The cost of guessing
The invisible cost of not deciding
Work that comes back leaves a trace. Decisions that were escalated unnecessarily leave none, and there are more of them.
Decisions this page is about, and where the line actually was
Changing the wording on a customer-facing error message
4 → 1It had been yours all along
Refunding a customer under fifty pounds
3 → 1A written policy you had never been shown
Telling a customer a date
3 → 6Dates are committed by one person here
Reordering your own week
4 → 1You asked twice before believing the answer
- 1Decide it alone and say nothing
- 2Decide it alone and mention it afterwards
- 3Check with one colleague first
- 4Check with the person who assigned it
- 5Needs a decision one level up
- 6Cannot be decided here at all
four decisions · three of them you had more room than you thought · one gave you less · the median gap was three steps on the scale
Four of the nine decisions counted on this site.
When responsibility is handed over without enough context, managers can read the practical overview and compare project time and workload before treating delay as an individual failure.
The wider organisational question can also be compared with resources from Health and Safety Executive.
Rework is the visible cost of a thin handover. It has a date, an artefact and a conversation. Everybody involved knows it happened.
The invisible cost is larger. It is every decision somebody escalated when they did not need to: the question that was asked, answered yes, and forgotten by both parties within the hour.
Why it leaves no trace
A return produces a changed artefact. An unnecessary escalation produces a message and a reply, which live in a channel and are indistinguishable from the necessary ones.
There is no moment at which anybody says "you did not need to ask me that". The manager answers, because answering is faster than explaining the boundary, and the question was perfectly reasonable from where the asker stood. The asker reads the yes as confirmation that asking was correct.
Both people behave well and nothing is learned.
What it costs per instance
Three things, none of which shows up anywhere:
Elapsed time. The question waits in a queue behind a manager's meetings. A five-minute decision becomes two days, and the work proceeds on the assumption of a yes, which sometimes has to be unwound.
The other person's attention. Each ask is a small interruption to whoever is usually the team's bottleneck. Fifteen of those a quarter is a real amount of somebody's week spent confirming things that were already true.
Judgement not exercised. A decision you escalate is one you do not get better at making. Somebody who has escalated every pricing question for a year has a year of seniority and no pricing judgement, and discovers it on the day they are finally told to decide alone.
The count, where it can be counted
On the record kept for this site, four of nine decisions had more room than the person assumed. Each one is an afternoon of waiting and a question that did not need asking.
Four of nine is not a measurement of anything general — it is one person's quarter, and the real figure is certainly higher, because the ones that resolved in a minute were never written down.
What the number does is make the category visible. Before counting, this person would have said their escalations were about right. The count is what turned "I check when I should" into "I check about twice as often as I need to".
How to count your own
One line per escalation, for a fortnight. What it was, who you asked, and what came back.
Then go down the list and mark the ones where the answer was an unqualified yes with no new information in it. Those are your candidates. They will cluster — small spends, customer wording, scheduling — and the clusters are what to take to whoever can set a rule.
A fortnight is enough. The pattern is obvious by the eighth entry, and the exercise costs about thirty seconds a day.
What the escalations cluster into
The decisions people escalate unnecessarily are not spread evenly across the work. On every record of this kind they bunch into a small number of kinds, which is what makes the problem tractable: each kind only needs answering once.
- Anything with money attached, however small. The presence of a figure triggers caution regardless of its size.
- Anything customer-facing, including wording, timing and tone. The fear is of committing the company to something.
- Anything involving another team's time. Asking a colleague in another team for two hours feels like spending something that is not yours.
- Anything that looks like a commitment. Dates, promises, orders of work.
- Anything in an area where somebody was once corrected. The correction generalises outward and keeps generalising.
Five categories, and between them they cover almost all of it. Which means the fix is not a general conversation about trust; it is five sentences, each naming a category and a boundary, said once.
On the record counted here all four unnecessary escalations fell into the first two. Two sentences would have retired every one of them.
For managers, the harder half
The unnecessary ask is invisible to you too, and worse: it looks like engagement. A person who checks in is a person who is being careful, and the alternative — somebody who decides without telling you — is the behaviour managers are trained to worry about.
So the signal has to be looked for rather than noticed. One question, asked once a month: "what did you check with me that you should not have had to?" It is uncomfortable, it takes four seconds to answer, and it produces the single most actionable list anybody on this site can hand you.