Our Methodology

Data Sources

We pull from 1,000+ sources globally. No single source tells the full story, so we cross-reference across categories until the numbers agree.

Data Collection

We combine primary and secondary research. Primary tells us the "why," secondary tells us the "how much." Neither works well on its own.

Primary Research

1. Consumer Surveys: Quantitative and qualitative surveys across demographics and geographies. We run these quarterly for core categories.
2. Expert Interviews: In-depth conversations with industry practitioners, category managers, and retail executives.
3. Focus Groups: Moderated discussions to get at the attitudes behind the numbers.

Secondary Research

1. Industry Reports: Published research from analysts and consultancies, reviewed for methodology before we reference it.
2. Government Publications: Official statistics, trade data, and regulatory filings.
3. Media Monitoring: Systematic tracking of news coverage, earnings commentary, and industry publications.

Data Analysis

Raw data needs structure to mean anything. Here's how we go from numbers to insight.

1. Regression Analysis: We map relationships between variables to understand what actually drives market outcomes.
2. Trend Analysis: Tracking changes over time, isolating seasonality, spotting inflection points early.
3. Sentiment Analysis: Consumer perception and emotional tone across digital channels and social media.
4. Thematic Analysis: Coding qualitative data into themes. This is where the "so what" behind the numbers comes from.
5. Cross-Category Correlation: We look at how shifts in one category ripple into others. These connections tend to get overlooked.

AI-Led Research

AI runs through every stage of our research process. We built the workflow around it from the start.

1. Data Ingestion: AI processes millions of data points from filings, news, and transaction data in hours, not weeks.
2. Pattern Recognition: ML models detect market shifts and emerging trends before they show up in traditional reports.
3. NLP for Unstructured Data: We use natural language processing to pull insights from earnings calls, news articles, and consumer reviews.
4. Forecast Refinement: Our models learn continuously from new data, improving forecast accuracy and flagging when assumptions need updating.
5. Human-in-the-Loop: AI handles volume. Our analysts handle judgment. Every AI-generated insight gets reviewed by a person before it reaches you.

Market Sizing & Forecasting

Our forecasts are built to hold up, not to make headlines. We use multiple estimation methods and reconcile them before publishing.

1. Bottom-Up Estimation: Building from transaction data, unit economics, and company-level inputs upward to category totals.
2. Top-Down Validation: Cross-checking against macro indicators, GDP data, and sector benchmarks.
3. Time Series Modeling: Historical trend projection with seasonality adjustments and economic cycle overlays.
4. Scenario Analysis: We publish base, bull, and bear cases. Single-point forecasts are misleading.

Validation & Quality

Every number we publish goes through multiple rounds of checking. We'd rather be late than wrong.

1. Cross-Validation: Testing findings against independent data sources. If two sources disagree, we dig until we know why.
2. Sensitivity Testing: We stress-test our assumptions. If a small tweak in inputs changes the conclusion, we flag it.
3. Expert Review: Industry practitioners review our output before it goes live. They catch what models miss.
4. Multi-Level QA: Analyst, senior analyst, and editorial review. Three sets of eyes minimum on everything we publish.