Overview
Predictive Narrative Extraction identifies recurring topics discussed across time-stamped documents, detects the evolving narrative arc, and projects forward. It's qualitative forecasting powered by your organization's own written record — not statistical extrapolation, but pattern-matched narrative trajectory.
This fills the gap between backward-looking BI dashboards and forward-looking strategic planning.
The Problem
- Business intelligence looks backward — dashboards tell you what happened, not what's coming
- Qualitative trends hidden in reports, memos, and meeting notes go undetected because no tool reads across time
- Strategy teams manually synthesize quarterly reports to identify trajectory shifts — a slow, subjective process
- Quantitative forecasting misses the "why" — it can project numbers but not the narrative drivers behind them
- Driver shifts (e.g., churn moving from price-driven to product-driven) are critical strategic signals buried in unstructured text
How It Works
- Temporal topic tracking — Identify recurring topics discussed across multiple time-stamped documents
- Sentiment and emphasis extraction — For each topic at each time point, extract tone, emphasis, quantitative mentions, and causal explanations
- Narrative arc detection — Classify trajectory: improving, declining, accelerating, stabilizing, pivoting, or emerging
- Driver shift identification — Detect when the explanations or causes cited for a trend change over time
- Forward projection — Project narrative trajectory using pattern matching (LLM-powered qualitative forecasting, not statistical extrapolation)
- Trajectory cards — Present as timeline visualizations with forward projections, confidence bands, and driver annotations
- Confidence scoring — Based on data point density, trend consistency, and recency of latest observation
User Story
Your last 8 quarterly reports discuss customer churn. Q1: "Churn remains stable at 4%." Q3: "Mid-market churn increased to 6%, primarily driven by competitive pressure." Q4: "Churn acceleration continues; product gaps now cited as primary driver alongside pricing." The system identifies this narrative arc and projects: "Based on the evolving language and metrics in your reports, churn is accelerating in the mid-market segment, with the primary driver shifting from pricing to product gaps. If the trajectory continues, expect churn to reach 8–9% by Q2 next year." Each claim is linked to the source document and time point.
Complexity & Timeline
| Aspect | Detail |
|---|---|
| Complexity | Complex |
| Estimated Build | 5–6 weeks |
| Platform Dependencies | Metadata extraction (timestamps), Inferences (trend type), Surfaces (timeline visualization) |
| New Infrastructure | Temporal analysis engine, narrative arc classifier, trajectory projection models, timeline UI |
Target Clients
- Personas: Strategy Directors, Board Advisors, Investor Relations, Chief Analytics Officers
- Verticals: Private Equity, Corporate Strategy, Consulting, Financial Services, Media
- Pitch: "See where your narrative is heading — not just where it's been."
Revenue Potential
High-value output that speaks directly to strategy teams and board presentations — the ultimate budget holders. Unique capability: no tool currently does narrative trend detection from unstructured text well. Supports premium pricing for executive-facing analytics. Natural bundling with Scenario Simulator for a "strategic intelligence" package. Competitive advantage: fills a gap that neither BI tools nor generic AI assistants address.
Feature Synergies
- Scenario Simulator — "Here's the trajectory → now what if we intervene?" — narrative trends provide the baseline for scenario analysis
- Decision Journal — Track whether decisions to intervene in a narrative trajectory actually changed the trend
- Living Battlecard — Detect narrative trends in competitor-related documents and reflect them in battlecard "trajectory" sections
Risks & Open Questions
- Qualitative forecasting is inherently uncertain — projection confidence is hard to calibrate
- Requires sufficient temporal density (multiple documents on the same topic over time) to detect meaningful arcs
- Risk of apophenia — finding narrative patterns in what is actually random variation
- Users may conflate narrative projection with statistical prediction, leading to overconfidence