A: There does seem to still be a little bit of confusion, which of course makes me very sad because I’ve spent the last four years of my life trying to get rid of that confusion. Our definition of a CDP is: packaged software that builds a unified persistent customer database which is accessible to other systems.
In somewhat plainer English, it means you pull together all of your customer data and create these unified customer profiles. Then you share them with whatever other system needs them. The core idea is that you have all this data, and you want to get it in one place so you can have a complete view of each customer. That’s really what a CDP does.
Q: Is the marketing group the key buyer of the CDP in an organisation?
A: Sometimes it is, but traditionally and in most cases I would say marketing is still the main buyer of the CDP, it was the case in the past. Now we’re seeing other departments in the company sometimes take the lead; it could be a customer service department that needs a complete view. But more likely, it’s central, like a data group or analytics group or even the IT group, which have been tasked with building this unified picture of their customers.
Of course, it’s useful throughout the company. Once the company realises a CDP has got uses outside of marketing, then the actual project management tends to migrate out of marketing as well, so the people who build it can share it with everyone else.
Q: Are all CDP offerings the same, what are some key differences?
A: If I said they were the same, this would be a very short interview, they’re not all the same. There are about 130 CDPs in the world we know about, and certainly no two are alike; they tend to fall into clusters. There are some CDPs which meet the definition of building a unified database, then they stop there, and they make it available to whatever system wants to use them.
There are CDPs which do this, plus provide analytical capabilities, predictive modelling, typically or sometimes marketing, measurement, or attribution features. They build the database, plus they do the analytics.
For the bulk of the market, we estimate about 70% of the industry are CDPs that both build databases and do some flavour of campaign management, customer engagement or personalisation. But they give the marketers a tool to do something with the data, not simply create the profiles other systems can use.
There’s some you can go further with, some would follow an email engine or web content management to deliver messages, not select them as those campaign CDPs do. Some are CDPs baked into an ecommerce system, an entirely separate platform that happens to have the CDP inside. It’s quite a variety, which is a good thing because it means you can find one to fit your needs.
Q: Most major marketing cloud vendors now have a CDP (Salesforce, Adobe, Oracle). Has this impacted the market?
A: Oddly enough, it hasn’t had much impact yet. We certainly expect it will, but those products are all relatively new, Oracle’s is about two years old, Adobe is about a year old. Salesforce is still something of a work in progress, although Salesforce had bought a couple of other companies with CDPs, depending on how you look at it, you can say they’ve been around as well. So far, these are still relatively immature and they are of interest primarily to the companies already using the suites of those vendors.
The independent CDP vendors still have a pretty good chance of selling in environments where the company doesn’t use Oracle, Salesforce or Adobe, which is the vast majority of companies. There are very few companies that will only use one of those vendors. As long as they’re using a couple of the vendors, there’s at least a little bit of room for a non-suite CDP, or an independent CDP to sneak in there and be neutral across all the other vendors. We expect eventually the big vendors will take a larger role, but so far, it’s been somewhat limited.
Q: Do you think the infrastructure the CDP vendor chooses is important?
A: The technical infrastructure makes a great deal of difference. It drives what the CDP can do now because our definition of CDP says you have to be able to serve all different kinds of data and you have to keep all the details. That is a fairly stiff set of technical requirements. There are only a few kinds of technologies that can meet those requirements, almost every CDP will use some sort of a NoSQL database, for example.
You really can’t strictly do what a CDP needs to do if you’re going to use an Oracle, SQL server or one of the relational databases. That being said, some companies call themselves CDPs that don’t meet our definition and have these more limited capabilities because they’re built typically to service the tools developed by a particular vendor, not really to share with everyone else.
Those systems might use a more limited technique, or they might use a relational database, for example. But if you’re going to use a relational database, it’s really hard to keep web browser logs, for instance, or to keep social media post history, the actual text itself, it doesn’t work that well in those kinds of databases. You’re almost always forced to use something else.
There’s a lot of databases now which do that, Snowflake is a popular one, Elasticsearch or even Redshift, which is a SQL database but it’s highly scalable. MongoDB is a document database, there are some different data stores we’ll see being used by the CDP vendors, but it’s almost always something else, it’s not simply a relational database.
Q: With the changes web browsers are making to cookie access and Apple is making to the device advertising ID, what effect does this have on the CDP?
A: By large, it’s a good thing for the CDP vendors because it makes it harder to accumulate third-party data, both with the cookies and with the Apple ID situation. Therefore, it makes companies pay more attention to their first-party data. CDPs are mostly about first-party data, they’re mostly about data on your actual customers. It’s been a selling point for CDP vendors to say: “hey, you want to do more through first-party data? Well, you need a CDP to do that”. Interestingly, I did see one analysis the other day where somebody said, ‘a lot of CDPs collect their web data through a third party cookie’, which is a true statement.
Collecting data is one aspect of what a CDP does, it’s only 5% of the value if you had to pick a number. The CDP will have to find some other way to get their hands on the web data. But there’s still plenty of other things a CDP does, it’s still going to add value. All this is not going to harm CDP use directly, and it will help them indirectly, quite a bit.
Q: How do CDP platforms enhance the customer experience?
A: The CDP’s primary mission is to pull together unified customer profiles. When the customer comes to the website, the call centre, the retail store, the mobile app, wherever, they’re interacting with you. Whatever that system is, it can call back to the CDP and know everything about the customer across all channels they’ve interacted with. It’s not that they get into the call centre, and the call centre agent can only look up the history of that customer in the call centre. They don’t know what the customer did on the web, because the call centre system won’t know that.
The CDP can share information with the call centre system, they get a complete view of the customer, which means the agent now can react to the customer with full knowledge and give them a better experience. We have seen a lot of research demonstrating people want companies to recognise them, understand them, and treat them consistently across channels.
They don’t necessarily want to get personalised advertising, because no one in history has raised their hand and said, “hey, send me more advertising”. But they do want you to give them better treatment in terms of appropriate discounts, or offers relating to the products they’ve purchased. Other kinds of important treatment include making it simple to do returns, customers want you to know about them. That’s what the CDP makes it possible for the company to give them.
Q: What are some untapped opportunities marketers have when implementing CDPs that they might not otherwise have thought of?
A: Marketers are very creative people. They think of many things, some of which you can talk about in public, some of which you can’t. In terms of CDPs, what marketers might not necessarily think about are some of the cross channel things. CDPs traditionally work in cross channels. But to be more advanced in predictive modelling, for example, and doing more advanced segmentation based on the predictive modelling, this is something often marketers don’t even realise could be done with those kinds of models, because they’re focused on getting the job done and doing relatively simple things.
There is a very interesting advanced technology out there that the CDP makes possible, because it assembles all this data, lets you do things and see relationships that you couldn’t see before, and therefore maybe offer some new services, for example. To take advantage of this cross channel view with customers, there’s quite a bit which can be done, but most marketers have a long list of things they’d like to do. Again, they’re creative folks, their problem is rarely they can’t think of what they might like to do next.
Q: There has been a massive growth of CDPs in the past few years, why do you think this is? Will this growth last?
A: The growth has been driven by the proliferation of channels and independent disconnected channels. As customers demand a connected experience, the company is forced to buy a CDP to connect those channels, because that’s what CDPs do, it is very difficult to do without a CDP. It’s been very much demand-driven, very much customer-driven, which is not always the way in marketing technology. What happens is some technologist comes up with a clever idea, “hey, we can do virtual reality tours that are on some crazy new social media channels you never thought of before”. They can have colour pictures that read your mind and do whatever cool thing they think of. Then somebody builds it in some market and they buy it, but the customers scratch their heads and say: “I wanted you to process merchants more efficiently”.
The customer is generally less excited about the technology than the marketers are. CDP is the other way around, CDP is the customer saying: “get all my data in one place, give me better treatment, understand what I’m doing, do what I want you to do”. That’s what companies have been forced to do, that’s part of the reason the big vendors like Salesforce, Adobe and Oracle were relatively late to market on this, because they didn’t see the need. Their customers saw the need and their customers’ customers saw the need, but it wasn’t so clear to them. They didn’t rush out and do it. It took them a long time to understand this was something serious their clients needed to serve their clients’ customers. Then they eventually came around and gave their clients what they wanted.
Q: What data trends have you seen over the past year that have shocked you, in either a good way or a bad way?
A: When it comes to data, in particular something that has very pleasantly surprised me, has been the increasing interest in privacy. For years, we’ve heard about privacy. There were this little coterie of privacy fanatics who were concerned about it and nobody else cared. The people who cared about it would make a lot of noise and they would get some people in the government excited. Nothing would ever happen because the industry would, I wouldn’t say paid anybody off, but they had influence.
In the last few years, privacy advocates have been much more effective. I’m not sure it’s because the privacy advocate got any smarter, however, consumers got a lot more worried about privacy.
Frankly, they saw a lot of data presented to them, about them, which they hadn’t intended to share, and it scared them. They worried about what data was being collected about themselves, and about their children. They realised this data was being used in many ways; some good, some not so good. There’s a substantial growth in actual mass consumer interest in privacy, no longer the relatively small number of advocates.
This led to considerable regulation, obviously GDPR in Europe, CCPA in California, there’re some in Australia as well. All of a sudden, there is some fairly serious concern about privacy and, of course, more recently, with Facebook and Google. Australia passed a very good law regarding this for the news channels. It’s all part of people pushing back against the companies that do suck up a lot of their data. I tend to think this is a good thing, but it may not be the easiest thing for marketers to deal with. But it’s a good thing overall, it’s been a pleasant surprise,
Q: As we step into a post-cookieless era, what are some of the myths about data you want to see disappear?
A: I want to see the mystique disappear around the thought you can’t do anything without third party cookies. Even though I’ve been saying it for quite some time now, the rest of the world has no choice but to deal with this reality. It’s weird, everybody’s been saying “oh, cookies are going away, the world is gonna end”. I thought, well human beings have been on this planet for a long time and they were doing marketing for most of that long time, they didn’t have cookies until maybe 20 years ago. Somehow, the world of marketing survived before there were cookies, and no doubt the world of marketing will survive after cookies.
If there’s a myth, it’s related to the notion that with cookies, and with other kinds of data collection techniques, we can predict with great accuracy what customers will do. It’s absolute nonsense. At one point in my checkered career, I ran a predictive modelling team, and the models are 80-90% more accurate than throwing a dart at a dartboard, but they’re not 100% accurate. The more complicated the behaviour, the less accurate they get. It’s not like we had all this absolute precision we are suddenly losing, we were never that good at actual predictions.
People need to understand there’s plenty of data out there, still. There’s plenty of tools to be used for modelling. Some of it is more effective data for modelling than cookies ever were, as cookies had a lot of issues anyhow. But we have to realise you only go so far, you have to listen to your customers and not try to figure out what they’re going to do as much as seeing what they’re doing and react to it. You really can’t force them to do things too much.
There are some exceptions around this, and some of them are pretty scary, like some of the kinds of things social media can do shape people’s behaviour. But even then, they’re taking people’s tendencies and exaggerating them, you’re not going to make me do something I have absolutely no interest in doing because I’m on social media. Social media exaggerates and moves people in more extreme directions than they move in otherwise, but even social media is not the root of all ill.
