Calculating Real Kaizen ROI (and What to Leave Out)

The board deck shows ₹3.2 crore in kaizen savings for FY25. The CFO opens the underlying spreadsheet. By tab 4, he's found ₹40 lakh of double-counting. By tab 7, he's removed ₹65 lakh of "estimated" savings with no verification. By the end of the meeting, the real number is ₹1.9 crore — and the kaizen programme's credibility is gone.

This is the most common quiet disaster in Indian plant kaizen tracking. The math looks impressive on the wall poster. It does not survive contact with anyone who knows accounting. Here's the formula that does.

The kaizen ROI formula that holds up

For each kaizen, only count what passes all four tests:

Verified Kaizen Saving (₹) =
    (Measured baseline cost) - (Measured post-implementation cost)
    [over the same measurement window]
    [minus implementation cost amortised over benefit life]
    [signed off by Finance designate, not the kaizen owner]

Five components, each non-negotiable:

1. Measured baseline. You have actual numbers from before the change. "We estimate scrap was around 4%" is not a baseline. "Scrap was 4.2% across 6,400 pieces in Jan" is a baseline.
2. Measured post-state. Same metric, same window length, same conditions. If your baseline was a normal month, your post-measurement is also a normal month — not the week after roll-out when everyone is hyper-focused.
3. Implementation cost subtracted. The jig cost ₹35,000. The cabling cost ₹8,000. Two engineers spent a week on it. Net the cost out of the saving.
4. Amortised over benefit life. If the saving runs for 2 years before the next equipment refresh, amortise implementation cost over 24 months. Don't pretend the saving starts at Month 1 and the cost is free.
5. Verified by someone who isn't the kaizen owner. Finance designate, plant manager, or a peer from another line. The person submitting savings should not be the person approving them.

The 5 categories of fake savings to stop counting

1. The "if we extrapolated this..." savings

"We saved 15 seconds per cycle. We do 800 cycles per shift, 2 shifts, 25 days a month, 12 months — that's 100 days of labour saved!"

Did you actually reduce labour? Did you let someone go, or stop hiring, or redeploy to another line? If no — the 15 seconds went somewhere, but it didn't go to your bottom line. Time saved is not money saved unless the freed time produces measurable output.

2. The "annualised from one good month" savings

One month after the kaizen, scrap was ₹12k lower than the prior month. Annualised: ₹1.44 lakh saved.

Wrong. Scrap varies month to month for reasons that have nothing to do with the kaizen — material lot quality, operator turnover, season. You need at least 3 months of post-data, ideally 6, before extrapolating. The pilot-month bump usually fades.

3. The "double-counted" savings

Productivity team claims ₹8 lakh from cycle-time reduction on Line 2. Maintenance team claims ₹6 lakh from the same machine running more uptime hours. Both numbers go into the kaizen tracker. Result: the same machine improvement counted twice.

Fix: every kaizen logs the cost centre, the metric, and the window. Finance can dedupe by overlap. If two improvements act on the same metric in the same window, only the larger one counts.

4. The "avoided cost" without baseline

"This safety improvement prevented a serious injury. A serious injury would cost ₹15 lakh in medical, legal, downtime. Saving: ₹15 lakh."

Avoided cost is a real category — but only when you have a frequency baseline. "We had 2 serious injuries last year averaging ₹14 lakh each. This kaizen targets that mechanism. If we achieve zero, that's a ₹28 lakh annual avoidance." Now you have a comparable. Without the baseline, avoided cost is wishful thinking dressed as math.

5. The "soft" savings — morale, image, team spirit

Real benefits, real outcomes. Do not put them in the ROI number. Track them separately under qualitative impact. The moment soft savings hit the rupee column, your CFO will discount the entire program.

An opinionated take

Plants should report HALF the savings they think they're getting, and report them with verification stamps from Finance. Discipline beats optics every time. The plant that claims ₹3 crore unverified loses credibility in one CFO meeting. The plant that claims ₹1.5 crore with every entry verified, baselined, and amortised wins the budget conversation forever. The CFO will then defend your programme to the board, instead of cutting it in the next quarterly review.

A worked example — small, real, complete

Field	Value
Kaizen	Relocate raw-material bin closer to assembly station on Line 4
Baseline measurement (Feb)	Average cycle 92 seconds across 4,800 pieces; non-value-added walk time 11 sec/cycle
Implementation	Bin relocation, new mounting bracket — ₹4,200 in material + 8 hours labour (₹1,800)
Post measurement (Apr, May, Jun — 3 months)	Cycle 84 sec; walk time 4 sec. Stable across all 3 months.
Output increase	From this freed cycle time, Line 4 redeployed to add 2,400 more pieces per month (verified — production records)
Contribution per piece	₹28 (per cost accounting)
Monthly verified saving	2,400 × ₹28 = ₹67,200
Implementation cost amortised	₹6,000 / 24 months = ₹250/month
Net monthly saving	₹67,200 - ₹250 = ₹66,950
Verified by	Plant Finance (R. Sharma) on 12-Jun-2026
Annual saving (if pattern holds — flagged, not extrapolated)	₹66,950 × 12 = ₹8.03 lakh (subject to monthly re-verification)

This is a kaizen that survives a CFO conversation. Notice: no extrapolation, no double-count, costs subtracted, Finance signed off. The annualised figure is flagged as a projection, not a verified saving — the verified saving is per-month, accumulated.

What we built

The kaizen module in LeanOS enforces this discipline structurally. Each kaizen requires baseline, post-state, implementation cost. Monthly savings are logged as ledger entries — not "annual × 12" — so the dashboard shows actual accumulated verified rupees per period. Finance can approve or reject each entry. Double-count detection flags overlapping cost-centre / metric / window combinations.

Factory ABC — auto-component plant in Pune, 200 employees — tracked ₹15.7 lakh in verified savings over 90 days on LeanOS. Every entry per-month, every entry verified, every entry survivable to a CFO walkthrough.

Calculate what your plant would credibly track — uses real pilot benchmarks, not vendor marketing.