Jaco van der Laan

Designing Exposure Views: Delivering Data from Dual SCD2 Layers to Consumers

16/09/2025  ·  Data Architecture

article base/exposure-views-consumption-views
‹ ›

Designing Exposure Views- Delivering Data from Dual SCD2 Layers to Consumers

Four exposure patterns that balance usability, reproducibility, and auditability.

So far, we’ve focused on historizing (dual SCD2), cleaning (source access views), and optimizing (PIT). But at the end of the day, data is only valuable if consumers can use it.

Different consumers need different cuts of history. A data scientist exploring churn behavior doesn’t want the same view as a compliance officer reconstructing a customer’s past. That’s why we design exposure views in multiple schemas, tailored to use cases.

Four categories of exposure

Model/DDA version views

  • Tied to a specific model or DDA version.
  • Crucial for reproducibility and regulatory alignment

Latest technical + latest functional

  • The “current state” view most business users expect.
  • Strips away historization, delivers the here-and-now.

Latest technical + all functional (functional snapshots)

  • Provides business-valid snapshots across time, ignoring re-ingestion noise.
  • Ideal for time-series reporting and trend analysis.

All technical + all functional (technical snapshots)

  • The full historized view, including re-ingestions and corrections.
  • Essential for audit, debugging, and root cause analysis.

How it all ties together

  • Source access views ensure consistency at the entry point.
  • Dual SCD2 captures both business and system time.
  • PIT tables make historized joins fast and reliable.
  • Exposure views package it all up into the right “slice” for each consumer.

A note on snapshots

As I’ve written before ( Why we use SCD2 with PIT tables instead of just snapshots), snapshots alone don’t cut it. Exposure views built on dual SCD2 and PIT tables do. They give you precision, performance, and flexibility — the trifecta of a usable historized platform.

Closing the Series

Together, these four articles form a journey:

  1. Clean the entry (Source Access Views).
  2. Build the foundation (Dual SCD2).
  3. Make it performant (PIT tables).
  4. Expose it flexibly (Exposure Views).

This is how we turn historized raw data into something both trustworthy and consumable.