CustomerLake adoption accelerates
Databricks CustomerLake is not another packaged CDP. It is a re-framing of where customer data lives. When the warehouse becomes the canonical store for identity, profiles, and events, every layer above it gets rebuilt around that gravity.
Early adopters report ~40% reduction in CDP spend within twelve months — not because the CDP failed, but because the workloads it owned (identity stitching, audience modeling, activation) collapsed back into the lakehouse where the data already lived.
The architectural consequence is bigger than the dollar number. Once the warehouse is the source of truth, identity resolution moves upstream, activation gets unbundled, and the packaged CDP stops being the integration hub.
Decisioning becomes the only premium real-estate left. Storage, transport, and activation commoditize toward the warehouse. What you decide to do with the data is the remaining moat.
How this propagates through the architecture
- Warehouse-native vendorsDatabricks, Snowflake, Hightouch, Census compound their gravity.
- Composable architecturesTeams already running reverse-ETL gain a clean upgrade path.
- Modern data teamsData engineering becomes the customer growth control plane.
- Enterprise
- Retail
- Financial Services
- B2C
- Marketplace
- Media
- Travel
What this changes about the optimal architecture
We believe the packaged CDP category is being unbundled in public. CustomerLake is the most credible warehouse-native challenger because the data is already there — it does not need to be moved, modeled, or duplicated to be useful.
For teams renewing a packaged CDP in the next 12 months, the right question is no longer 'which CDP?' but 'what should the warehouse own, and what gets composed on top of it?' That is a different architecture conversation, and most procurement processes are not asking it.
The optimal architecture in 2026 has the warehouse as the system of record, an activation layer above it (Hightouch / Census / native), and a decisioning layer above that. The packaged CDP becomes optional rather than central.
Every source, transparently cited
- Vendor DocumentationDatabricks CustomerLake announcementPublished Mar 27, 2026 · Updated 21 hours agoTrust92
- Analyst ReportHightouch — Warehouse-Native State of the StackPublished Apr 02, 2026 · Verified yesterdayTrust84
- NewsletterChiefmartec — CDP unbundlingPublished Apr 15, 2026 · Updated 22 hours agoTrust78
- Community DiscussionMeasureCamp — Reverse-ETL replacing CDPPublished May 01, 2026 · Updated this weekTrust64
- Customer InterviewFortune 500 retailer architecture reviewPublished May 22, 2026 · Verified yesterdayTrust88
- Customer architectures143
- Public sources47
- Vendor announcements12
- Community discussions61
- Product launches4
This perspective evolves as evidence arrives
- v3Incorporated 143rd customer architecture. Confidence 86 → 91.Jun 26, 2026 · 1 day ago
- v2Added Hightouch warehouse-native survey. Confidence 78 → 86.Apr 30, 2026 · 2 months ago
- v1Initial publication.Mar 27, 2026 · 3 months ago
Personalize this perspective
Add your current stack to Architecture Studio. Valence will tell you exactly how this perspective affects the architecture you actually run.
Open Architecture Studio“We believe the packaged CDP category is being unbundled in public. CustomerLake is the most credible warehouse-native challenger because the data is already there — it does not need to be moved, modeled, or duplicated to be useful.”
For teams renewing a packaged CDP in the next 12 months, the right question is no longer 'which CDP?' but 'what should the warehouse own, and what gets composed on top of it?' That is a different architecture conversation, and most procurement processes are not asking it.
Continue down the reasoning graph
- Product UpdateΔ +5.1Iterable launches Nova AIIterable's Nova is the credible answer to Braze + OfferFit. It moves Iterable from 'execution platform with AI features' to 'engagement platform with native decisioning' — and resets the Braze vs Iterable comparison teams have been making for two years.Read perspective
- Product UpdateΔ +7.5Hightouch launches AI DecisioningActivation-layer vendor moves up the stack into autonomous audience and journey decisions.Read perspective
- AIΔ +4.0Snowflake releases Cortex AgentsNative agent framework reduces the need for standalone decisioning layers above the warehouse.Read perspective
- AcquisitionΔ +6.4Braze acquires OfferFitBraze acquires OfferFit to bring self-learning AI decisioning natively into Canvas. This materially raises the floor on engagement-layer AI and changes the case for a standalone decisioning vendor in the Braze stack.Read perspective
What does Valence think?
Every Valence Perspective has a permanent URL. Share it with your board, your team, or the operator who needs it.
