When does information actually become knowledge?

“I have plenty of information and I still cannot decide.” It is one of the sentences we hear most, and it is almost always treated as a data problem. It is not.

The information exists. It is in monthly reports, dashboards, spreadsheets and in the heads of four people. What is missing is structure: an explicit link between what gets measured and what has to be decided.

Why does having more data not help you decide?

Because the difficulty is not in the quantity, it is in the distance. Between a number and a decision there is interpretive work nobody has done, and adding more numbers increases that work instead of reducing it.

There is also a perverse effect. The more indicators exist, the easier it is to find one that confirms what you already wanted to do, and the harder it is to tell the indicator that should have changed the decision from the one that merely accompanied it. A dashboard with forty metrics is not forty times more informative than one with four. Almost always it is less.

What are the steps between a piece of data and an action?

There are four: information, knowledge, decision and action. Most reporting investment deals only with the first and assumes the other three happen on their own. They do not.

Information is data with context: the number alongside whatever it is compared with, be that last month, the target or the rest of the market. Knowledge is information interpreted by someone who knows what it means in that organisation, at that moment, and who can tell a normal variation from a signal. A decision is knowledge with a consequence: someone chose to do, or not to do, something because of it. Only then does action exist.

Each step depends on a different condition. The second needs someone with enough context to read the numbers; the third needs that person to have the mandate to decide; the fourth needs someone to have the time to execute.

Who should interpret the data in an organisation?

Someone close to the decision with context on the operation, and that responsibility has to be assigned, not assumed. In most companies, whoever produces the numbers is far from whoever decides, and whoever decides receives the numbers without the reading that would give them meaning.

The simplest way to close that distance is to require every report to come with a written interpretation: what changed, what it means and what is being proposed. It sounds like little, but it changes the nature of the work. Whoever prepares the report stops being a compiler and has to form an opinion. And whoever reads it discusses a proposal, instead of trying to decipher a table.

How do you know whether a management indicator is useful?

By asking it one question: which decision does this number change, and whose? If there is no answer, the indicator is not management information, it is well-presented noise. It may go on existing for legitimate reasons (a legal obligation, history, curiosity), but it does not belong on the table where decisions are made.

The question has a second half, and it is the one that usually stings: what value would change the decision? If there is no threshold, the decision did not depend on the number. It had already been made, and the number was only keeping it company.

Why can so many management reports be eliminated?

Because most indicators were created to answer a question nobody asks any more. Someone needed it once, the report stayed, and the monthly cadence gave it an appearance of necessity that was never questioned again.

Applying the decision test usually cuts reporting in half. And it is normally in that half that management starts finding what it was looking for, not because the numbers improved, but because they stopped being surrounded by numbers that decided nothing.

The secondary gain is time. Whoever produces the reporting produces less and explains more, which is precisely where the gap was.

What is tacit knowledge, and why is it so hard to preserve?

It is the knowledge that lives in the judgement of experienced people and never reached any system. It is hard to preserve because it is not solved by asking those people to write down what they know: in part, they themselves have no conscious access to everything they use to decide.

It is the manager who looks at a proposal and knows, without being able to explain quite why, that this client will pay late. It is the operations person who recognises a pattern in a complaint and anticipates a problem that has not happened yet. Neither of these judgements is in any report, and they are worth more than most of the ones that are. The risk is well known: that knowledge walks out of the door when the person does, and the organisation only notices late.

What ways are there to capture tacit knowledge?

The ones that work best observe decisions instead of asking for explanations: recording the reasoning at the moment a decision is made, examining the cases where an experienced person went against what the numbers said, and putting less experienced people alongside real cases instead of manuals.

None of these captures everything. They capture fragments, and the fragments are already worth a great deal, because they show which criteria are in use even when nobody can state them. The cases where the expert and the data disagree are the richest: either the expert saw something the data does not show, or the data shows something the expert has stopped seeing. Either way, something is learned.

The project that came from this: Arcane.Engine

But one question remains unsettled: how to turn those fragments into something the organisation can use without depending on whoever produced them. That is the question that led us to Arcane.Engine, and it is not closed yet.