Designing Metadata That Actually Gets Used

Metadata only earns its place when users populate it consistently and find it useful. Most metadata designs fail not because of the technology, but because the columns are too many, too vague, or do not match how users think. Here is how to design metadata that survives contact with reality.

Start from how people search

List the questions users actually ask of the library:

• "Show me all current policies"
• "Find the contract with Acme that expires this year"
• "What are the templates for the legal team?"

Each question implies a column. Status (current/superseded), counterparty (Acme), expiry date, audience (legal). Design columns to answer the real questions.

Keep the count low

5 to 7 columns is the sweet spot for most libraries. More than 10 and users start ignoring them. If you need 15, split the content into 2 libraries.

Choose the right column type

• Choice: when there is a fixed list (status, department, region). Beats free text every time
• Managed metadata: when the list is large, hierarchical, or shared across libraries (taxonomies, product catalogues)
• Person: for owners, approvers, contributors. Links to the directory and supports filtering by team
• Date: for expiries, effective dates, deadlines
• Number / currency: for amounts, counts, percentages
• Single line of text: only when nothing else fits. Free text is the enemy of filterable views

Make required columns truly required

Mark only the columns you absolutely need to be populated. Required columns block uploads, which frustrates users when they cannot find the right value. Use defaults wherever possible so the column is populated without forcing user input.

Set defaults at the folder or library level

If a folder always contains contracts, set Document type = Contract as the default for that folder. Users do not need to set it manually for every upload.

Test with one team before scaling

Pilot the metadata design with one team for a fortnight. Watch what gets populated, what gets skipped, and what causes confusion. Adjust before rolling out across the library or to other libraries.

Connect metadata to retention and access

Once metadata is reliable, use it to drive:

• Retention labels (auto-apply based on document type)
• Sensitivity labels (auto-apply based on audience or classification)
• Views and dashboards (filter by status, owner, date)
• Search refiners

Review annually

Business needs change. Columns that were essential 2 years ago might be dead weight today. Review and trim.

For metadata design and governance support, submit a support ticket.