Interview, Fireside Chat
Analyzing Billions of Transactions to Understand Consumer Behavior - Michael Babineau and Kevin Hale
- Foundational Shift: Co-founders Mike and Lillian pivoted from a video game background (Electronic Arts, Zynga) to financial analytics after realizing hedge funds lacked in-house coding teams to process messy transactional data.
- Core Business Model: The company builds an analytics platform that empowers users (investors and corporations) to answer their own questions using credit card transaction data, rather than selling pre-written research reports.
- First Customers: The very first customer was a Venture Capitalist (VC) who invested in the company during their Summer 2015 YC batch; most current Bay Area VCs are now clients.
- Data Granularity: The platform processes 50+ billion transactions, containing 1 billion unique transaction descriptions, which requires complex entity resolution to map to roughly 1,000 target companies.
- Key Use Cases:
- Venture Capital: Used for due diligence to verify startup growth metrics (e.g., comparing Bird vs. Lime unit economics) and identify market share trends.
- Public Equity: Used for competitive intelligence, such as determining if Peloton surpassed Soul Cycle in active members.
- Corporate Strategy: Used by non-financial companies (e.g., Stitch Fix, Amazon) to analyze customer behavior and cross-selling opportunities.
- Data Science Team: The company employs 60 people, with a 50/50 split between engineers and data scientists; one-third of the staff hold PhDs in diverse fields like genetics, neuroscience, and climate science.
- Hiring Philosophy: The team prioritizes "scientists with a capital S" who can solve problems from first principles, specifically screening for the ability to identify data "dragons" (errors, biases, and inconsistencies) rather than just executing tool-based queries.
- Data Complexity: Transaction data is unstructured with high cardinality (e.g., Macy's has 3 million text variants per location) due to human input errors, franchise point-of-sale variations, and processing chain perturbations.
- Editorial Strategy: A dedicated team of data scientists and writers uses the product internally ("dogfooding") to generate blog posts and press quotes for outlets like the Wall Street Journal and Financial Times.
- Investment & Funding:
- Series A: Raised with a lead from Bessemer Partners and co-lead from Goldman Sachs.
- Strategic Investors: Citi Bank participated, recognizing the company's ability to solve the specific "messy data" pain point they face internally.
- Investor Profile: Goldman Sachs and Citi represent a strategic push into "alternative data," specifically for credit card insights which was previously untapped by these institutions.
- Product Strategy: The company avoids selling raw signals (which lose value as more users subscribe) in favor of selling a toolset that allows users to generate their own unique "information edge" by asking creative questions.
- Growth Model: Achieved 150 clients via 100% inbound growth and word-of-mouth virality within the YC and VC community, without any outbound sales efforts.
- Market Limitations: The data is effective for direct-to-consumer (B2C) US transactions but cannot effectively track business-to-business (B2B) or non-credit-card based models like grocery store chains (e.g., General Mills).
- Forward-Looking: The company intends to continue expanding its entity resolution capabilities and de-biasing techniques to ensure its longitudinal consumer panels remain representative of the broader US population.
- Strategic Evolution: The team broadened its focus from purely "investors" to "anyone needing to understand company performance" after realizing that corporate clients also sought the same competitive intelligence data.