Root Cause Analysis
Find the Root Cause of Any Metric Change — Automatically
When revenue drops, Fig traces the causal chain through your knowledge graph, queries your warehouse for evidence, and delivers a structured report with suspected drivers and confidence levels — so your team investigates causes, not symptoms.
The Problem
Root Cause Analysis Shouldn't Take Days
Your revenue dropped 12% this month. Here's what happens next:
4 dashboards opened
30 minutes
6 SQL queries written
2 hours
3 meetings to discuss findings
3 hours
2 more days of investigation
2 days
Someone has a theory
Maybe.
Fig does this in 30 seconds. Structured root cause analysis with evidence, confidence levels, and recommended next steps.
How It Works
Three Steps to Root Cause
Fig's RCA agent follows a structured methodology: detect the anomaly, trace the causal chain, verify with evidence.
Detect
Fig continuously monitors your metrics. When an anomaly score crosses the threshold — or when you ask 'why did X change?' — the RCA agent activates.
Anomaly detection uses z-scores and historical baselines, not arbitrary thresholds. A 5% drop in revenue might be noise; a 5% drop in a metric that never moves more than 1% is a signal.
Trace
The RCA agent traverses your knowledge graph, following DERIVED_FROM and USES_DIMENSION relationships to identify every metric and dimension that could have caused the change.
If revenue dropped, Fig traces to its components: volume, price, mix. Then traces volume to pipeline, conversion, deal size. Then traces each of those to their upstream drivers. The graph tells Fig where to look.
Verify
For each suspected driver, Fig executes SQL queries against your warehouse and rates each one as low, medium, or high confidence based on the evidence.
No speculation. Every suspected driver comes with a specific query result: 'Discount rate increased from 12% to 18% in Enterprise segment (high confidence — explains ~40% of revenue decline).'
Output Format
What an RCA Report Looks Like
Every root cause analysis produces a structured report — not a wall of text. Here's the actual output format.
Headline
Revenue declined 15.2% MoM driven primarily by Enterprise discount rate increase and elevated return rates in APAC.
Summary
October revenue of $3.4M represents a $612K decline vs. September. Analysis of 14 upstream metrics identified two primary drivers accounting for ~78% of the variance. Enterprise discount rates increased 6.2pp following Q4 promotional approval. APAC return rates spiked to 14.3% (vs. 8.1% baseline) correlated with a shipping carrier change on Oct 3.
Suspected Drivers
Enterprise discount rate: 12.1% -> 18.3%
HighExplains ~$340K (55%) of decline. 23 deals closed with >20% discount in Oct vs. 4 in Sep.
APAC return rate: 8.1% -> 14.3%
HighExplains ~$142K (23%) of decline. Spike begins Oct 3, correlates with carrier switch.
SMB pipeline volume: -18% MoM
MediumExplains ~$85K (14%) of decline. Lead volume dropped but conversion held steady.
Recommended Next Queries
"Show me all Enterprise deals closed in October with discount >20%"
"Compare APAC return rates by shipping carrier for the last 90 days"
"What is the SMB lead-to-close funnel for October vs. September?"
Assumptions & Limitations
Analysis based on data through Oct 31. Discount rate sourced from deals table; some manual overrides may not be captured. APAC return rate excludes RMA items processed after Nov 5 cutoff.
Real Scenarios
Root Cause Analysis in Action
Every business metric has upstream drivers. Fig traces the causal chain, runs the queries, and tells you exactly what changed and why.
Revenue Declined 15%
The Problem
Monthly revenue dropped from $4.0M to $3.4M. Leadership wants answers by the board meeting.
Fig Traces
Fig traces through the knowledge graph: Revenue -> Volume x Price -> Segments -> Discount Rates + Return Rates. Identifies Enterprise discount rate increase and APAC return rate spike as primary drivers.
Outcome: so that the CFO walks into the board meeting with a specific, evidence-backed explanation and a remediation plan — not a theory.
Customer Churn Spiked
The Problem
Monthly churn rate jumped from 2.1% to 3.8%. CS team suspects product issues but has no proof.
Fig Traces
Fig traces: Churn Rate -> Cancellation Reasons -> Support Ticket Volume -> Resolution Time -> Product Areas. Finds that average ticket resolution time increased from 4.2 hours to 11.7 hours after a support team restructure.
Outcome: so that the VP of Customer Success can pinpoint the operational change that caused churn and reverse it — instead of launching a broad, unfocused retention campaign.
Gross Margin Eroding
The Problem
Gross margin declined 3.2pp over two quarters. The trend is subtle but accelerating.
Fig Traces
Fig traces: Gross Margin -> COGS Components -> Raw Material Costs + Labor + Product Mix. Identifies that raw material costs for two SKUs increased 22% while product mix shifted toward those SKUs by 8pp.
Outcome: so that the COO can renegotiate supplier contracts on the specific materials driving the cost increase — before margin erosion compounds further.
Ready to Trace Root Causes Automatically?
Start with free credits. Connect your data warehouse. See your first root cause analysis in under 5 minutes.