have live chat information going in there, or monthly revenue going in as such. So have got any examples of customers that our listeners could look into, and sign up for their stuff, maybe to get the experience of uber-personalization or anything along those lines?

You know, I don’t carry that around my head. But you know, if you look at our website, at cdpinstitute.org we have 40 or 50 case histories of specific vendors, specific customers who do that. Most of the big retailers nowadays actually are going to be using it, but again, from the user’s perspective, the stuff doesn’t come with a label that says, “This data came through a CDP.” It would be nice if it did, for the CDP vendors anyhow, but the customers probably wouldn’t care.

If you look at the case histories, there are few very specific things that are classic CDP case histories. One is the one that I just mentioned about giving the call centre agents access to all the customer data, so they can go on the screen, they can see the stuff they can’t see already. So a more complete picture of what that customer is up to, which will immediately translate into better service. Okay, so that’s a classic, CDP use case.

Another classic CDP use case is to take your retargeting audiences and update them more quickly. So we’ve all been chased by the famous pair of red shoes that chase us around the internet because we were looking at red shoes or whatever shoes you look at, that’s between you and whoever. But what the CDP will do is if you were to go off and buy those shoes, it can take two or three days for that purchase to make its way through the systems to whatever system is feeding your email retargeting or your social media retargeting. And in meantime you’re getting those ads, right? So the CDP can do that in minutes. So it can update that list in minutes so that when you make the purchase, you stop getting the advertisement for the product you’ve already purchased, a very classic CDP use case.

A few things like that, that we just see coming up over and over again. And the real common thread is that it was always involved in moving data between two systems. That’s what makes it a CDP, that’s what makes it a CDP use case. It’s not that it does better analytics, although many CDPs have great analytical tools, it’s that I usually cannot combine the data from two different systems in one use case. And the use case requires that. A lot of companies can’t do it. And that’s what use the CDP lets you do that you couldn’t do before. So it’s kind of a classic, almost hallmark of the CDP use case; “Does it require multiple systems?”, “Yes”. Probably a CDP job.

 

Just to be difficult and go another way on it is, is there a risk with this system? If you’re pulling in from so many different sources, you actually end up with too much data and you end up over-analysing what your customers are doing and then you go off on the wrong track. Is that a risk? And? Or is it not something to worry about?

Well, you know, over-analysis is always a risk, you can have a tiny bit of data and over-analyse that as well. So that probably goes back to your proclivities and then it’s not the fault of the CDP. I think people do need to be careful about collecting data, not for the sake of collecting data, but because there’s some value to it, it’s easy to kind of be a data hoarder. Like just you know, it’s like storage is cheap, right? How many times have you heard that’s like having a house with a big attic, “I’ll just throw everything up in the attic? someday? You know, maybe someday I’ll use that tennis racket again and restring it?”

Data is the same way. The thing about data is it degrades, right? So a three-year-old tennis racquet? Probably okay, but a three-year-old piece of data, probably not so useful. So you do want to be a little more rigorous. So yeah, that’s just, you know, good kind of good business practices, any system is going to have the same issues. CDP might make it easier to hoard that data, but you want to be careful and be thoughtful and purposeful in how you deal with it. I don’t think most companies’ problem is not that they’re over-analysing, most companies’ problem is that they’re not analysing at all.

 

A little bit of insight would actually be useful to some companies. And David, I’ve heard the terms of omnichannel, multi-channel and CDP kind of thrown together with the kind of similar terminologies. Are they actually kind of separate things that have their own function, features and uses?

You know, there’s an entire industry that does nothing but comes up with these names. “Oh, we’re a multi-channel engagement hub!” Okay, I don’t know what that means either. Then, people who write things write papers about you know exactly what they are? What’s the difference between omnichannel engagement, digital engagement and multi-channel engagement? Beats the hell out of me, you got to read their patter. I don’t write those papers.

I already gave you my definition, you notice it was relatively concise, and we try not to use too many big words. I have a three-syllable rule. “Try not to use any word more than three syllables.” Well, I think that definition might violate it a little bit. So yeah, we don’t get too hung up on that, and certainly, our definition of CDP says you have to take data from all sources and you have to share the data out with all systems. So in that sense, it supports multi-channel or omnichannel, that’s a distinction I do understand. The use cases are multi-channel when you deal with multiple channels but they’re disconnected, uncoordinated. Omni-channel, they’re coordinated, that’s my understanding. So the CDP supports both multi-channel and omnichannel. CDPs have tools built in that do orchestration, either omnichannel or multi-channel orchestrations, across channel orchestration. So, some of them have message delivery capabilities, have email actually built in. What makes it a CDP is that it builds those unified profiles.

And then there are systems that have lots of other capabilities but also happen to be a CDP. For example, an e-commerce system that “Oh, by the way, yes, I have CDP functionality, but I’m really an e-commerce system. You know, that’s my focus. And this is just a feature for me.” So from my perspective, okay, you’re a CDP. Wherever else you are, you’re also a CDP, or you have CDP functionality. So, that’s part of the confusion about CDPs, there are so many different systems that include a CDP functionality that do other things that it’s sometimes hard to find a common denominator. And some of those systems are going to be omnichannel engagement hubs or whatever.

 

I think that’s a very important point. Because through this talk with you and I, the main common denominator that’s coming out of it is what label you give it. But in a CDP perspective, it’s having that data in one central point where you can make some insight and actions on it. Whatever you want to label it; CRM, automation, whatever. But a CDP gives you that central depository, and then you can make those actions and hopefully use those channels based on that that knowledge, which I think quite a few marketers would be better off doing, rather than some of the kind of blast sending marketing tactics that go out there, or at least engaging that CDP data.

So with that, Dave, would you do you recommend different types of tactics based on the traditional marketing funnel? I guess that’s what that data gives you. So for example, if you’ve got a load of web visits, coming to your website that you can see in your platform, you might have a certain level of content, which you call “top of the funnel” content, whereas if you can see there are people on your website, and they’ve been calling your sales line or your contact centre, as you were saying you might have a certain kind of content that brings them down the funnel a little bit more, that tries to show their behavioural intent or a buyer intent. Is that the kind of thing that you could use that data for?

You can certainly use that data for that. It can, the nice thing about it is it will give you a more complete view of that customer, so you’ll have a clear picture of where they sit in the funnel. And that in turn will allow you to send more appropriate content to them, messaging or whatever, whatever treatment you’re doing with them. Again, not particularly a need for a CDP there. That’s good marketing, that you should be doing, hopefully, the CDP just lets you do a little more easily and effectively. But the CDP is not this magic elixir that you drink it and suddenly you’re like Super-Marketer. You’ve still got to know how to do marketing,  it’s not really going to solve that problem for you.

 

No, I guess it won’t. You’ve written some pieces on predictive profiling. And I just wanted to see your thoughts on how having all that data will help you do that predictive profiling and lead scoring a little bit more accurately. At least, it gives you more options to add scores or take away scores, based on behaviour, interactions and inputs.

That’s right. I mean, anything that involves any kind of predictive analysis, whether it’s automated or just somebody looking at the data, that will be more effective when there’s more data to look at more data points, more up to date, more current data points. And CDPs generally will ingest data in real time. If it is available to be ingested in real time, of course, many sources can’t provide it in real time. but if the source can provide it to the CDP, you can get it in there and available to you. So as a result, the better the data, the better the prediction. The quality of data is by far the biggest predictor of the quality of your predictions. Much better than which algorithms you use or anything like that, It really is 90%-85% just the data itself. So at least marginally better data makes marginally better predictions. Some of the CDPs have their own predictive tools built in. So if you go from a company that has no predictive tools and limited data to a company that has unified data and predictive tools, of course, that’s going to be a bigger leap, a big improvement. And that’s what you get by buying a CDP. Some of the CDPs do actually have some pretty advanced orchestration capabilities. So beyond simply creating individual models, some of them can actually kind of manage that customer lifecycle in the funnel, and automatically move people from one status to the next and send the right message for that particular status. At the funnel stage, they’ve entered into some of them, even now, or at least they claim, can automate the design of those things. So they can figure out what the final stages are and figure out what the right content is, as opposed to actually hitting the rules. So some of them get quite powerful.

 

What level of adoption are we at generally? I’ve got to be honest, CDP is only a term I’ve started hearing probably in the last 12 months or so. And obviously CRMs are, not popular knowledge, but people are a lot more familiar with them. Are business that are picking up CDPs still early adopters, for a lot of industries? I know you mentioned the retail sector is pretty developed.

Well, I think we’re past early adoption. You’ll see all kinds of different numbers and different surveys, and the problem is usually that the survey was done by a CDP vendor of their own house files. And guess what, most of those people have heard of a CDP or they wouldn’t be on that list! We survey our membership at the CDP Institute. But it’s the same thing, obviously, these are people who are more tuned to CDPs than average.

Every now and then we’ll see a survey that was not really based on CDP in one way or another. And those tend to come in around over 20-25% penetration. Getting into definitions, “Do you know what a CDP is?” The people are answering the questions. There was a famous survey where 50% of the people said they were using the Salesforce CDP, before Salesforce even launched the CDP. Maybe we’ll take that with a grain of salt. So there are those issues. But realistically, we think penetration is around 20%. Higher in retail for sure. Higher in big business than small for sure. Yeah, sometimes you can just see relative levels of penetration, which are going to be accurate, regardless of the issues with the audience prior. Really small business taking under 3 million revenue, much less common just to see and they don’t have the skills or the resources, the technical staff to deploy or manage one of those things, or the marketing sophistication to get the incremental benefit again, they usually have simpler things they need to work out and get straight before they get the incremental benefit of that better data that a CDP will give them.

 

Okay, I was about to follow up on that and say, if a business is listening to this, and says “Oh, we really need to have a CDP.” What kind of questions do they need to be asking, what kind of groundwork do they need to do before they dive in? Sounds like you’ve got to consider what data sources are particularly useful for what you do, but what else do you need to think about before you start this journey?

 

Well, it starts with what do I want to do? So “What can’t I do and why can’t I do it, and will a CDP solve my problems?” is the process to go through. So “What do I want to do? What are the marketing programs or the business activities”, because it’s not just marketing, “that I really feel would be of great value?” “What are the big pain points? The big things that I really want to do that I can’t do today?” Okay, so ask yourself those questions. Then ask yourself, “Well, what’s stopping me from doing that? And is that a problem that a CDP would solve?” If what’s stopping me is I can’t combine data from multiple sources to use together, ok, that’s what a CDP does. If what’s stopping you is “I don’t have a predictive modelling tool.” Yeah, the CDP might provide that but you can also buy one. If that’s your only problem, if you already have your data assembled, if it exists all in one system to begin with. If the problem is my staff doesn’t know what the heck they’re doing, you know, train them for God’s sake. And maybe it’ll turn out after the training that they need a CDP, but a CDP is not going to train your staff. So you have to understand what problem we’re going to solve. And again, the particular one that’s most likely to be a CDP problem is one that involves multiple data sources coming together. So if you have a lot, a lot of things you can’t do, because you’re missing that, that’s pretty much the indication that a CFP should be on your list.

 

I think that’s a really good point to wrap this up on. We’ve given people plenty to think about, explored what these platforms are, what they can do what they can’t do. So that’s probably a great place to say thank you very much for being on the podcast and we’ll leave it at that.

Okay, great to catch up!