Data Warehouses Power Vertical Experiences

Diving deeper into

Charles Chretien, co-founder of Prequel, on the modern data stack’s ROI problem

Interview
data warehouses are going to go vertical and start powering vertical experiences instead of just being horizontal platforms.
Analyzed 5 sources

The big shift is that warehouses are starting to sell finished outcomes, not just raw storage and compute. A horizontal warehouse stores tables and runs queries. A vertical warehouse packages that same engine into a workflow a business team can actually use, like marketing audiences, observability dashboards, or AI monitoring. That matters because those products are easier to budget for, easier to adopt, and much easier to tie to revenue or cost savings.

  • CDPs are the clearest example of why this happens. Instead of giving a team raw customer data and asking data engineers to model it, a CDP connects a small set of sources, creates audiences, and pushes them into ad and messaging tools. That lets a marketing team see lift in conversion, CTR, or paid media efficiency, which is a cleaner ROI story than general purpose data processing.
  • The warehouse vendors are already moving this way through acquisitions and product expansion. Snowflake closed its Observe acquisition in February 2026 to add observability. ClickHouse acquired HyperDX for infrastructure observability and Langfuse for LLM observability. Databricks has expanded from Spark into warehousing, AI, and industry specific workflows, showing how a broad data platform can keep adding adjacent products.
  • This also changes who owns the relationship. In the older stack, warehouse vendors sold to data teams and tools like Fivetran, dbt, and reverse ETL sat on top. In the newer stack, the warehouse wants to own the business workflow itself. That is why reverse ETL compressed into CDP, why Fivetran bought dbt, and why SaaS vendors increasingly ship native warehouse connectors as a feature.

Going forward, the winners will be the data platforms that can turn warehouse primitives into packaged applications for specific jobs. Observability, marketing activation, and AI evaluation are early examples. The next phase is a market where Snowflake, Databricks, ClickHouse, and adjacent platforms compete less on query engines alone, and more on which high value workflow they can own end to end.