There are lot of tools to gather and utilize data. There are still CRMs and data warehouses. These tools don’t sounds like real time data, dynamic data, digital footprints, or data activation in RT, do they? The turf battle is between audience data technologies now, between CDP and DMP. Even if ~50% of functionality could be duplicated in these two tools.
What the buzzword CDP means?
CDP stands for Customer Data Platform. There is a word “customer” inside and we should stick by it. It is a tool used for collecting and matching data from different sources into a single pool to create create a 360-degree view of the customer followed by personalized customer experiences. The unified customer data are granular ones, it means you can pull a single user out of the pool and built targeted experience for this particular user.
The main data source for CDP are CRMs and transactional systems as these strongly rely on PIIs. These data could be extended by website interactions or app interactions, as long as users’ identifiers can be swapped and matched between platforms. Do we want collect offline data there? No problem at all, you can add your offline sources as well. Can we add some external – 2P / 3P data there? No problem, as long as we can adjust identifiers.
CDP as an answer for cookieless world?
It is slightly confusing why CDPs gained attention and were named as the answer for 3P cookies problem. They are addressing a different segment of the market. Most technologies are able to step into advertising world only thanks to 3P cookies (sic), the same they were killing a few paragraphs earlier. Many vendors use a 3P cookie delivered by AppNexus in their identity frameworks. This means that you can identify a user on media channels as long as 3P cookies are alive. Implementing that kind of technology a few months before the change could be a big fail.
Are insights from CDPs valuable? Of course they are! Any unified data among teams, departments can give more value for every organization. It is valuable for marketing teams, sales teams, analytics teams. Any investment made till date in audience data technologies will not be wasted as there are still multiple sources of audience data, that should be consolidated on your side.
What is and what isn’t a CDP?
There are many cases of dressed up technologies as CDPs. There are CRMs dressed up (Microsoft Dynamic 365), analytics platforms (Piwik Pro), marketing automation technologies (Salesmanago, Bloomreach) and many others. They are as good at syndicating data to external partners as sending a database on a floppy disc to another vendor, what, by the way many clients and agencies still do with Google and Facebook, except this floppy ;).
There are finally ETL and connector providers, that can call themselves as CDP and there is a big element of truth in it. Some top CDPs claim to help you resolve users identity. In my opinion that is a clear and important differentiator that tells the difference between a real CDP and a CRM or any other marketing platform. Only technologies that could address a customer with different identities, without AppNexus 3P cookie, and finally not only based on client’s email address, should survive in the marketplace.
CDP is a quite complex issue. You should select it carefully and deliberately otherwise you can end up with just a half of technology you need. You can contact us if you need any support.
And what is a DMP?
DMP – data management platform was another buzz word few years ago. Similarly to CDPs, many vendors used to call their technologies DMPs. Mostly vendors, who provide data for sale. Therefore still many people links the term DMP with 3P data. That is half true. Yes, many vendors came from that part of the world but they have offered technologies to collect and manage 1P data on clients’ side as they did it with 3P data. These technologies were set firmly in the advertising marketplace and were able to swap identifiers with all major media vendors and programmatic open market, based on 3P cookies of course. Many publications mix up 3P data with 3P cookies (Microsoft, Treasure Data, Segment). No, basing on 3P cookies (what is still required by the programmatic world) doesn’t make you a 3P data management technology
Later on all major players developed identity management tools (identity graphs) that swap external identifiers with internal ones (client number, loyalty card number, 1P cookie, fingerprint, etc) joining together own media with media channels.
How DMP works?
DMPs aggregate anonymous users data originating from multiple sources, from the website, apps, campaigns. DMPs must work with anonymous IDs, like cookies, device IDs to exchange information (anonymously) about audiences with the advertising world. All personal identifiers are not allowed to be kept in the platforms to protect users’ privacy. It is a huge advantage of DMPs. It make the solution more GDPR compliant, especially for clients who do not use any cloud based system to collect personal identifiers. In this case a DMP could be a bridge between 1P data with PII (on premise CRM), 1P data behavioral data without PII, to resolve and swap all identifiers of customers and users.
As mentioned above, client’s first party behavioral data can be extended by other 1P sources – CRM, transactional system, marketing automation, and by 3P (public) and 2P (partners’) data to get even better view of our own users and customers. These data, mixed or not mixed up together, can be used for targetting, in most cases in real time – for retargetting, for personalization, for A/B testing among others.
A DMP stores data in two places. There is a behavioral pool and swap pool – a complex solution to swap identifiers with all advertising vendors in real time.
Difference between DMP and CDP
The main difference between DMP and CDP is in granularity of data collected. In CDP you should be able to get a single grain, a single user and get his/her 360° view. DMP collects data in buckets. There is no sense to target a single user in media channels, it is neither allowed by vendors, nor by regulations. It violates the privacy of users. Therefore, you have a taxonomy tree there to easily access users.
Resolving users’ identifiers is a basic functionality of a DMP platform as well. It used to base on 3P cookies, but over time identity graphs were developed to swap more IDs in real time. The next step for DMPs is to adopt other methods of swapping identifiers.
An example of data you can access in CDP and DMP
A user bought e.g. a training suit, size 152 on your site.
In a CDP you have stored the following information: email address, phone number, first name, surname, delivery address, zip code, purchased item name, sku, product category, value, date of transaction, last transaction data (purchased product, amount, value, date). Like in SQL, you can retrieve all users who bought that product, retrieve their email addresses and send an email campaign to them. Probably you did it before manually, CDP automates it a bit.
In a DMP you can conclude from this single event that the user has a teen kid, even assume a gender. Probably you already know some geo-related data (country, region, city). A single event activates many rules that pushed data into separate buckets (taxonomy nodes). You will have family status, child gender, child age, country, region, city, a status of the transaction, transaction value (range), purchased product category, purchased product value and a hashed email, a hashed phone number (to match data with the CRM or swap IDs across devices). Clicking a taxonomy node purchased product “training suit” and “purchased product value 100-200€” and “within last 30 days” you will get a segment you can use for targetting. But you will not retrieve any information you had not categorized before (e.g colour of the suit), exact amount of the transaction (although it depends on the granularity of your taxonomy) or date of the transaction (DMP rather uses data recency like “within last 30 days”). It means that if you have 2 transactions of 2 products categories within 30 days – all data will be mixed up. You would rather not retrieve specific product price purchased for the user lands in two taxonomy nodes, e.g. 100-200€ and 200-300€ and you never knows how these nodes are related with others.
Having both a CDP and a DMP on a board can bring some additional values for every marketer – ease of targetting in mass media from one side and sequential data selection from the other.
Final thoughts. CDP or DMP?
Both CDPs and DMPs are audience data technologies and could merge together in the near future. Fading up of 3P cookies does not mean you should change from DMP to CDP. These technologies still complement each other. But if you have to select just one, the answer depends on your data strategy, what I’m going to cover in the next episode of “Cookieless future”.