In 2019, the Customer Data Platform Institute launched its RealCDP initiative, intended to help marketers understand which vendors in a landscape of over 100 so-called CDPs are actually CDPs, capable of fully satisfying the broad range of critical CDP use cases.
In order to be a “RealCDP,” the Institute states that a vendor must:
- Ingest data from any source
- Capture full detail of ingested data
- Store ingested data indefinitely (subject to privacy constraints)
- Create unified profiles of identified individuals
- Share data with any system that needs it
While this list of features is still generously inclusive of many CDP vendors (and of course AgilOne has RealCDP certification as well), I’d like to take a moment to drill a little deeper into requirement #3: “store ingested data.” While this may seem like an obvious feature for a CDP, understanding the exact way a CDP vendor stores data reveals a lot about how well a vendor can meet specific CDP use cases. Analytical use cases, identity stitching use cases and use cases that require a lot of processing power are a few that come to mind…
Data Resides in a Centralized Location
Enterprise CDPs provide their own data warehouse, which is a central repository of data for the CDP after identity resolution, predictive intelligence and unique configurations for a specific business take place on that data.
CDPs that shuffle data around (e.g., tag management systems) based on data access patterns present fundamental problems and huge limitations for businesses that are serious about understanding their customers and activating customer data for orchestrated engagement. With no persistence of cross-silo data, there can be no master customer record -- and without that, many “customer-first” initiatives fall flat on their face.
Enabling Powerful Identity Resolution
Without a centralized data store, identity resolution is stifled. If identifiers are pulled across each application, but not reconciled into a persistent stored record, there can be no elements of probabilistic matching or other intelligence that stitches first and third-party data into complete, single profiles. At best, identity resolution would be relegated to simple device stitching -- not enterprise-grade true identity resolution.
By contrast, AgilOne and other mature CDPs were designed under the premise that no customer insight or activation can take place until the marketer has a true understanding of their clean, deduped, stitched customer records (including identities and events) into a “golden profile,” which serves as the basis of all customer marketing and engagement. Furthermore, identity resolution works best when it’s configured to meet the exact needs of the organization -- and such configurability is only possible with an enterprise CDP.
There is also the issue of efficiency (or lack thereof). A non-centralized data model requires scoring each customer’s record, their browsing history, their purchasing history, and their campaign engagement — all of which reside in separate non-centralized databases — using deterministic and probabilistic matching algorithms against that population every single time the scoring query is submitted. Inefficient, right? Wouldn’t you want to do the record matching one time instead of every single time?