Every organisation you have worked for has paid a bill it never opened. It is paid in the hour spent hunting for an answer nobody wrote down, in the report redone because two departments meant different things by one word, and in the decision that has been "under consideration" for fourteen months while two teams quietly staff for both futures.
Nobody approved this bill, nobody budgets for it, and it does not appear in the annual report. That does not make it small. It makes it unexamined.
What the research can and cannot tell us
The best-known measurement of the search part of the bill is from the McKinsey Global Institute. In its 2012 report The social economy, it estimated that the average "interaction worker", meaning a manager or professional whose job is mostly coordination and judgement, spends about 28% of the working week managing email and nearly 20% looking for internal information or tracking down colleagues who can help with specific tasks.1
Nearly 20% is about one working day in five. Two cautions belong with it. The figure is an estimate from 2012, made for the purpose of sizing the value of social technologies, and it lumps together finding documents and finding people. It tells you the order of magnitude of what coordination costs. It does not tell you what any single organisation's number is.
The same caution applies to the figure most often quoted for the cost of poor data. Gartner puts it at an average of $12.9 million a year per organisation. The number traces to a 2020 survey in which customers of data quality software vendors estimated their own costs.2 It is a published benchmark built on self-reported estimates, and it is more honest to use it as a prompt for measurement than as a measurement.
The part that never shows up as search
Search is the visible line item. Two others are harder to see, because they look like normal work.
The decision left open. Leaders delay hard calls to keep options open, and it feels free. It is not. While the decision sits, people plan for two futures, effort is duplicated, and when the call is finally made, it arrives into suspicion. Sometimes the right answer really is not to decide today. Then the honest form is a small, funded option with a date and an owner. An open decision without an owner is not caution. It is a cost nobody is carrying on purpose.
The rule only one person can apply. Software engineers have a blunt name for this: the truck factor, the smallest number of people whose departure would leave a project without anyone who understands it. It has been studied properly. Avelino and colleagues proposed an algorithm for estimating it from commit history and applied it to 133 GitHub projects; in a survey of developers, most agreed that the people the algorithm named were the system's main authors.3 The idea carries well beyond code. Every organisation has an exception-approval rule, a pricing special case or a reconciliation that works because one long-serving person carries it. The finding is about the rule, never the person. When we show teams this, the usual reaction is relief: someone is finally going to write it down so that people can take leave.
What we measured in an illustrative sample
In the sample we built for the keynote described below, a fictitious claims department of 180 people lost about one working day in five to compensating for decisions that were never finished: searching for answers, redoing work after arguments about definitions, waiting for approvals, and meetings whose only purpose was to read a dashboard aloud. The sample is illustrative and says so. Its job is to show how the bill can be itemised, not to forecast your own.
Three ways to open it this week
- List every decision open for more than ninety days. Give each an owner and a date. If you cannot, you have found a cost.
- Add one column called "source" to your most important management table. Observed, declared, taken from a published source, or estimated, with a range. People stop arguing about whether a number is right and start asking how much weight it can bear.
- Ask where the truck factor is one. Pick the five rules the business cannot run without and ask who could apply each one tomorrow.
None of this needs us. The bill is cheaper to open than to keep paying.