Skip to content
Let ideas be heard and improvements last.
Gemy

How is a Kaizen gain calculated? Worked example, exchange rates, inflation and retention

In brief

How do seconds saved become hours a year, and hours become an amount? Why do estimate, measurement and finance approval stay apart? How do you compare amounts from different years? Is the gain still there a year later? The calculation rules of verified value, line by line.

29 October 2026 · 4 min read

Sample company · demo dataVerified value board: verified, open estimate, awaiting validation and eroding amount cards
Value board

A Kaizen gain is credible when it answers three questions: where does the calculation come from, was it measured, and does it last? The product's verified value module, which the company switches on, asks them in order: claim → estimate → measurement → finance validation → retention.

At the end of every Kaizen record there is a gain line, and most of the time it is an estimate. An estimate is not bad; it says how important the idea might be. The problem is adding estimates up as “gains” in the year-end report. In the Lean Enterprise Institute's definition of PDCA, the “check” step is measuring the result of the change; an unmeasured gain does not close the loop.

Three numbers, three questions

  • Estimate: what does the author expect? The calculation should be written and its assumptions visible.
  • Measurement: what happened on the floor? It should come from an indicator or an evidenced manual entry.
  • Finance approval: does the company accept this amount? Accept, correct or reject with a reason.

The difference between the three numbers is information, not failure. A measurement above the estimate says the idea's value was underrated; one below says an assumption was wrong. Both make the next Kaizen better.

Why estimate, measurement and finance approval stay apart

An improvement's value goes through three stages: it is estimated, then measured in the field, and finally approved by finance. Summing these stages into one number means presenting unrealised gains to management. The value board prevents it:

Sample company · demo dataValue validation: pending value claims, estimate and measured total
Value validation
  • Estimated, measured and validated amounts on separate cards.
  • Each record's confidence level: how strong is the evidence?
  • Persistence: does the gain continue after the measurement period?
  • Currency and unit settings: which currency, which rate?
Sample company · demo dataConfidence level: score by evidence strength and its explanation
Confidence level

Worked example (sample figures)

  • Old time 45 seconds, new time 30 seconds: 15 seconds saved per piece.
  • 400 pieces a day × 250 working days = 100,000 pieces a year.
  • 15 seconds × 100,000 pieces = 1,500,000 seconds ≈ 417 hours.
  • Amount = 417 hours × the company's hourly cost. The company enters the hourly cost; the product assumes no value.

Each line appears under “How was this calculated?”. Cash savings, time or productivity gains and avoided cost sit on separate cards and are never merged into one total. Estimated and validated amounts are not added together either.

Sample company · demo dataValue card: estimate and verified amount kept apart, confidence score and measurements
Value card

Who validates?

The person who submits a gain cannot validate it. The measurement comes from an indicator or from a manual entry with evidence; finance accepts, corrects or rejects it with a reason. The decision and its reason stay in the record; every amount labelled “validated” in the management report has this decision behind it. Validated amounts are kept in an append-only ledger.

Sample company · demo dataFinance decision: accept, correct or reject, with a reason
Finance decision screen

Verification is a matter of respect

Verifying a gain can look like doubting the author; it is the opposite. A verified gain is the author's work entering the company's official records. A gain described as “roughly this much” is forgotten at the next budget meeting; a gain approved by finance stays in the record under the author's name. Keeping the verifier separate from the submitter protects the author: nobody can say “they approved their own sums”. When the floor says “we gained” and finance says “we did not see it”, the bridge is a visible calculation: if every assumption is written down, the argument is about the assumption rather than the figure. A reasoned correction by finance is not disrespect; a rejection without reason or a silent forgetting is.

Exchange rates and inflation

Amounts from different years cannot be compared directly. The product shows values “at the prices of a selected month” using an index from the same series and base: amount × target-month index ÷ validation-month index. Watch the base: Türkiye's statistics office TÜİK replaced the 2003=100 base of the consumer price index with 2025=100 from January 2026. Values on different bases are never divided by each other. The official source for exchange rates in Türkiye is the central bank's Electronic Data Delivery System (EVDS); the rate used at validation is stored with the record, and a missing rate is shown as “rate missing”, never as zero.

Is the gain still there?

The gain is measured again 30, 90, 180 and 365 days after validation (the days and thresholds are company settings; the default thresholds are 80% and 50%). With the default settings, a retained share of 80% or more shows “Retained”, 50% to 80% “At risk”, and below that “Eroded”. With no measurement it says “Not enough data”; the gain is not counted as lost by itself. An eroded gain is not a matter for blame but a standard problem: the standard has loosened and the old habit has returned; the answer is retraining or a new Kaizen. A 0–100 confidence score is also calculated from the measurement source, evidence and finance status with a transparent formula; no AI is used. Retention checks, the confidence score and exchange-rate / inflation adjustment are features the company switches on separately.

Sources

You may copy and share this article with attribution (Gemy, https://www.gemydigital.com/en/blog/kaizen-kazanci-nasil-hesaplanir-dogrulanmis-deger).

Related posts

Cookies

Essential cookies are needed for the site to work. Analytics cookies load only if you allow them; declining is as easy as accepting. You can change your choice at any time via the footer link.

Cookie policyPrivacy notice