Three ways to say no, and the portfolio that survives

A no is a decision, and most of them are never taken

Wanagiri finished with six programmes running, two items that were never proposed, and three refusals. The refusals took the longest, because a factory has no habit for them. An approval produces a document, a budget line and an owner. A decline produces a silence, and the next salesperson walks into it eighteen months later with a new set of slides and nobody in the room who remembers why.

There are three different noes and they are not interchangeable.

Not now: the dated deferral

This is lesson 4's register: a fact, a date, a name and a default. It is the only no that has an expiry, and it is the one most often used when one of the other two is meant. The test is the first field. If nobody can name the fact that will be known, it is not a deferral.

Not by us at this price: the item whose benefit belongs to somebody else

Programme F is Meriden Row's supplier portal. Wanagiri would key its own order confirmations and shipping advices into the buyer's system, and the buyer's team would stop keying them. The benefit is real, measurable, and entirely the buyer's.

The instinct is to refuse, and the instinct is wrong. "The benefit belongs to somebody else" is not a reason to decline. It is a reason to invoice, and course 14.3 owns how that conversation is run. Wanagiri asked for USD 0.0210 a piece across the buyer's 940,000 pieces a year, which is USD 19,740 and is roughly what the work costs it. It settled at USD 0.0120, or USD 11,280 a year.

At that price F is worth USD 6,036.07 over three years if it is done first. In the portfolio it is not done first. On the sequence below it would go live at week 87.0, and from there it is worth minus USD 4,712.69, because the same annual benefit has sixty-nine weeks to run instead of a hundred and forty-seven, against costs that do not shrink.

Solve for the price that makes it worth doing at that slot and the answer is USD 0.0158 a piece.

Where an item sits in your queue changes the price at which it is worth doing at all. That is a fact about your factory, not about the buyer's budget, and the buyer cannot dispute it. That makes it the strongest thing you can bring to the conversation, and it is available to anybody who has built the sequence in lesson 2.

Not at all: the decline, with the condition that would reverse it

Programme D records defects against the operation that caused them, and it answers the third-largest loss on Wanagiri's ledger: USD 24,100 a year of rework nobody can attribute. Its own cost is modest, at USD 29,600 and 190 specification hours.

It also needs every bundle to carry an identity at operation level, and programme A was delivered at bundle level. Course 24.1 sets out at length why that choice cannot be reversed afterwards, and why bundle was the right answer for Wanagiri's mix. Re-doing A at piece level costs USD 116,000 and 214 more of Retno's hours.

Declined. And the register records what would change the answer: A's terminals reach end of life in year six, and the unit decision is open again then at no extra cost. Without that line the decline is a mood rather than a decision.

The queue decides which items exist

Here is the portfolio, ordered by benefit rate per hour of Retno's diary, exactly as lesson 2 derived it.

Every one of those eight is positive on its own. Two of them are negative in the portfolio, and nothing about them changed. They simply arrive with too little of the window left to pay for themselves. So the last two come out, the duplicate module moves up, and the answer is six items:

• eight items in the best order: USD 164,856.12
• the same eight in the order the proposals arrived: USD 142,914.77
• six items, the last two dropped: USD 183,622.94

Dropping two positive programmes added USD 18,766.82, and ordering the list properly was worth another USD 21,941.35. The six use 406 of Retno's hours and finish her involvement at week 82.1, leaving 47.4% of the three-year window with the constraint free. That is not slack going to waste. It is where the next proposal, and the recovery from the first thing that goes wrong, will have to come from.

Can you not simply buy more of the constraint?

It is the right question and Wanagiri costed it. Training a second person to specify and accept, a planner with six months of shadowing, costs USD 14,800 and 60 of Retno's hours. Those sixty hours are 11.5 weeks of delay to every surviving programme, and the six survivors together earn USD 3,779.83 a week, so the delay alone costs USD 43,613.43.

The only way to enlarge a portfolio is to spend the constraint on enlarging the constraint, and the hours come out of the front of the queue where they are worth most. Inside Sundari's three-year window that loses badly. It breaks even a little past four years and is clearly right at five.

This is the one place in the course where her window gives the wrong answer, and it is worth saying plainly rather than hiding. A convention that suits ordinary purchases is exactly wrong for an investment in capacity, whose whole return is beyond it. A factory that expects to be here in five years should train the second person, and should do it in a year when the queue is short.

What to take away

1. There are three noes: not now with a fact and a date, not at this price, and not at all with the condition that would reverse it. Using the wrong one produces a decision nobody can find later.
2. A benefit belonging to somebody else is an invoice, not a refusal, and the price at which it is worth doing depends on where it sits in your queue.
3. A permanent decline is a decision about a solution and never about the problem. Wanagiri's USD 24,100 loss stays on the ledger, unanswered, reported every year.
4. An item's value depends on its place in the queue. Two of Wanagiri's eight positive items were negative in the portfolio, and dropping them added USD 18,766.82.
5. Ordering the same list properly was worth USD 21,941.35, and the six-item portfolio finished the constraint's work by week 82.1 with 47.4% of the window free.
6. Buying more of the constraint costs the constraint. Over three years it loses. Past four it pays, which is an argument about the horizon rather than about the training.