Archive · 26 Oct 2025
Data Delusion: More Data Doesn't Always Equal Better Decisions
Imagine you're trying to pick a winning horse.
At first, a few key facts help, and you pick the winner some of the time.
But if you get tons of info, your picks don't get much better.
The study showed that even when experts got more data, they weren't much better at predicting who'd win, even though they felt more confident.
Too much information can be hard to use.
Sometimes the extra data isn't helpful, and you might think it is.
This is like businesses spending lots on data but not seeing results.
To fix this, decide what you need to improve, and focus on the important data.
A study with horse race handicappers showed that more data did not improve prediction accuracy, though confidence increased.
Organizations may experience underwhelming business outcomes despite investing in data assets.
The value of information is determined by its impact on improving business decisions; the cost is its acquisition.
Data's effectiveness depends on the decision; some business decisions have greater inherent uncertainty than others.
Focus on improving the governance of existing structured data to improve decision quality.
- Who
- A risk management and AI consultant examined the impact of data on business outcomes, using a horse racing study as an example.
- What
- The study suggests that more data doesn't necessarily lead to better business decisions, and can sometimes lead to poorer outcomes. The article explores the value versus cost of information in business decision making.
- When
- In 1974, a study was done regarding prediction and data.
- Where
- The study involved horse handicappers.
- Why
- To understand how the availability of more data impacts the quality of business decisions.
Optimizing Data Strategies
Avoiding Information Overload
Use of Data
Optimizing Data Strategies
Focus on gathering all possible data, even if low value, to expand insight.
Avoiding Information Overload
Prioritize high-quality structured data and relevant information to drive effective business decisions.
- Experiment Subject
- Horse handicappers
- Experiment Goal
- Predict winners of horse races
- Attributes per Horse
- 40
- Prediction Rounds
- 4
- Initial Prediction Success Rate
- 17%
- Data Infrastructure Failure Rate
- Over 60%