shipping production AI · since 2020 NAICS 541511 / 541512 / 541519  ·  CMMC-aware
Refinery Report / Data Engineering / post · urable
Data EngineeringData GovernanceEnterprise AIAnalytics

BigQuery Friction Log: Instrumenting the Discovery Journey to Make Governance, Trust, and Self-Service Measurable

Most "data governance" fails at the same point: the moment a real person tries to find and use data. The BigQuery Friction Log fixes that blind spot by treating data discovery as an observable workflow, not a private mental struggle.

D
DSE-Experts
Operator-led practice
December 21, 2025
4 min · 931 words

Executive Summary

Most “data governance” fails at the same point: the moment a real person tries to find and use data in BigQuery. They hit a wall—permissions, unclear meaning, stale tables, missing lineage, duplicate datasets—and the system records almost nothing. Leadership then funds more cataloging and more policy work without proof of what broke, where, and for whom. The BigQuery “Friction Log” fixes that blind spot. It treats data discovery as an observable workflow, not a private mental struggle.


The Governance Blind Spot

Data governance investments often fail to deliver because they lack visibility into the actual user experience. Leadership funds: - More cataloging efforts - More policy documentation - More training programs - More metadata initiatives

But without measurement, there’s no proof of what’s actually broken, where it breaks, and for whom.

The Discovery Journey Problem

What Users Experience What Systems Record
Can’t find the right table Nothing
Don’t understand the schema Nothing
Get blocked by permissions Maybe a denied access log
Use wrong data by mistake Nothing until it’s too late
Give up and ask someone Nothing

What the Friction Log Changes

The BigQuery Friction Log captures discovery steps and obstacles as structured, queryable events. Teams can then measure where users stall and why.

From Invisible to Observable

Before Friction Log: - Discovery is a private struggle - Failures are anecdotal - Improvements are guessed at - ROI of governance is faith-based

After Friction Log: - Discovery is an observable workflow - Failures are measurable - Improvements are verifiable - ROI of governance is evidence-based


The Accountability Shift

Done well, a friction log changes accountability. The platform stops hiding behind “uptime” and starts answering for user outcomes:

Metric What It Measures
Time-to-first-successful-query How long until users get value
Abandoned searches Where discovery fails completely
Repeated permission denials Access control friction
Rework from semantic confusion Cost of unclear definitions

A Micro-Scenario: Revenue Table Discovery

An analyst searches for “revenue,” opens a table, runs a query, and gets blocked by a policy tag.

What the Friction Log Records

  1. Search: Query “revenue” entered
  2. Result click: sales.monthly_revenue selected
  3. Schema view: User reviewed column definitions
  4. Query attempt: SELECT statement submitted
  5. Permission denied: Policy tag PII_RESTRICTED blocked access

The Context Captured

The Remediation

The owner now has specific, actionable information: - Update the policy mapping, OR - Add an approved view without PII, AND - Document the definition of “revenue” for this table - Set an SLA for future access requests

The metric moves or it doesn’t. If it doesn’t, the fix was wrong.


ROI-Grade Evidence for Data Investments

The friction log yields evidence for investments in:

Investment Area How Friction Log Proves ROI
Metadata quality Reduced semantic confusion, fewer wrong-data incidents
Catalog coverage Faster discovery, fewer abandoned searches
Lineage documentation Faster trust verification, reduced rework
Policy design Fewer permission denials, appropriate access
Documentation Reduced time-to-first-query

This borrows the discipline of product analytics and observability, but applies it to the discovery journey inside BigQuery.


Implementation Approach

Phase 1: Instrument (Weeks 1-4)

  1. Define discovery events to capture - Search queries and results - Table/dataset interactions - Query attempts and outcomes - Permission checks and results

  2. Build the logging infrastructure - Structured event capture - Queryable storage - Real-time and batch analysis

  3. Establish baseline metrics - Current time-to-first-query - Current abandonment rate - Current permission denial rate

Phase 2: Analyze (Weeks 5-8)

  1. Identify top friction points - Which searches fail most often - Which tables cause the most confusion - Which policies block the most legitimate access

  2. Segment by user type - Analysts vs. engineers vs. executives - New users vs. experienced users - Teams with high vs. low data literacy

  3. Prioritize remediation - Impact × frequency × effort

Phase 3: Remediate and Verify (Weeks 9-12)

  1. Fix the top friction points - Update policies - Improve documentation - Add approved views - Clarify definitions

  2. Measure the impact - Did the metrics improve? - By how much? - For which user segments?

  3. Iterate - Move to the next friction point - Build a continuous improvement loop


Strategic Implications

For Data Platform Teams

For Data Governance Leaders

For Enterprise Architects


Conclusion

The BigQuery Friction Log represents a fundamental shift in how organizations approach data governance. By treating discovery as an observable workflow, teams can:

Governance stops being a policy exercise and becomes an engineering discipline. Trust becomes debuggable. And self-service becomes measurable.


Sources


Related Insights

P
Founder · Principal Engineer
Data & AI engineer · 10+ yrs hands-on

Writes most of the long-form here. Lives in the codebase. Active on GitHub and LinkedIn.

One long-form a week. No marketing.

Subscribe to the Refinery Report. Practitioner deep-dives on AI engineering, security, and the realities of running production systems. Unsubscribe in one click.

~12 issues / quarter