Skip to content

Handle a newly nullable field without fabricating defaults

Propagate unknown values explicitly when a returned field no longer always contains data.

Guides

Handle a newly nullable field without fabricating defaults — what should the operator record?

A field once populated in every inspected row may later return null. That can be valid source behavior or an operational mismatch, but replacing null with zero or an empty label erases the distinction. Review the field's downstream role and keep unavailable values visible through the working representation.

Practical operations check
Propagate unknown values explicitly when a returned field no longer always contains data.
Record field_path, previous_assumption and observed_null_state, checking whether the field controls identity, eligibility, display or calculation. A null descriptive title can be shown as unavailable while a null primary identity may block joining; handle the field's operational role rather than treating all nulls equally.
Name which operations accept nulls and which remain blocked, avoiding a claim that every source row has become incomplete or invalid. Explain which fields were newly observed as nullable in this input, preserving uncertainty about broader behavior instead of asserting a universal change from one receipt.
01 / Guides

Identify affected semantics

Record field_path, previous_assumption and observed_null_state, checking whether the field controls identity, eligibility, display or calculation. A null descriptive title can be shown as unavailable while a null primary identity may block joining; handle the field's operational role rather than treating all nulls equally.

02 / Guides

Define explicit handling

Keep null as unknown or unavailable according to actual response meaning; block operations that require a real value instead of supplying a plausible default. An explicitly absent metric should not be converted to zero just to satisfy a numeric output type; maintain typed unknown state alongside any validated numeric column.

03 / Guides

Test downstream cases

Use fixtures with populated, null and absent forms to verify joins and calculations fail or retain uncertainty as the contract requires. Include a downstream calculation fixture where one operand is null and verify it does not produce a misleading exact delta, ratio or completed interval match.

04 / Guides

Review accepted rows

Inspect whether earlier transformations already converted unknowns into values and issue a scoped correction where observed evidence supports it. If earlier artifacts used a default, list only the affected revisions supported by lineage rather than claiming every table or every source record was damaged.

05 / Guides

Publish the updated contract

Name which operations accept nulls and which remain blocked, avoiding a claim that every source row has become incomplete or invalid. Explain which fields were newly observed as nullable in this input, preserving uncertainty about broader behavior instead of asserting a universal change from one receipt.

Keep the boundaries clear.

  • Propagate unknown values explicitly when a returned field no longer always contains data.
  • Operational record: Identify affected semantics. Context badges do not certify access.

Start with a read-only check: ask your agent to report actual source coverage before collecting records or exporting an audience.

Primary references

Documentation reviewed 2026-11-15. Product capabilities, client availability and plans can change. Third-party marks identify their products, not an affiliation.

Know what social media shows about you.
Start with a question.

Get started
montage