The CDP question European operators stopped asking, and the one they started
The category that was sold as a foundation and rediscovered as a feature
Customer data platforms had their commercial peak in Europe between 2019 and 2022. Twilio Segment, after the 2020 acquisition, sold the platform as the indispensable foundation of any serious customer data architecture. Tealium, mParticle, and BlueConic occupied the more specialist tiers. Hightouch and Census reframed the reverse-ETL conversation. The pitch into European marketing teams was that a unified customer profile was a strategic asset, and that without a CDP the data lived in twenty disconnected places and was structurally untrustworthy.
The pitch was partially correct. The implementations that delivered on it were a minority. Forrester's 2024 survey of European CDP buyers reported that thirty-eight percent of implementations were rated by their own buyers as not having delivered the promised unified profile two years after go-live. The reasons were familiar to anyone who had implemented Salesforce or SAP: data quality was worse than assumed, the use cases that justified the spend did not survive contact with engineering reality, and the team that operated the platform did not have the cross-functional authority to enforce identity resolution at the points of data entry.
The other thing that changed underneath the category was the third-party cookie. Chrome's full deprecation in 2025 did not produce the apocalypse some vendors had warned about, but it did produce a clearer picture of which data sources a CDP could actually unify and which it could not. The platforms that had built their value proposition around stitching together cookie-based behavioural data discovered that the stitching they relied on was getting less reliable each quarter.
How the post-cookie European stack looks
A European mid-market team rebuilding its customer data architecture in 2026 is doing a quieter version of the work than the 2020 cohort. Twilio Segment remains the default first choice for companies under €100M ARR for the same reasons it was in 2020. The product is mature, the SDK story is clean, the EU data residency posture is documented, and the developer ergonomics are still better than the alternatives. The buyer's expectations are lower than they were. A CDP is a data router and an identity resolver. It is not a strategy.
Tealium continues to land in the European retail and travel mid-market accounts that need the consent management integration deeply baked in. BlueConic, which has a Dutch parent and an explicit EU data residency story, continues to win in regulated industry accounts where the procurement question about data location has a binary answer and an American vendor cannot give the right one without elaborate qualifications.
The more interesting structural change is reverse ETL. Hightouch and Census reframed customer data activation around the data warehouse rather than around a CDP. The architectural argument is that if the warehouse already contains the cleanest version of the customer record, the activation layer should read from there directly rather than maintaining a parallel store inside a CDP. The arithmetic on this for a European mid-market team is favourable. The team is already paying for Snowflake or BigQuery and already has dbt models that define the customer entity. A reverse ETL layer on top of the warehouse costs a fraction of a full CDP licence and avoids the second source-of-truth problem that broke many of the 2020 implementations.
The EU specific reading
The European specifics on this shift are three. The Schrems II compliance posture for any tool processing customer-level identifiers is harder for US-incorporated CDPs than the marketing materials suggest. The data residency question for behavioural and identity data has tightened across DPAs since 2024, and a vendor that cannot promise the data does not leave the EU is increasingly losing renewals. The cookie deprecation has shifted the data sources that matter, and the platforms designed around third-party cookie stitching have had to rebuild their identity layer in a way that often surfaces capabilities the buyer did not realise they were paying for.
Real examples make this concrete. A €150M GMV French fashion retailer moved off a US-headquartered CDP to a Hightouch-plus-Snowflake architecture over nine months in 2024. The licence saving was €280k annually. The team's reported view was that the architecture is simpler, the data lineage is clearer, and the procurement conversation with the DPO is easier. A Berlin-based subscription business at €40M ARR stayed on Twilio Segment but moved its consent management to a European vendor and rewrote its identity resolution to run inside the warehouse rather than inside Segment. The cost did not change much. The compliance posture became defensible in a way it had not been before.
What the productive question now is
The question that produced bad answers in 2020 was "which CDP should we buy." The question that produces better answers in 2026 is "where does our customer data already live, and what is the smallest router or activation layer we need to put on top of it." For most European mid-market teams the answer is now closer to a warehouse-centric architecture with a reverse-ETL tool than to a full CDP procurement. The teams that have done the rebuild quietly are operating with less software, less vendor risk, and a clearer story to tell their DPO. The category will continue to exist. Its shape inside European mid-market accounts is converging on something more modest and more defensible than the 2020 pitch deck promised.