​
Overview

Data prioritization automatically kicks in whenever you use a Group By clause.

Within each group outlined by your Group By clause, the query executor would:

  • Sub-group the data further by their provider and their Source Type;
  • Keep only data from the sub-group that has the highest source type priority, or the highest provider priority if multiple sub-group exists for the same source type.

Data prioritization works only in a Group By context.

You are querying Sleep Score grouped by every 1 day:

{
    "select": [
        { "func": "newest", { "value_macro": "sleep_score" } }
    ],
    "group_by": [
        { "date_trunc": { "value": 1, "unit": "day }, "arg": { "index": "sleep" } }
    ]
}

Assuming this timeline:

  • Day 1: The user connected Oura.
  • Day 5: The user connected Apple HealthKit.
  • Both Oura and HealthKit have been consistently sending sleep data every day since connected.

If your team priority states that Apple HealthKit > Oura

The query result would be:

  • Day 1-4: Sleep Score computed from Oura sleep data
  • Day 5-7: Sleep Score computed from Apple HealthKit sleep data

If your team priority states that Oura > Apple HealthKit

The query result would be:

  • Day 1-7: Sleep Score computed from Oura sleep data

Apple HealthKit data are ignored in this case, because Oura is available throughout all days and has higher provider priority.

​
Previewing the prioritization effect

You can experiment and preview the effect of prioritization using the Query API.

By specifying a list of provider slugs in config.provider_priority_overrides, you instruct the query executor to treat these providers as the highest priority — above the team provider priority — specifically in this query invocation.

​
Managing Provider Priorities

You can manage the provider priorities for each summary type through the Data Priority section of the Vital Dashboard.

A larger number indicates a higher priority.

​
Source Type Priorities

We use pre-assigned, non-configurable Source Type priorities that follow the general expectation of data reliability.

(sorted from highest priority to lowest priority)

  • lab
  • automatic
  • watch
  • ring
  • chest_strap
  • scale
  • cuff
  • fingerprick
  • manual_scan
  • phone
  • app
  • multiple_sources
  • unknown
Using Query APIExamples