1. Video Streaming Platform
  2. Features
  3. Source data
Feature · Source data

Source data

Capture source snapshots with coverage and freshness, and keep reviewed field mappings from each source system.

OverviewSource Snapshot Freshness
Video Streaming Platform · Source data

Source Snapshot Freshness

Source snapshots

New
NameSource systemCaptured atCoverage startCoverage endRecord count
Snapshot 774SS-92723 Oct28 Oct01 Nov23
Snapshot 132SS-94114 Oct21 Oct23 Oct18
Snapshot 152SS-65527 Oct29 Oct01 Nov1
Snapshot 736SS-75805 Oct07 Oct12 Oct13
Snapshot 810SS-91521 Oct29 Oct31 Oct24
6 of 6 records
Snapshot 774Source snapshots · SS-10461
Fresh
Source systemSS-927
Captured at23 Oct
Coverage start28 Oct
Coverage end01 Nov
Record count23
Example screen with sample data

What you can do

Know what the numbers were built from

Playback, payments and catalog data arrive from other systems. Each time a source is captured, a snapshot records the source system, when it was captured, the period it covers, the number of records and a hash of the content. Its freshness is marked Fresh, Stale or Failed. Metric observations point to the snapshot they came from, so a number can always be traced to the capture behind it.

Source mappings describe how a field in a source system becomes a field here: the source and target field, their types, the transform rule, the version and the date it takes effect. Mappings are reviewed before use and versioned when a source changes its format.

How the work is organized

  • Snapshot with coverage. Every snapshot states the start and end of the period it covers and how many records it holds. Gaps in coverage are visible.
  • Freshness before use. Only fresh snapshots should feed results. A stale or failed snapshot is a reason to hold or invalidate a result.
  • Versioned mappings. A changed source format gets a new mapping version with its effective date. Earlier results keep the mapping they used.

What to agree during setup

Agree the source systems and how often each is captured. Agree what makes a snapshot stale, for example more than a day old for playback. Agree how record counts are checked against the source system’s own totals. Decide who reviews new mappings and how transform rules are written so they can be read by an analyst.

Check the handoff on an actual case

The playback snapshot for March 3 is captured at 06:00 the next morning and covers March 3 in full, with 1.2 million records. The Data Steward compares the record count with the player’s own daily total and marks it Fresh. On March 5 the capture fails partway; the snapshot is marked Failed and no results are calculated from it.

In April the player renames its watched-time field. The steward adds a new mapping version effective April 1 that maps the new field to watched minutes. March results keep the earlier mapping, and April results use the new one.

What your team sees

  • Source Snapshot Freshness — Snapshots by source system with capture date, coverage start and end, record count and freshness.
  • Invalid Metric Results — Results invalidated by a stale or failed source, for the steward to resolve.

Follow the work through

Plan your integrations before the first capture.

What it keeps track of

Data model
Source snapshots
Captured source-system snapshots with coverage and freshness evidence
Source mappings
Reviewed field mappings from source systems into platform fields
Metric observations
Observed metric values captured per source period and unit

Who uses it

Permissions
Data Steward
Captures snapshots and maintains mappings.
Metrics Analyst
Reads snapshots and mappings when results look wrong.
Admin
Manages access.

Set up Video Streaming Platform your way.

Start free, then use Proto to change fields, steps and approvals before your team goes live.

How to get started