Time: one hour to set up at the end of the build; 30 to 45 minutes each month for twelve months. Output: a baseline that can’t be disputed, a twelve-month calendar, and a monthly review that names every slip the month it shows up.

Before you start

Chapter 20’s rule is simple and, in the book’s words, non-negotiable: track the savings for at least twelve months after the engagement ends.

The most common sustainability failure isn’t the reversion itself. It’s the moment the tracking stops. The final report goes out, the savings get posted, and the weekly and monthly comparisons quietly end with the engagement that created them. Reversion then happens in a data vacuum, where it compounds before anyone names it.

The tracking is itself a sustainability mechanism. An operation that knows it’s being measured against its improvement behaves differently from one that knows nobody is looking anymore.

The monthly review is an accountability conversation, not a report. It’s the four elements applied to the transformation itself: the outcome, the standard, the timeline and the owner. In the book, the practitioner asks for the numbers. If you’re running this internally, the person who asks should be someone who doesn’t own the numbers being reviewed: the general manager for a plant, the plant manager for a department, the operating partner for a portfolio company. In a small operation, a peer from another department works. What matters is that the person asking doesn’t own the numbers.

Results trail routines. A system can be running well before the money shows up, and the money can hold for a while after the system starts to slip. That’s why this page tracks the numbers monthly and Exercise 32 inspects the system quarterly. You need both.

Before Step 1: close out the build

Chapter 20’s first step at the end of the build is a final assessment, compared with the entry assessment. It also becomes the end-of-build picture that Exercise 32 compares against every quarter and Exercise 34 reports at month 12. Take it now, before the tracking starts.

Instrument	Re-run	At entry	At the end of the build
EMI for every department, with the calibration	Exercises 6 to 8, and 7
Plant EMI and the lowest-scoring department	Exercise 8, Step 3
MOS Gap Analysis: Present / Partial / Absent, of 29	Exercise 19
Trace the Loop	Exercise 23
Stage readiness checks and the coach-absent test	Exercise 25
Technology inventory and connections	Exercise 26, and Exercise 29, Step 5
Can operators answer the three questions?	Exercise 1, Part A, with two or three operators

Use the same question sets, department boundaries and evidence standards as the entry assessment, so the comparison is fair (Exercise 32, Step 3 explains why). File this page with Exercise 9. Exercise 32, Step 1 starts from it.

Step 1. The engagement baseline

Chapter 20 asks for three numbers before the work starts: where the operation is, what the target is, and the conservative projected return. You set the first two in Exercise 9. Add the result at the end of the build: that’s the level you’re now trying to hold.

Bring in the lines you marked Yes in Exercise 9, with their mechanisms.

Metric (from Exercise 9)	Baseline, before the build	Target (next EMI level)	Result at build end	Projected return, conservative ($ per year, realizable only)	Mechanism (sell more, remove cost, avoid penalties, defer capital)
OEE
Throughput
Unplanned downtime
First pass yield
Labor cost per unit
On-time delivery
Total

The total adds lines that name a mechanism, using the same overlap rules as Exercise 9. Deferred capital stays out of the annual total. Enter dollars per year as numbers.

Use the same overlap rules as Exercise 9. Each financial benefit must be linked to a documented operational improvement, a defined conversion mechanism, and evidence that it was actually realized. Don’t recognize the same benefit under more than one metric.

Keep deferred capital out of the annual total, as in Exercise 9. Record it on its own line as a one-time amount, reported separately from operating benefits; finance decides the treatment.

	Answer
Outcome: what the work committed to deliver, in one sentence
Standard: the total above, and the EMI level by department at build end
Timeline: twelve months from build end, through (date)
Owner of the twelve-month result (one name)
Who asks for the numbers each month (not the owner)

Step 2. The twelve-month calendar

Set every date now, at the end of the build. Dates set later tend not to get set at all.

Month	Quarterly audit (Exercise 32)	Other checks due
1
2
3	Q1 audit
4
5
6	Q2 audit	Exercise 33 (culture)
7
8
9	Q3 audit
10
11
12	Q4 audit and final assessment	Exercises 33 and 34

In “Other checks due,” add the day 91 checks for any technology you deployed or reactivated (Exercise 29, Step 6), and the quarterly coach-absent test (Exercise 25, Step 5).

A missed review is recorded, not rescheduled. If a monthly review or a quarterly audit doesn’t happen on its date, write it down as missed, with the reason. It’s a governance failure, and it’s usually the first sign the tracking is about to stop.

Step 3. Agree what “slipping” means, before month 1

Monthly numbers move. A single bad month is often noise, and chasing it wastes the review. A slow slide over three months is the pattern Chapter 20 warns about, and it’s easy to miss if every month gets explained away. Agree the rule now, while nobody’s numbers are on the line.

	Our rule
A metric is holding when
A metric is slipping when (for example: below the build-end result two months running, or outside the normal range seen during the build in any one month)
A slip is escalated to the quarterly audit when

Step 4. The monthly review

One page per month. Same numbers, same baseline, same discipline as during the build.

Month 1 review

Review date:

Metric	Baseline	Build end	This month	Holding or slipping?
OEE
Throughput
Unplanned downtime
First pass yield
Labor cost per unit
On-time delivery

Two leading signals. Bring two routine measures that move before the results do. Good choices: huddles held on every shift with four-element expectations (Exercise 16), and actionable deviations answered within their window (Exercise 22). A routine measure that’s slipping while the results still look fine is the earliest warning you’ll get.

Leading signal	Last month	This month

When the numbers are holding, keep it brief: acknowledge the result, reinforce the behavior that produced it, and name the next constraint.

When a number is slipping, the review gets harder. Name the slip, trace it, own it.

Slipping metric	Likely cause: human barrier returning, system component drifting, or process barrier not fully fixed?	Evidence	Response (outcome, standard, timeline, owner)	Expectation for next month

Month 2 review

Review date:

Metric	Baseline	Build end	This month	Holding or slipping?
OEE
Throughput
Unplanned downtime
First pass yield
Labor cost per unit
On-time delivery

Two leading signals. Bring two routine measures that move before the results do. Good choices: huddles held on every shift with four-element expectations (Exercise 16), and actionable deviations answered within their window (Exercise 22). A routine measure that’s slipping while the results still look fine is the earliest warning you’ll get.

Leading signal	Last month	This month

When the numbers are holding, keep it brief: acknowledge the result, reinforce the behavior that produced it, and name the next constraint.

When a number is slipping, the review gets harder. Name the slip, trace it, own it.

Slipping metric	Likely cause: human barrier returning, system component drifting, or process barrier not fully fixed?	Evidence	Response (outcome, standard, timeline, owner)	Expectation for next month

Month 3 review

Review date:

Metric	Baseline	Build end	This month	Holding or slipping?
OEE
Throughput
Unplanned downtime
First pass yield
Labor cost per unit
On-time delivery

Two leading signals. Bring two routine measures that move before the results do. Good choices: huddles held on every shift with four-element expectations (Exercise 16), and actionable deviations answered within their window (Exercise 22). A routine measure that’s slipping while the results still look fine is the earliest warning you’ll get.

Leading signal	Last month	This month

When the numbers are holding, keep it brief: acknowledge the result, reinforce the behavior that produced it, and name the next constraint.

When a number is slipping, the review gets harder. Name the slip, trace it, own it.

Slipping metric	Likely cause: human barrier returning, system component drifting, or process barrier not fully fixed?	Evidence	Response (outcome, standard, timeline, owner)	Expectation for next month

Month 4 review

Review date:

Metric	Baseline	Build end	This month	Holding or slipping?
OEE
Throughput
Unplanned downtime
First pass yield
Labor cost per unit
On-time delivery

Two leading signals. Bring two routine measures that move before the results do. Good choices: huddles held on every shift with four-element expectations (Exercise 16), and actionable deviations answered within their window (Exercise 22). A routine measure that’s slipping while the results still look fine is the earliest warning you’ll get.

Leading signal	Last month	This month

When the numbers are holding, keep it brief: acknowledge the result, reinforce the behavior that produced it, and name the next constraint.

When a number is slipping, the review gets harder. Name the slip, trace it, own it.

Slipping metric	Likely cause: human barrier returning, system component drifting, or process barrier not fully fixed?	Evidence	Response (outcome, standard, timeline, owner)	Expectation for next month

Month 5 review

Review date:

Metric	Baseline	Build end	This month	Holding or slipping?
OEE
Throughput
Unplanned downtime
First pass yield
Labor cost per unit
On-time delivery

Two leading signals. Bring two routine measures that move before the results do. Good choices: huddles held on every shift with four-element expectations (Exercise 16), and actionable deviations answered within their window (Exercise 22). A routine measure that’s slipping while the results still look fine is the earliest warning you’ll get.

Leading signal	Last month	This month

When the numbers are holding, keep it brief: acknowledge the result, reinforce the behavior that produced it, and name the next constraint.

When a number is slipping, the review gets harder. Name the slip, trace it, own it.

Slipping metric	Likely cause: human barrier returning, system component drifting, or process barrier not fully fixed?	Evidence	Response (outcome, standard, timeline, owner)	Expectation for next month

Month 6 review

Review date:

Metric	Baseline	Build end	This month	Holding or slipping?
OEE
Throughput
Unplanned downtime
First pass yield
Labor cost per unit
On-time delivery

Two leading signals. Bring two routine measures that move before the results do. Good choices: huddles held on every shift with four-element expectations (Exercise 16), and actionable deviations answered within their window (Exercise 22). A routine measure that’s slipping while the results still look fine is the earliest warning you’ll get.

Leading signal	Last month	This month

When the numbers are holding, keep it brief: acknowledge the result, reinforce the behavior that produced it, and name the next constraint.

When a number is slipping, the review gets harder. Name the slip, trace it, own it.

Slipping metric	Likely cause: human barrier returning, system component drifting, or process barrier not fully fixed?	Evidence	Response (outcome, standard, timeline, owner)	Expectation for next month

Month 7 review

Review date:

Metric	Baseline	Build end	This month	Holding or slipping?
OEE
Throughput
Unplanned downtime
First pass yield
Labor cost per unit
On-time delivery

Two leading signals. Bring two routine measures that move before the results do. Good choices: huddles held on every shift with four-element expectations (Exercise 16), and actionable deviations answered within their window (Exercise 22). A routine measure that’s slipping while the results still look fine is the earliest warning you’ll get.

Leading signal	Last month	This month

When the numbers are holding, keep it brief: acknowledge the result, reinforce the behavior that produced it, and name the next constraint.

When a number is slipping, the review gets harder. Name the slip, trace it, own it.

Slipping metric	Likely cause: human barrier returning, system component drifting, or process barrier not fully fixed?	Evidence	Response (outcome, standard, timeline, owner)	Expectation for next month

Month 8 review

Review date:

Metric	Baseline	Build end	This month	Holding or slipping?
OEE
Throughput
Unplanned downtime
First pass yield
Labor cost per unit
On-time delivery

Two leading signals. Bring two routine measures that move before the results do. Good choices: huddles held on every shift with four-element expectations (Exercise 16), and actionable deviations answered within their window (Exercise 22). A routine measure that’s slipping while the results still look fine is the earliest warning you’ll get.

Leading signal	Last month	This month

When the numbers are holding, keep it brief: acknowledge the result, reinforce the behavior that produced it, and name the next constraint.

When a number is slipping, the review gets harder. Name the slip, trace it, own it.

Slipping metric	Likely cause: human barrier returning, system component drifting, or process barrier not fully fixed?	Evidence	Response (outcome, standard, timeline, owner)	Expectation for next month

Month 9 review

Review date:

Metric	Baseline	Build end	This month	Holding or slipping?
OEE
Throughput
Unplanned downtime
First pass yield
Labor cost per unit
On-time delivery

Two leading signals. Bring two routine measures that move before the results do. Good choices: huddles held on every shift with four-element expectations (Exercise 16), and actionable deviations answered within their window (Exercise 22). A routine measure that’s slipping while the results still look fine is the earliest warning you’ll get.

Leading signal	Last month	This month

When the numbers are holding, keep it brief: acknowledge the result, reinforce the behavior that produced it, and name the next constraint.

When a number is slipping, the review gets harder. Name the slip, trace it, own it.

Slipping metric	Likely cause: human barrier returning, system component drifting, or process barrier not fully fixed?	Evidence	Response (outcome, standard, timeline, owner)	Expectation for next month

Month 10 review

Review date:

Metric	Baseline	Build end	This month	Holding or slipping?
OEE
Throughput
Unplanned downtime
First pass yield
Labor cost per unit
On-time delivery

Two leading signals. Bring two routine measures that move before the results do. Good choices: huddles held on every shift with four-element expectations (Exercise 16), and actionable deviations answered within their window (Exercise 22). A routine measure that’s slipping while the results still look fine is the earliest warning you’ll get.

Leading signal	Last month	This month

When the numbers are holding, keep it brief: acknowledge the result, reinforce the behavior that produced it, and name the next constraint.

When a number is slipping, the review gets harder. Name the slip, trace it, own it.

Slipping metric	Likely cause: human barrier returning, system component drifting, or process barrier not fully fixed?	Evidence	Response (outcome, standard, timeline, owner)	Expectation for next month

Month 11 review

Review date:

Metric	Baseline	Build end	This month	Holding or slipping?
OEE
Throughput
Unplanned downtime
First pass yield
Labor cost per unit
On-time delivery

Two leading signals. Bring two routine measures that move before the results do. Good choices: huddles held on every shift with four-element expectations (Exercise 16), and actionable deviations answered within their window (Exercise 22). A routine measure that’s slipping while the results still look fine is the earliest warning you’ll get.

Leading signal	Last month	This month

When the numbers are holding, keep it brief: acknowledge the result, reinforce the behavior that produced it, and name the next constraint.

When a number is slipping, the review gets harder. Name the slip, trace it, own it.

Slipping metric	Likely cause: human barrier returning, system component drifting, or process barrier not fully fixed?	Evidence	Response (outcome, standard, timeline, owner)	Expectation for next month

Month 12 review

Review date:

Metric	Baseline	Build end	This month	Holding or slipping?
OEE
Throughput
Unplanned downtime
First pass yield
Labor cost per unit
On-time delivery

Two leading signals. Bring two routine measures that move before the results do. Good choices: huddles held on every shift with four-element expectations (Exercise 16), and actionable deviations answered within their window (Exercise 22). A routine measure that’s slipping while the results still look fine is the earliest warning you’ll get.

Leading signal	Last month	This month

When the numbers are holding, keep it brief: acknowledge the result, reinforce the behavior that produced it, and name the next constraint.

When a number is slipping, the review gets harder. Name the slip, trace it, own it.

Slipping metric	Likely cause: human barrier returning, system component drifting, or process barrier not fully fixed?	Evidence	Response (outcome, standard, timeline, owner)	Expectation for next month

Step 5. Month twelve

At month twelve, close the loop with an honest number.

Metric	Baseline	Build end	Month 12	Share of the improvement held: (month 12 minus baseline) divided by (build end minus baseline)

The share calculates from the three numbers, with no cap at 100%. Enter each metric in its own unit.

For metrics where lower is better, such as downtime or labor cost per unit, the same formula works: both differences are negative, so the share is still positive.

Result	What it means
Build end equals baseline	Report “N/A: no measurable improvement at build end.” The formula doesn’t apply.
Below 0%	Performance is worse than the original baseline
Exactly 0%	None of the original improvement remains
Above 0% through 100%	Some or all of the original improvement remains; the number says how much
Above 100%	Performance has improved beyond the build-end result

Don’t cap the result at 100%. Doing so hides improvement that continued after the build.

	Answer
Realized return over twelve months ($, with evidence for each line), against the projected return in Step 1
As a percentage of the projection
EMI level by department: entry, build end, month 12 (from Exercise 32)
What held, in one sentence
What didn’t, and why

The share held and the realized return answer different questions. The share held is operational, measured at one point: month twelve. The realized return is financial, summed over all twelve months. Report both; one doesn’t stand in for the other.

Report the real percentage, even when it’s 60 and not 100. Chapter 22 calls that data courage. Chapter 20 says it’s how the work earns the right to the next engagement: not by being compelling in the presentation, but by being honest in the measurement.

Diagnostic or permanent? Permanent for twelve months. Many operations keep the monthly review after that, as the operation’s own management review.

How you’ll know it’s working: all twelve monthly reviews happen on their dates (routine measure); every slip is named, traced and owned in the month it appears (behavioral measure); the share of the improvement held at month twelve is reported honestly, and is higher than the operation managed last time (results measure).

Ties to: Chapter 20; Exercise 9 (the Opportunity Gap baseline); Exercises 16 and 22 (leading signals); Exercise 29 (day 91 checks); Exercise 32 (the quarterly audit); Appendix A (the Opportunity Gap as a quarterly number); Appendix D, Stage 4 (monthly review).