The Rise of Customer Data Platforms
17 Nov 2021
I hear you say, all the key players in the Marketing Automation space are going to market with their own CDP. That is true, but what they are really selling is a set of tools that allow construction of a custom CDP that meets your needs. That’s where having an implementation partner that knows data better than they know their children’s faces will come in rather handy.
So what does a CDP project look like?
It will vary greatly depending on your endpoints, where the data will be heading once it’s combined in the CDP into elegant new forms. But whatever the technology you are using, it will typically follow these distinct phases (and assumes that provisioning will be handled by your CDP vendor).
Phase 1. Workshops
Kick off phase to identify all the key stakeholders, data sources/integrations, user roles within your business and what the requirements actually are for the new CDP system you will be assembling.
Phase 2. Data Ingest
The first hands on phase, intended to setup and configure the data sources, and define the keys. This can be very simple if the vendor supports all your data sources right away, or very hard if custom integration development is involved, which we always recommend to avoid wherever possible.
Phase 3. Data Mapping
Identify the relationships between the different tables, based on the keys you have previously described to the system. Expect there to be a lot of back and forward to the previous phase as additional requirements and dependencies become clear.
Phase 4. Identity resolution
Once all the data has been defined, mapped and imported, it is time to look at defining your rules to manage and dedupe the data. Remember, it comes from diverse sources, may not all be normalised (one record per customer), and almost certainly includes some crossover, perhaps some of it contradictory – what do you do when two data sources contain the same customer but with different field information – which one should you prefer?
Phase 5. Define Destination channels
Once you are reasonably happy that you have a draft CDP, consuming data from diverse sources, and some semblance of order imposed onto it, then it is time to configure connections to recipient/consumer systems, where you will be sending your newly created synchronised views of the customer.
Phase 6. (optional) Segmentation design
Some of the more advanced CDPs actually provide a capability to develop segmentation ready for syndication to the downstream systems, which can be extremely useful if you have multiple destination systems that could consume the same target segments, saving you from unnecessary duplication of effort, and possible inconsistencies of interpretation.
Phase 7. Reporting & Analytics
The final phase of this (and many) engagements is to configure your reporting requirements to provide correct levels of Management and Quality reporting in order to ensure that all stakeholders have a clear, and real-time view of the health and accuracy of the CDP.
How long can it take to implement?
That is a tricky question to answer with so many variables, not to mention variations in approach from different vendors. As a ballpark, a lapsed 3-6 months should be allowed depending on complexity, assuming the involvement of suitably qualified Technical Architect and Business Analyst resources. For every custom integration that needs to be developed, add on another chunk of time to allow that to take shape.