Strategic Analysis
12 Analysis Algorithms Built for Business Questions
Fig doesn't just query your data. It runs structured analysis algorithms — concentration risk, growth decomposition, variance detection, funnel analysis, cohort retention, and more — so every answer comes with evidence, not guesses.
The Problem
SQL Answers Questions. Algorithms Answer Business Questions.
When your CFO asks "why did revenue drop?", a SQL query returns a number. But the real answer requires decomposing growth into volume, price, and mix effects. It requires checking whether the drop is concentrated in a few accounts or spread broadly. It requires comparing this quarter's cohort retention against prior cohorts.
These aren't SQL problems. They're analysis problems — and they require structured algorithms that know how to frame the calculation, handle edge cases, and present results in business context.
Most AI tools convert natural language to SQL and stop there. Fig goes further: it selects the right analysis framework, executes it against your warehouse, and returns a structured answer with evidence.
SQL Query
Revenue is $4.2M
Basic AI
Revenue dropped 12% vs. last quarter
Fig Analysis
Revenue dropped 12% — 70% driven by volume decline in Enterprise segment, 20% by pricing pressure in APAC, 10% by mix shift toward lower-ACV products
The Engine
12 Algorithms. Zero Guesswork.
Each algorithm is purpose-built for a specific class of business question. Fig selects and chains them automatically based on what you ask.
Concentration Analysis
Calculates
Pareto 80/20 distribution, cumulative share curves, and top-N contribution percentages.
When Fig Uses It
When Fig detects revenue, customer, or SKU concentration questions.
Example: Identify that your top 3 customers represent 52% of revenue so that you can quantify churn risk and diversify proactively.
Growth Analysis
Calculates
Current vs. baseline period comparison, absolute and percentage growth, ranked by growth rate.
When Fig Uses It
When a question involves change over time across segments or categories.
Example: Rank product lines by growth rate so that you can double down on winners and investigate laggards.
Variance Analysis
Calculates
Z-score outlier detection, high/low/average categorization, deviation from expected values.
When Fig Uses It
When Fig needs to find anomalies, outliers, or deviations from plan.
Example: Surface the 4 regions with spending 2+ standard deviations above budget so that finance can investigate before quarter-end.
Metric Trend
Calculates
Time-series decomposition across monthly, weekly, or quarterly intervals with direction indicators.
When Fig Uses It
When questions involve 'over time,' 'trend,' or 'trajectory.'
Example: Visualize monthly churn rate across 12 months so that you can spot seasonal patterns and plan retention campaigns.
Metric Comparison
Calculates
Side-by-side segment comparison with absolute and relative differences.
When Fig Uses It
When comparing performance across regions, teams, products, or any dimension.
Example: Compare ARPU across enterprise, mid-market, and SMB segments so that you can validate pricing strategy.
Top-N Ranking
Calculates
Best and worst performers ranked by any metric, with percentile positions.
When Fig Uses It
When questions ask for 'best,' 'worst,' 'top,' or 'bottom' performers.
Example: Rank sales reps by quota attainment so that you can identify coaching opportunities and replicate top-performer behaviors.
Distribution Analysis
Calculates
Histogram buckets, percentile breakdowns (P25/P50/P75/P90), spread and skew metrics.
When Fig Uses It
When Fig needs to understand the shape of a metric across a population.
Example: Map deal-size distribution so that you can set realistic targets and identify whether your pipeline is healthy or top-heavy.
Rate Analysis
Calculates
Conversion rates, churn rates, retention rates, and win rates with period-over-period changes.
When Fig Uses It
When questions involve ratios, percentages, or rate-based metrics.
Example: Calculate monthly churn rate by customer segment so that you can prioritize retention efforts where they matter most.
Driver Analysis
Calculates
Decomposition of metric changes into contributing factors with magnitude and direction.
When Fig Uses It
When Fig needs to explain why a metric changed.
Example: Decompose the $2.3M revenue shortfall into price, volume, and mix effects so that you can assign accountability to the right teams.
Composition Analysis
Calculates
Share/breakdown of a whole into parts, with percentage contribution of each segment.
When Fig Uses It
When questions ask about breakdown, mix, or share of total.
Example: Break down revenue by channel so that you can understand the true cost of each acquisition path.
Funnel Analysis
Calculates
Stage-by-stage conversion rates, drop-off points, and bottleneck identification.
When Fig Uses It
When questions involve sequential processes like sales pipelines or user journeys.
Example: Map the lead-to-close funnel so that you can identify the stage with the biggest drop-off and fix it first.
Cohort Analysis
Calculates
Cohort-based retention curves, period-over-period cohort comparison, and decay rates.
When Fig Uses It
When questions involve retention, loyalty, or behavior over customer lifetimes.
Example: Compare Q1 vs. Q2 customer cohort retention so that you can measure whether onboarding improvements actually moved the needle.
Analysis Blueprint
How Fig Chooses the Right Algorithm
Fig doesn't guess. Its planning agent reads your question, identifies the business intent, maps it against your data's structure, and selects the analysis algorithm that fits — before writing a single line of SQL.
Parse the Question
Fig's planning agent reads your question and identifies the business intent: are you asking about concentration, trend, comparison, ranking, or something else?
Select the Algorithm
Based on the intent and the structure of your data, Fig selects one or more algorithms. A concentration question triggers Concentration Analysis. A 'why did X change?' question triggers Driver Analysis.
Execute with Evidence
Fig writes the SQL, runs the algorithm against your warehouse, and returns a structured result with specific numbers, not vague summaries. Every claim is backed by a query you can inspect.
Algorithm Selection Is Deterministic, Not Probabilistic
Unlike LLMs that generate freeform analysis (and hallucinate numbers), Fig's algorithms are deterministic code paths. When the planner selects "Concentration Analysis," it runs a specific, tested algorithm that calculates Pareto distributions the same way every time. The AI chooses the tool; the tool does the math.
Chained Analysis
Multi-Phase Business Review
The real power isn't any single algorithm — it's how Fig chains them. Ask "give me a quarterly business review" and Fig runs a multi-phase analysis that would take an analyst days.
Phase 1
Baseline & Trend
Fig runs Metric Trend on your core KPIs to establish what happened over the review period.
Phase 2
Decomposition
For any KPI that moved significantly, Fig runs Driver Analysis to explain why it changed.
Phase 3
Deep Dives
Fig follows up with targeted analyses: concentration checks on key accounts, funnel analysis on conversion, cohort retention on new customers.
Phase 4
Synthesis
Fig combines all findings into a structured report with executive summary, detailed findings, and recommended next questions.
Result: A complete business review with trends, drivers, risks, and recommendations — in minutes, not days.
Ready to Move Beyond Basic Queries?
Start with free credits. Connect your warehouse. Ask your first business question and see structured analysis in action.