Lesson 22 of 23 · 6 min read
Measure honestly: identified vs confirmed
In Lean: identified versus confirmed
Builds on: Did it hold?
In one minute
Improvement work is easy to oversell, so its numbers need firm rules. An identified figure is the time a person estimated a friction costs them, typed when they reported it; a confirmed figure is typed at a check-in after the change, by the person answering it. The two are kept apart and never added together, and nobody invents or forecasts a figure to fill a gap. When there is no figure, the honest entry is "no figure", which is not the same as zero.
The idea
Honest numbers are part of the method, not decoration. Once figures are inflated, people stop trusting them, and the team's choices about what to fix next get worse. So every number you show should trace back to something a person typed or confirmed.
Two figures, two moments.
- What the reporter estimated at the start (call it identified). When someone reports a friction, they can say how often it happens and how long it takes each time. That gives a weekly estimate of their own time the friction costs, made before anything changed. It shows where time goes, not what has been saved.
- What a check after the change showed (call it confirmed). After the change, a check-in asks whether it held, and whoever answers can type the weekly time the change now saves. If it held and no new figure was typed, the figure from the report can stand in, labelled as coming from the report. Either way, a person typed it.
- Never added, averaged or divided into one figure or a ratio.
- No figure is not zero. A friction reported without a number shows “no figure”.
- Nothing projected. One person’s figure is not multiplied across a team or a year.
The rules.
- Never add them, average them or divide one by the other. Identified plus confirmed counts the same friction twice. Confirmed divided by identified looks like a success rate, but it compares an early estimate with a later check. Show them side by side.
- No figure is not zero. A friction reported without a number is still a real problem; it shows as "no figure". Writing 0 would claim it costs nothing, which nobody said.
- Partly counts only what was typed. If a change partly held, do not assume half of the original estimate. If the person typed a figure, use it; if not, there is no confirmed figure.
- The latest check decides. The answer at a month outranks the answer at two weeks, whenever each was given. A change that held at two weeks and reverted at a month confirms nothing now. The earlier answer stays as history.
- No projected savings. Do not multiply one person's figure across a team, a department or a year and present it as a result. Each person reports their own time; colleagues with the same friction report their own.
- Neither figure caps the other. A confirmed figure can be larger than the estimate, or exist without one. Keep both as typed.
These rules make results look smaller than a generous estimate would. That is the point. A small number that everyone trusts is worth more than a large one that collapses at the first question.
Example
Illustrative example, not a real company.
A finance assistant reports that she rebuilds a reconciliation file by hand. She says how often and how long, so her friction has an identified figure. A colleague reports a related friction and gives no figure.
The change goes in. At the one-month check-in she answers held and types the time it now saves her each week, a little less than her first estimate. Her colleague answers partly and types nothing.
The honest summary for their manager: her identified figure and her confirmed figure, shown side by side; her colleague's friction, partly held, no figure. The tempting version, "we saved this many hours a year", would add her estimate to her confirmed figure, guess a number for the colleague, and multiply it all by fifty-two weeks. Every step of that is unsupported.
Common mistakes
- One "impact" number. Adding identified and confirmed counts the same time twice.
- Filling gaps. Writing 0, or a guess, where nobody gave a figure.
- Projecting. Multiplying one person's result across people or months.
Words to know
- What the reporter estimated at the start (call it identified): typed at report time; an estimate.
- What a check after the change showed (call it confirmed): from a check-in answer; after the fact.
- No figure: no number was given; never shown as zero.
Check your understanding
Try it on your own work
Describe one recurring task that wastes your time, and see these ideas applied to it.
Worksheet for this lesson: Countermeasure Plan Template (free download, no sign-up).
Where this comes from
Judging a change by what was observed after it, rather than by what was hoped, is the Check of the plan-do-check-act cycle, and respect for facts runs through the whole Lean tradition.