Redactions on Spreadsheets

Spreadsheets are not flat documents. Production files bundle formulas, hidden rows and columns, pivot tables, and workbook metadata—each of which can carry or echo sensitive values outside the cells a reviewer is staring at.

Redacting only what is visible can still leak data: a formula can recompute a redacted cell, a pivot can resurface excluded rows, and metadata can retain identifiers or paths teams assumed were gone.

Existing workflows forced context-switching between selection, redaction, and QA checks. That introduced manual handoffs, duplicated effort, and inconsistent review behavior across teams.

Before this project, redactions in spreadsheets were effectively handled one cell at a time. That approach was manageable for small files but broke down quickly on enterprise-scale datasets.

Review teams had to repeat manual actions across large grids, which increased fatigue and made it easier to miss sensitive values during production review.

Before interaction details, we mapped how production spreadsheets actually behave: formulas, hidden rows and columns, pivot tables, and metadata can all propagate or echo sensitive values. That dependency view defined what the product had to respect—not just what reviewers see in the grid.

From that map, a core move was modeling spreadsheet content as explicit redaction objects so the system could consistently track what was selected, what was redacted, and what still required validation across those relationships.

We anchored the design around three principles: expose risk before action, keep reviewers in-context, and make validation unavoidable at the right moments. Selection patterns reflected spreadsheet mental models—ranges, headers, repetitive structures—so reviewers could batch work without losing precision.

Validation cues stayed explicit rather than subtle. We traded visual minimalism for clarity so teams would not ship false confidence on sensitive productions.

Selection, redaction controls, and validation feedback were unified in one workflow so users could move from action to confirmation without losing orientation.

Review states and safeguards were designed to support both speed and defensibility, especially for teams reviewing large spreadsheets collaboratively.

Post-launch, teams reported fewer QA round-trips and more consistent review behavior across matters, especially in large spreadsheet productions.

The product reduced redaction anxiety by making risk states visible and traceable, which improved reviewer confidence in final outputs.

Business-wise, this strengthened Everlaw's ability to support high-stakes enterprise workflows where defensibility is a core buying criterion.