CDP Myths #3: ID Stitching for a "Golden 360-Customer View"
As before, I wrote this message without any LLM input.
In CDP RFI/Ps, enterprises usually ask for the sophistication of ID stitching capabilities. Here's some thinking on how to ask vendors the right questions on ID stitching:
The CDP value equation is simple: Unified data = better data = better decisions = better commercial outcomes.
The foundation for better data is ID Stitching. Some call it ID Resolution, but that's more of a legacy term used in "CRM-Onboarding" land.
ID Stitching means that any identifier per customer (email, phone, MAID, cookies, etc.) is stitched (= unified, consolidated) to a distinct profile of a customer. There are a couple of things to stress-test when selecting a CDP vendor:
1. How are non-logged-in customers treated? Is the CDP vendor able to differentiate between different cookie types? How about 3rd party cookies? Once the user logs, is the cookie-based or MAID-based profile stitched to an existing persistent profile, e.g. based on an email-to-MAID/cookie link?
2. Deterministic vs. probabilistic stitching. Probabilistically assuming that a certain email is the same person as a certain MAID (mobile inApp ID) or cookie can be risky in strict privacy environments, such as GDPR, as you end up stitching two people into one profile, where one of them may not have given consent for a certain action.
3. Does the CDP allow you to flexibly set primary and secondary IDs for stitching? Is it possible to define priorities of stitching?
4. How are households treated? Is the CDP vendor able to detect that two members are part of a shared household? They might share the same cookie (same laptop), but two different emails. Are these IDs merged into one profile by default or is the CDP's ID stitching advanced enough to allow the creation of two distinct profiles? In retail and insurance, certain products, such as white goods (washing machines) are most likely only bought once per household. Hence, the effective use of suppression campaigns requires identifying household users to suppress the entire household from promoting a product for customer acquisition campaigns. Otherwise, you end up wasting your budget on folks who will not buy.
5) Are there limitations in how many data attributes (events, behavioural data etc.) can be stitched to a profile? Some Marketing Cloud CDPs have limitations here, while some industries have more than 400 attributes per profile and need flexibility. Think of football clubs.
In fancy marketing terms, a properly unified profile is often called a "360-customer" view or "golden record". Make sure your CDP actually has the power under the hood to deliver that.
Projjol Banerjea Sathish K S Pritika Shetty Harsha Korasala Peter Mielke Chang-Won Hwang Mitch (Mircea) Gavril Robert Smart Reetesh Nimje Christine Yeh Javier M. Mohamed Ali
Great points on how consent driven marketing builds long term customer loyalty. In a world where privacy is a top concern, this approach will make a huge difference for brands