Attribution Rules

The Attribution Rules page is where you manage the rules that determine how Databricks costs are assigned to teams in your organization. Rules are evaluated against every billing record — the right set of rules can attribute 90%+ of your spend with high confidence.

For the conceptual background on the attribution model, confidence tiers, and evaluation flow, see Cost Attribution & Confidence Tiers.

Rules list

The main view shows all configured attribution rules:

Column	What it shows
#	Priority — evaluation order (lower number means higher precedence)
Type	Exact, Pattern, or Overhead (proportional)
Name	Descriptive label (and description, if set)
Match	Summary of match criteria (resource ID, pattern, SKU, tags, domain)
Team	Which team the rule attributes to, split count, or distribution basis
Active	Toggle to enable or disable the rule

Rules are displayed in priority order. The evaluation engine processes rules from lowest priority number to highest, stopping at the first match.

The list API supports filtering by workspace_id, rule_type, and active_only query parameters.

Creating a rule

Click Create Rule and choose the rule type:

Exact rules

Match a specific resource by type and ID. Use these for known high-cost resources where you’re certain of the owner.

Field	What to enter
Name	Descriptive label
Priority	Evaluation order (1–50 recommended for exact rules)
Workspace	Required — exact rules are always workspace-scoped
Resource type	Cluster, Warehouse, Job, Pipeline, Endpoint, or App
Resource ID	The Databricks resource identifier
Attribution	Team, department, or org unit assignment (direct or split)

Pattern rules

Match resources by criteria. All conditions use AND logic — every specified condition must match. Conditions you leave empty are treated as “match anything.”

Field	What to enter
Name	Descriptive label
Priority	Evaluation order (50–200 recommended for pattern rules)
Workspace	Optional — leave empty for a global rule
Resource type	Cluster, Warehouse, Job, Pipeline, Endpoint, or App (optional)
Resource pattern	Regex against resource name or ID (e.g., ^prod-analytics-.*)
Principal domain	Email domain suffix (e.g., @analytics.company.com)
Tags	Databricks custom tags — key-value pairs that must all match
Attribution	Team, department, or org unit assignment (direct or split)

Note

Pattern rules with no workspace specified are global — they apply across all workspaces. This is useful for organization-wide tag mappings or naming conventions.

Proportional rules

Distribute platform overhead costs across teams based on their compute spend.

Field	What to enter
Name	Descriptive label
Priority	Evaluation order (200+ recommended)
SKU pattern	Regex to match SKU names (e.g., (?i)(STORAGE|DBFS), (?i)(NETWORKING|EGRESS))
Usage type	Exact match on usage type (optional, combined with SKU pattern)
Distribution basis	Compute spend proportion

Proportional rules are evaluated in a separate path from exact and pattern rules — they match against SKU names and usage types rather than resources. They’re designed for platform overhead costs that don’t belong to any single team.

Attribution modes

When configuring a rule’s attribution, choose one of these modes:

Direct

Assigns 100% of the matched cost to a single team, department, or org unit. You can optionally associate a project. If the project is marked as shared, the resource is treated as shared infrastructure and grouped by its shared bucket label (like shared:platform:analytics) in reports.

Split

Distributes cost across multiple teams by percentage. Percentages must sum to 100%.

Team	Percentage
Data Infrastructure	60%
ML Ops	40%

You can have up to 20 splits per rule. The UI provides a “Distribute evenly” helper to auto-balance percentages.

Date bounds

Rules can have optional “Valid From” and “Valid To” dates, configured under the Options section of the rule form:

Use case	Configuration
Temporary override	Set start and end dates for the override period
Migration	Old rule ends Dec 31, new rule starts Jan 1
Retroactive correction	Set the start date in the past to fix historical attribution

Date bounds are stored as metadata on the rule for organizational purposes.

Testing rules

Simulate

Use the Simulate tab to test how a specific resource would be attributed without creating billing records:

1. Enter the resource details: resource type, workspace, resource ID, and optionally the owner email.
2. Click Run Simulation.
3. Review the evaluation chain — each active rule is shown in priority order with its match result and reason.

The simulation shows the first matching rule, the team it would attribute to, and the confidence level (exact or pattern). If no rule matches, the result indicates that costs would use fallback attribution.

Tip

Always simulate a new pattern rule before activating it. A regex that’s too broad can inadvertently capture resources that belong to other teams.

Priority management

Rules are evaluated in priority order. Managing priorities effectively is key to accurate attribution.

Recommended priority ranges

Range	Use
1–50	Critical exact matches for known high-cost resources
50–100	Specific pattern rules (narrow regex, specific workspace)
100–200	General pattern rules (broad regex, global scope)
200+	Catch-all and proportional rules

Priority conflicts

When multiple rules share the same priority:

1. Workspace-specific rules take precedence over global rules.

You can reorder priorities by editing the priority number directly or using the reorder API.

Editing and deleting rules

• Edit — Modify any field of an existing rule. Changes take effect on the next attribution cycle.
• Toggle active — Use the Active toggle in the rules list to deactivate a rule without deleting it. Useful for testing.
• Delete — Permanently remove a rule via the actions menu. If the rule is referenced by existing cost allocations, you will be prompted to deactivate it instead.

All rule changes are recorded in the Audit Log.

Caution

Deleting a high-priority rule can cause a large volume of costs to shift to lower-priority rules or become unattributed. Always simulate the impact of removing a rule before deleting it.

Common workflows

Getting started with attribution

1. Start with the Tag Governance page to map existing Databricks tags to teams.
2. Create exact rules for your top 10 most expensive resources.
3. Create pattern rules for naming conventions (e.g., ^prod-.* for production resources).
4. Set up proportional rules for platform overhead (networking, storage).
5. Monitor the Cost Explorer Attribution tab — aim for less than 15% unattributed.

Improving attribution coverage

1. Check the Cost Explorer Attribution tab for the current unattributed percentage.
2. Drill into unattributed costs to see which resources and categories are unmatched.
3. Create rules targeting the highest-cost unattributed resources first.
4. Use simulation to verify rules before activating them.
5. Repeat until unattributed costs reach an acceptable level.

Handling organizational changes

1. When teams are restructured, update attribution rules to reflect the new ownership.
2. Use date bounds to make the transition clean — old rule expires on the reorganization date, new rule starts.
3. Test the new rules with simulation before the transition date.

Next steps

• Cost Attribution & Confidence Tiers — The attribution model and evaluation flow
• Tag Governance — Tag-based attribution shortcuts
• Cost Explorer — Monitor attribution quality
• Organizational Hierarchy & Budgets — The team structure that rules target