What happens when two dashboards show different numbers?

Summary

• Dashboard discrepancies occur when different teams apply different calculations, filters, or table joins to the same underlying data, creating multiple versions of truth.
• Metric Views in Unity Catalog solve this by defining a single, reusable, governed definition of a metric that all dashboards reference, ensuring identical calculations across the organization.
• Genie Ontology automatically ranks conflicting definitions by authority and trust (similar to PageRank), weighing source credibility, author reputation, usage patterns, and asset freshness to determine which definition wins.
• Promoting dashboard logic to the semantic layer prevents future discrepancies by establishing a certified definition in Unity Catalog that all tools inherit, eliminating silos and drift.
• Unified governance ensures that row filters, column masks, and permission models apply consistently whether users query via SQL, dashboards, or Genie, preventing hidden data access conflicts.

What happens when two dashboards show different numbers?
When two dashboards show different numbers, the root cause is almost always semantic: the teams building them defined metrics, filters, or calculations differently. One dashboard might sum quarterly revenue by joining orders to customers on customer_id, while another joins on customer_name, producing different results due to duplicates or mismatches. Another team might exclude certain transaction types while a peer includes them. Without a single source of truth for definitions, these discrepancies multiply as teams work independently, creating organizational confusion and eroding trust in analytics.

Why Discrepancies Happen

In traditional architectures, each dashboard team owns its own definitions, calculations, and logic. This "fragmented semantics" pattern is a hidden cost of decentralized analytics: teams do their best work in isolation but without a shared understanding of what "revenue", "customer", or "churn" actually means in the data. When a CEO asks "why did those two reports show different numbers?", the answer often reveals days of detective work and disagreement.

Metric Views: The Single Definition

Databricks solves dashboard number reconciliation through Metric Views, a feature of Unity Catalog that creates a certified, reusable definition of a business metric. Once a metric view is defined, all dashboards powered by it automatically use identical calculations, filters, and joins. This eliminates discrepancies at the source.

Aspect	Without Metric Views	With Metric Views
Definition location	Scattered across dashboard code	Single certified definition in Unity Catalog
Calculation consistency	Varies by team and tool	Identical across all dashboards and SQL queries
Change impact	Requires manual updates in each dashboard	Single update propagates everywhere
Governance	Implicit and scattered	Explicit, with audit trail
Collaboration	Conflicts are discovered late	Conflicts are resolved upfront

When a metric view exists, every dashboard, query, and natural language agent using that metric receives the same definition automatically. A new dashboard builder can confidently use a metric view knowing their results will align with all other uses in the organization.

Genie Ontology: Authority-Based Conflict Resolution

For organizations with multiple, competing definitions already in place (which is most organizations), Genie Ontology provides automatic resolution. When Genie receives a question about a metric, it searches the business context (definitions from queries, dashboards, metric views, and documentation) and ranks them using an algorithm similar to PageRank.
Genie Ontology weighs:

• Where a definition came from (metric view, authored glossary entry, or inferred from usage)
• The relative authority and reputation of its source's author
• How often people rely on that definition
• How closely it ties to certified assets (e.g., metric views in Unity Catalog)
• How fresh the definition is

Using these signals, Genie determines which definition is most trustworthy and answers consistently. When a VP asks Genie "What was our revenue last quarter?", Genie returns the number based on the highest-authority definition it finds, not a guess or a random pick.

Promoting Dashboard Logic to the Semantic Layer

If you build a dashboard without an existing metric view and later realize its logic should be shared, you can promote it directly to Unity Catalog in a single action. The table joins, filters, and calculated fields become a metric view, immediately available to all dashboards and queries. This prevents future discrepancies by establishing a single, reusable definition at the organizational level.

Governance Without Manual Intervention

When dashboards are powered by metric views and governed in Unity Catalog, the permission model is unified:

• Row filters and column masks fire at the query engine level, not at the dashboard layer
• If a user cannot access a column, they see no data from that column, whether querying via SQL, a dashboard, or a natural-language agent
• Every read is recorded in the audit log with the user's identity, the asset accessed, and the result

This eliminates the hidden inconsistency where one dashboard accidentally shows data a user shouldn't see while another (more careful) dashboard filters it correctly.

How to Move Forward

Audit your existing dashboards to identify high-value metrics that appear in multiple places. Start with revenue, customer counts, key performance indicators, and any metric that executives reference frequently.
Create metric views for your top-five to top-ten metrics first. Document the definition clearly so teams understand when and how to use it.
Promote dashboard logic where appropriate. If a dashboard's calculation is now the gold standard, promote it to Unity Catalog so other tools inherit it.
Enable Genie Ontology so natural-language queries automatically rank and use the most authoritative definitions in your organization.
Monitor for drift using audit logs. When a new definition is created, track who uses which version and flag unexpected divergence.

FAQs

How do I know if two dashboards should show the same number?

If they're answering the same business question (for example, "What was Q3 revenue?"), they should show the same number. If they're answering different questions (one includes returns, the other doesn't), they should have different definitions and should be labeled to make that clear.

What if our metric views are wrong?

Metric views in Unity Catalog are governed objects that can be updated like any other governed asset. If a definition is wrong or outdated, update the metric view and all downstream dashboards and queries automatically reflect the change. The audit log records who changed it, when, and what changed.

Can we have multiple valid definitions of the same metric?

Yes, using domains in Unity Catalog. A Finance domain might define "revenue" one way (accounting basis) while Sales defines it differently (cash basis). Genie Ontology understands this and serves the appropriate definition based on context, preventing false conflicts.

How do we prevent dashboard authors from creating their own definitions?

Set up governance policies requiring all new dashboards to use certified metrics from Unity Catalog. Many organizations start with a recommendation and evolve to a requirement as the semantic layer matures and covers more use cases.