What information does a CEO need to run an organization as effectively as possible? It's a simple question, but I don't think most leaders spend nearly enough time thinking about it.

Your first answer might be "All of it. The more information the CEO has, the better!" Big data is hot, and we tend to think that more information means we can do a better job leading.

But in my experience, flooding a CEO with raw data is one of the quickest ways to reduce his ability to drive company performance.

I like to quote a paper by chemist Michael Schultz:

Decision making is impeded when evaluating information that is wrong or excessive, and thus [data] should be limited to the absolute minimum and most relevant available.

Big data is powerful; it's simply that it's not very good for helping a CEO fulfill her core responsibilities. Big data is fantastic when:

  • You have high-volume, high-velocity data sets.
  • You have the organizational capacity to mine them.
  • A machine is making automated decisions based on the data.

For example, I own a quant-based hedge fund. Every day a computer analyzes the entire trading pattern of the S&P 500 and automatically purchases or shorts numerous securities. The computer can process an incredible amount of information very quickly. And it works because we allow the computer to make the decision.

Running a company is a different story. Here, a human needs to make decisions, and the most relevant data to the CEO is often anecdotal and hard to quantify--you can't create an algorithm to make sense of it. As this New York Times op-ed puts it, "The things we can measure are never exactly what we care about."

Because of this, "small data" is more useful to the CEO than big data. That term has been used in several different ways, but let me boil it down to three characteristics every CEO should look for in the information he collects from the organization.

1. You need information that is normalized.

Most functional leaders are eager to aim the data fire hose at the CEO; they abstract their data into charts and graphs and push it up the organization. But no matter how smart the CEO, it just isn't realistic to expect her to understand and monitor functional data from finance, marketing, sales, etc. Without proper context, the mountain of data doesn't coalesce into anything informative or usable.

Instead of this bottoms-up approach, the key is to reverse our thinking and come at the problem from the top down. As CEO, ensure that you've established a clear destination for the organization, broken that into a few success criteria, and asked functional leaders to communicate their department's data in terms of those criteria.

The members of your executive team can then crunch the numbers (which they understand best) and deliver normalized information to you, tailored to what you need to know about the company's priorities.

2. You need information that is predictive.

CEOs set the future direction of the business--future being the most important word. Unfortunately, data is by definition historical. It tells you about what has already happened. You need to know what is likely to happen.

Therefore, one of the best forms of small data the CEO can collect is the answer to this simple question: "Will your group achieve its goals on time?" In answering, the functional leader can mentally place his group's data in the context of organizational priorities (normalization) and make an informed forecast.

This is a great way to transform executive team meetings from lengthy data reviews into brief, future-focused sessions.

3. You need information from the entire organization.

Finally, the best CEOs build systems for gathering information from the entire organization, not just the executive team.

If you were Henry Ford back in 1908, you could look out at the assembly line and easily monitor production. Today's work is much harder to measure--and you often can't see the same opportunities or issues that the knowledge worker sees, even though these insights may be of material value to the business.

In many classic business failures, someone within the organization knew there was a problem, but there was no effective system for communicating that problem up the chain.

How does a CEO go about collecting insight from the entire organization? The "walking around" method works, though you'll get a relatively small sample size. Suggestion boxes and surveys are another method, though low participation and hard-to-interpret responses are possible. Especially in large companies, aggregating insight from the whole team is a major challenge.

Big data is great, but when it comes to the CEO's duty to lead, small data (normalized, predictive information from the entire organization) is better. Fortunately, it's there for the taking--no data scientists required.