From a request for new plant to the outcome behind it

Most projects arrive as a request for a thing: a machine, a building, a system, a refit. The request is usually sincere and often well researched, and it carries an unstated claim that the thing, once delivered, will cause something the business needs. This chapter separates the thing from what it is meant to cause. It takes one request for new plant, asks what must be different when the work is finished, names that outcome in four measures, and then makes a first, cheap test of whether the requested plant could deliver the outcome at all.

The test takes an afternoon with records the business already holds, and it changes the direction of the whole project. The chapter ends with the brief that went to vendors before the test was made, what came back, and how the brief was rewritten.

The request as it arrived

This is an illustration. A family-owned commercial laundry in a regional New South Wales city, run by its second generation, washes, dries, irons and delivers linen for a group of four private hospitals and for 22 hotels and serviced apartments. It processes 60,000 kg of clean linen a week over six production days, 3,120,000 kg a year, with eight washer-extractors, a bank of dryers and two flatwork ironer lines, the older of which dates from 2009. It employs 42 full-time-equivalent staff. The two sibling owners have watched competitors install tunnel washers, have taken calls from vendors, and know their wash room is crowded on busy days. They ask for “a new plant”.

Their brief lists five items. All amounts in this book are Australian dollars excluding GST.

• T, a tunnel washer with 12 compartments, a press and a shuttle: A$3,400,000.
• D, three new gas dryers: A$450,000.
• E, a 600 m² building extension to house the tunnel: A$1,600,000.
• R, RFID linen tracking: A$420,000.
• M, refurbished staff amenities and office: A$250,000.

The total is A$6,120,000, slightly more than the laundry’s whole annual revenue of A$6,052,800. Nothing in the list is unreasonable on its own. The vendors say a tunnel uses less water and gas per kilogram than the present machines, the dryers are old, and the amenities are tired. What the list does not say is what the business will be able to do afterwards that it cannot do now, how anyone would know, or by when that must be true. Those three gaps are where this book starts.

Outputs, outcomes and benefits

Guidance on benefits management draws a line between three things that a request tends to run together. The UK Infrastructure and Projects Authority’s Guide for Effective Benefits Management in Major Projects (2017) separates the outputs a project delivers, the outcomes that follow when those outputs are put to use, and the benefits those outcomes produce. In substance, it defines a benefit as a measurable improvement that results from an outcome, that one or more stakeholders see as an advantage, and that contributes to one or more of the organisation’s objectives.

Applied to the laundry, the three layers look like this:

• An output is something the project hands over: a tunnel washer commissioned, an extension with its occupancy certificate, an RFID system live.
• An outcome is a change in how the business performs once the outputs are in use: hospital linen signed for before the 06:00 cut-off, complete; fewer litres of water per kilogram washed.
• A benefit is the measurable improvement that matters to someone with a stake in the business: the hospital contract retained at the new price, and a margin the family can live on.

The distinction matters because each layer can fail on its own. A tunnel can be commissioned on time and on budget, an output success, while deliveries stay late, an outcome failure, and the contract is lost, a benefit failure. When the project succeeds and the strategy fails works through that pattern after approval, when it is already expensive. This book applies the same distinction before approval, when acting on it costs little.

A request almost always names outputs. Sponsors find outputs easy to picture, vendors sell them, and progress on them is easy to report. Outcomes and benefits are harder to see and slower to arrive, which is exactly why they need to be written down first.

Asking what must be different

The useful question to put to any request is what must be different when it is finished, and by when. The owners’ first answers were about the plant itself: newer, faster, using less water. Asked what would happen if they did nothing, they named the event behind the request.

The hospital group sends 55% of the laundry’s volume, 33,000 kg a week or 1,716,000 kg a year, at A$1.85/kg. Its contract is to be re-tendered. The tender closes at month 12 from now, and the current contract ends at month 18. The group has signalled three things:

1. It expects a lower price. The laundry’s own estimate is A$1.76/kg, 4.9% less.
2. It will require 98% of deliveries on time and in full (OTIF) over the trailing 26 weeks, and will ask for the evidence with the tender.
3. It will score bidders on the water and energy they use per kilogram of linen.

The 22 hotels and serviced apartments, the other 45% of volume, pay A$2.05/kg and have no tender pending, but they receive the same late and short deliveries and some of them have begun to say so.

With that, the request can be restated. The business does not need new plant as such. It needs to win the hospital tender at the lower price, which means showing 98% OTIF over the 26 weeks before the tender closes, at a cost that still leaves a margin, with water and energy figures that score well. That is an outcome. The tunnel washer is one possible means to it, and nobody has yet checked whether it is a good one.

Figure 1The request and the outcome statement

	As requested	As an outcome
What is asked for	A new plant: tunnel washer, dryers, extension, RFID, amenities	Keep the hospital contract through the month-12 tender at a lower price
Success looks like	Plant installed and commissioned	98% of deliveries on time and in full over the 26 weeks before the tender
Money	A$6,120,000 of capital	A cost per clean kg low enough to hold the margin at the new price
Water and energy	Newer machines use less	Litres and megajoules per clean kg at or below the tender's scoring thresholds
Deadline	Whenever the plant is ready	Evidence in hand before the tender closes at month 12
How a proposal is judged	Vendors' quotes and specifications	How far it moves all four measures, and how soon

The four outcome measures

The outcome statement contains four things the business can measure and the hospital group will judge:

1. On-time-in-full delivery (OTIF): the share of deliveries signed for before the customer’s cut-off and complete.
2. Cost per clean kilogram: what it costs, fully loaded, to process and deliver one kilogram of clean linen.
3. Water intensity: litres of water per clean kilogram.
4. Energy intensity: megajoules of energy per clean kilogram.

Chapter 3 defines each measure tightly enough that it cannot drift, Chapter 4 measures where the business stands on each, and Chapter 5 derives each target from the purpose. One number is worth having now, because it frames everything else. Over 26 weeks of customer-signed delivery dockets, 4,680 deliveries in all, 412 failed: OTIF was 91.2% against the 98% the tender will require.

There are four measures, not one, because each guards against a different way of winning the wrong game. An option that lifts OTIF by adding people and vans may raise the cost per kilogram beyond what the new price can carry. An option that cuts water may do nothing for delivery. An option that cuts cost by running fewer, fuller loads may make deliveries later. Scoring every option on all four keeps each one honest against the whole outcome. Four is also few enough to track every month, with one person responsible for each.

None of the four mentions a tunnel washer, an extension or RFID. They are written so that any option, including ones nobody has proposed yet, can be scored on them. That is the practical test of whether a measure describes an outcome or an output: if it can only be met by one particular piece of equipment, it is a specification in disguise.

Working backwards in practice

The habit of starting from the result and working back to the means has several published forms. Amazon’s Leadership Principles state it under customer obsession: “Leaders start with the customer and work backwards.” Colin Bryar and Bill Carr, in Working Backwards (2021), describe how the company turned that into a routine. Before a product is built, the team writes the press release that would announce it and a set of frequently asked questions, with answers, that customers and executives would raise. If the release cannot say plainly why a customer would care, the product is not ready to build. The documents are cheap to write and cheap to throw away, and they expose a weak purpose before anyone spends on design.

Public infrastructure guidance reaches the same order by another route. Infrastructure Australia’s Assessment Framework (2021) runs in four stages: defining problems and opportunities, identifying and analysing options from a longlist to a shortlist, developing a business case, and a post-completion review. Its first stage asks for the problem to be defined and evidenced before any option is put forward. Infrastructure Australia has noted that the framework is changing, but the order of problem before options is what matters here.

For a family laundry, the equivalent of the press release is one page the owners could hand to the hospital group’s procurement manager, saying what will be different for the four hospitals by the time the tender closes, in the group’s own measures. If that page cannot be written about the tunnel washer, the tunnel washer is not yet the answer. Plan slowly, deliver quickly: where a project’s money is exposed in Front-End Planning by the Numbers shows why this definition work belongs at the start: weeks spent defining a project cost little, while weeks spent delivering the wrong one cost a great deal. One project or twelve: starting from the outcome in Modular Projects by the Numbers gives a second example, twelve packing cells defined by cases a minute and pallets an hour before any machine was chosen.

A first test of the request

The cheapest test of a request is to ask how much of the gap it could close if it worked perfectly. That needs the failures sorted by cause. The operations manager went through each of the 412 failed dockets with the dispatch supervisor and the production records for that day, and coded each failure to one primary cause: the first step in the process, counting back from the customer, that made the delivery late or incomplete.

• ironer backlog on peak days, flatwork not ironed in time for loading: 157 (38.1%);
• breakdowns of ironer line 1: 74 (18.0%);
• dispatch, loading and routes: 111 (26.9%);
• wash capacity, washers full and soiled linen waiting: 37 (9.0%);
• soiled linen received late from the customer: 33 (8.0%).

Figure 2412 failed deliveries in 26 weeks, by primary cause

Ironer backlog on peak days157 failures38.1%

Dispatch, loading and routes111 failures26.9%

Ironer line 1 breakdowns74 failures18.0%

Wash capacity37 failures9.0%

Soiled linen received late33 failures8.0%

The tunnel washer, dryers and extension address washing. Suppose they worked perfectly from the first day and removed every one of the 37 wash-capacity failures. There would still be 375 failures in 4,680 deliveries, and OTIF would rise from 91.2% to 92.0%. The other 91.0% of failures arise after washing or have nothing to do with it. A faster wash room would, if anything, deliver damp linen sooner to ironers that are already the bottleneck on peak days.

This is not yet a full diagnosis. A primary cause can hide a second one behind it, 26 weeks may miss the worst season, and some failures had more than one cause. Chapter 6 works the causes through properly, with capacity figures for the ironers and the washers. But the first test does not need precision to be decisive. Even with generous allowance for misclassified dockets, washing is too small a share of the problem for the requested plant to carry the outcome. If a third of the 342 dispatch and ironer failures had really been wash failures, a perfect wash plant would remove 37 + 114 = 151 failures and the ceiling would be 94.4%, still well short of 98%.

There is a second, independent problem: timing. The evidence the hospital group will ask for covers the 26 weeks before the tender closes. Plant that needs a building extension and a development approval is unlikely to be installed, let alone settled, inside that window. Chapter 8 puts numbers on the timing, and on an option derived from the outcome instead of from the request.

The error: a brief written as a deliverable

This is an illustration. Before the docket analysis was done, the owners’ first brief went to three equipment vendors. It read, in full: “Tunnel washer, 1,200 kg/h, with press, shuttle and dryers; building works to suit.” It described a deliverable, with a capacity figure taken from a brochure.

The vendors did what the brief asked. Their quotes came back for washing capacity: tunnel sizes, compartment counts, water reuse, press cycles, and payback calculations based on labour saved in the wash room and on water and gas per kilogram. Every proposal was competent and every payback looked attractive. None mentioned ironing, dispatch or delivery times, because the brief did not.

The problem surfaced when the operations manager set the quotes beside the docket analysis. A perfect wash plant could remove at most 37 of the 412 failures, so the OTIF ceiling with the requested plant was 92.0% against a requirement of 98%. The quotes were careful answers to a question that did not need answering first. Had the owners signed one, they would have spent most of A$6.12 million and arrived at the tender with deliveries almost exactly as late as before.

The correction was not to discard the vendors’ work but to rewrite the brief. The second version stated the outcome and the four measures, the date by which evidence was needed, and the constraint that hospital linen must meet hygiene requirements throughout. It invited proposals for any change to plant, process or arrangements with customers that would move those measures, and asked each proposer to say which failures it would remove and what it would do to cost, water and energy per kilogram. A brief written that way can still lead to a tunnel washer. It cannot lead to one by default.

Figure 3Tests of a project brief

• Do: States what must be different afterwardsthe outcome, not the equipment
• Do: Names the measures and how each is calculatedOTIF, cost per clean kg, L/kg, MJ/kg
• Do: Gives the date by which the outcome must be shownevidence in hand before the month-12 tender
• Do: Admits any means that would workplant, process or customer arrangements
• Do: Asks each proposal for its effect on every measurefailures removed, A$/kg, L/kg, MJ/kg
• Avoid: Specifies a machine and a rate before the cause is known"tunnel washer, 1,200 kg/h"

The sequence this book follows

The rest of the book does for the laundry what the rewritten brief asks of vendors, in a fixed order. Each step uses only what the steps before it have established, and the order is deliberate. Measures come before baselines, because a baseline in a loosely defined measure is a number of unknown meaning. Targets come after baselines but are derived from the purpose, not from the baseline. Options come only after the causes of the gap are known, so that each option is aimed at a cause.

Figure 4The outcome-first sequence, chapter by chapter

1. 1Chapter 1: from request to outcomeseparate the deliverable from what it must cause
2. 2Chapter 2: ends, means and valuethe objectives hierarchy and the value at stake
3. 3Chapter 3: measures from questionsfour measures defined so they cannot drift
4. 4Chapter 4: baselineswhere the business stands, with its variation
5. 5Chapter 5: targetsvalues derived from the purpose, with periods and dates
6. 6Chapter 6: causeswhere the gap comes from, by Pareto and capacity
7. 7Chapter 7: benefits mapped to changesbusiness changes, enablers, owners and dis-benefits
8. 8Chapter 8: optionsevery option scored against all four targets
9. 9Chapter 9: requirements and scopeevery item traced to a target
10. 10Chapter 10: acceptance and trackingthe same measures after go-live

The test of the method is visible. If the option derived from the outcome turns out to be the one requested, the work has still been worth doing, because the owners will know why their plant is right and what it must achieve. In this case it is not: the option that meets all four targets costs about A$2.44 million and looks quite different. The next chapter, Ends, means and the value at stake, sorts the owners’ aims into ends and means and puts a dollar value on the hospital contract. Working back from the gap to its causes returns to the 412 failures with capacity figures.

Exercise: removing causes one group at a time

Using the 412 failures by primary cause, calculate OTIF over the 4,680 deliveries if (a) only the wash-capacity failures were removed, (b) only the ironer failures, backlog and breakdowns together, were removed, and (c) both groups were removed. Compare each result with the 98% requirement, which allows at most 93 failures in 26 weeks (Chapter 5 derives that limit), and say what the comparison implies for the scope of the project.

Worked answer. (a) 412 − 37 = 375 failures, so OTIF is (4,680 − 375) ÷ 4,680 = 91.99%. (b) Ironer failures are 157 + 74 = 231, leaving 181, so OTIF is (4,680 − 181) ÷ 4,680 = 96.13%. (c) 412 − 37 − 231 = 144 failures, so OTIF is (4,680 − 144) ÷ 4,680 = 96.92%. Even with every wash and ironer failure gone, 144 − 93 = 51 failures more than the limit remain. Dispatch, at 111 failures, must change as well, and probably the arrangements for soiled linen that arrives late. The requested plant acts only on the wash group, 9.0% of the failures, so on its own it cannot meet the delivery requirement however well it performs.