Data Drift Recommendations
Solid detects when your warehouse schema drifts from your semantic model and surfaces recommendations to keep them in sync.
Your warehouse is always changing — columns get dropped, new ones gain traction, tables evolve. Data Drift Recommendations keeps your semantic model aligned with those changes automatically, so drift doesn't silently degrade your answers.
What Solid detects
Solid continuously monitors the relationship between your warehouse schema and your semantic model — no benchmark run required.
| Signal | What it means |
|---|---|
| Removed column | A column in your model no longer exists in the warehouse — it was deleted upstream. Leaving it in the model causes SQL failures and inaccurate results. |
| Rising-usage column | A column not currently in your model is seeing increased query activity in your warehouse. It may be worth adding. |
Where recommendations appear
Recommendations surface in the Optimize view of your semantic model. Each recommendation includes:
- Type — column removal or addition candidate
- Why this matters — a plain-language explanation of why Solid flagged it
- Impact — High, Medium, or Low, based on how frequently the affected column is referenced
- Apply Fix — a one-click action to apply the recommended change immediately
Model owners also receive an email notification the moment new recommendations appear, so nothing slips through unnoticed.
Applying a fix
Each recommendation comes with a pre-built fix. Click Apply Fix to preview the change and apply it — the model is updated and a new version is created automatically, preserving your full version history.
Click Dismiss for any recommendation that doesn't apply — for example, if a removed column was intentionally deprecated and the model reference should stay for historical context.
Why no benchmark run is required
Data Drift detection works by comparing the current warehouse schema directly against the model's asset list — a lightweight check that doesn't require generating or executing SQL. This means drift signals arrive continuously, not just after a scheduled run.
The bigger picture
Data Drift Recommendations is the first step toward a semantic layer that stays in sync with your warehouse automatically. As Solid adds more drift signals over time, this view will grow into a continuous quality feed — flagging issues before they reach your answers.
Updated 1 day ago