AI summary by TwinMind
The Real Reason Companies Have Become Obsessed With Your Data
How Money Works
11m 48s
Data as Trillion Dollar Asset
- Personal data (age, income, gender, preferences, sharing habits) combined with thousands of other data points = trillion dollar savings account for biggest tech companies
- Next 5 years: companies will start making withdrawals from this data wealth
- Facebook Cambridge Analytica scandal: data firm with Trump 2016 campaign ties collected private info from 50+ million Facebook users without permission
- Washington DC filed lawsuit against Facebook over Cambridge Analytica scandal
- Big tech companies collecting petabytes of data on every user they can lure onto platforms
Data Management Companies Worth Billions
- Companies like Snowflake, Teradata, Palantir worth billions of dollars
- Do nothing but compile/manage billions of data points and turn into useful information
- Snowflake so attractive that Warren Buffett changed investing strategy around Tech IPOs to include them in portfolio
- Companies harvesting data worth trillions (not just managing it)
- Information useful for much more than targeted advertising
Big Data Becoming New Big Oil
- Memory storage became orders of magnitude cheaper in 2000s
- Big tech companies realized data highly monetizable through commerce and direct sales to third-party data brokers
- Started collecting as much data as possible
- Data brokers take breadcrumbs gathered about users, combine with other obtainable data, share with businesses for marketing
- Back then: high volume data center memory cost $10,000 per terabyte, stored on highly inefficient hard drives
Early Data Collection Strategy
- All web companies had to store transactional user data (login info, names, account status)
- Big Data companies went further: collected thousands of non-essential data points
- How long users stayed on certain pages
- Time of day people interacted with services
- What caused users to leave platform
- Business value obvious but too expensive to store
- Compiling large data sets into usable information took weeks/months with available computers
- Companies bet storage would become cheaper and processing faster
- As technology improved, collected data became proportionally more valuable
- Companies starting earliest took biggest risks but built biggest data moats
Data Moats as Competitive Advantage
- Investors love companies hard to compete with (competition = lower margins, greater risk of losing customers)
- SEC and FTC make anti-competitive behavior illegal
- Companies build moats around markets - advantages incumbent market leader has over would-be competitors
- Detailed user data = huge advantage for one company over another
- Microsoft executives found providing business-to-business sales executives with detailed market analytics improved productivity by 30%
- New companies won't have user data until they get users
- Hard to get users when competing against entrenched companies with existing user data
Tesla Data Strategy Example
- Tesla has significantly higher earnings multiplier than legacy car brands
- Made collecting driver data priority
- Currently using data to develop full self-driving features
- If Tesla never successfully implements autonomous driving:
- Millions of hours of driving data could be sold to another automaker developing self-driving
- Could use data moat to make cars better in other ways
- Predicting driving habits to increase efficiency
- Determining where to put new superchargers on network
- Anything to make it harder for other electric automakers to catch up
Third-Party Data Analytics Tools
- Decade ago: companies needed expensive in-house tools to analyze data
- Developing software: time consuming, expensive, needs maintenance
- Many large corporations tried building data management tools in-house - most were clunky and slow
- Cost more than they earned in improved business operations
- Large tech companies with software experience were only ones that got it right
- Other companies left using basic regression analysis by analyst teams
Modern Data Analytics Market
- Market opportunity filled by third-party data analytics specialists
- Companies: Snowflake, Teradata, Amazon Redshift, Google BigQuery
- Services: ingest company database, provide tailored tools to work with customer data
- Price range: $250,000 to $5 million per year
- Insignificant investment for major companies that would have spent that much on internal tools anyway
- Customer turnover rates from biggest data analytics providers extremely low
- Modern tools use machine learning and advanced results-based algorithms
- Use data in ways never possible with human intuition alone
Algorithm Examples
- YouTube algorithm suggested video based on:
- Interest in finance and business
- Previous video watching patterns
- Watch time on different topics
- YouTube algorithm delivers result of maximum watch time on platform
- More watch time = more ads delivered + more data collected
- Advanced algorithms now accessible to all companies that can afford subscriptions
Data Cleaning and Integration
- Having billion data points means nothing if not organized in usable format
- As companies migrated data between databases, formats regularly changed/lost/duplicated
- Made much unusable without manual correction of millions of entries
- Modern tools fix problems without human intervention
- Make different data sources work together in one central database
EA Gaming Example
- EA gets data from Steam, Xbox, PlayStation, Nintendo, in-house Origin platform
- All platforms send user data in different formats
- Uses Snowflake to monitor:
- How long users play certain games
- What modes create highest player engagement
- How to most effectively push microtransactions
- Right balance of skills-based matchmaking for optimal dopamine
- Same algorithms as YouTube recommendation system
- Aim to keep users engaged by balancing winning/losing to prevent boredom or rage quitting
- Also helps predict server usage and optimize costs
Data as New Currency
- User data: highly valuable, cheap to store, easy to sell
- Become new type of currency impossible to tax
- Can be spent as many times as company wants
Hedge Fund Data Monetization
- Hedge funds quietly buying consumer credit/debit card transaction data
- Monitor which retailers outpacing/underperforming earnings projections
- Buy or short companies ahead of public announcements
- Profit off share price shifts following transactions
- Transaction data providers charge millions per year for real-time database access
- Some banks sell data directly through private agreements with hedge funds
- Others go through service providers like Bloomberg's Second Measure
- Subscription on top of regular Bloomberg terminal ($24,000/year)
- Hedge funds pay large premium to be exclusive purchaser of customer data
- Make profitable trades on technically public information nobody else has
- Major hedge funds subscribe to multiple data providers for comprehensive overview
- Get customer activity data before even the companies themselves
- Not strategy for Robinhood account with few thousand dollars
- Incredibly profitable for firms that can pay to play
