Stages and gates: organising the front end

Twenty-two weeks of front-end work is not one long study. Left unstructured, it drifts: design starts before the option is chosen, studies are commissioned because someone thought of them, and the estimate is refreshed whenever a board meeting approaches. Structured, it is a short sequence of stages, each with a question to answer, a budget, a list of things it must produce and a gate at which someone decides, on that evidence, whether to spend the next stage’s money.

This chapter sets out the usual three front-end stages and their names, places them beside two Australian public-sector frameworks, and shows what each of Northbank’s stages produced and cost. It works through the three options compared at Gate 2, lists what each gate asks for, and ends with an options stage that, in its first draft, had only one option in it.

The front end as stages

The idea of dividing development into stages separated by decision points has a long history. Cooper (1990), in Business Horizons, set out the stage-gate model for new product development: work proceeds in stages, and between them sit gates where management decides, against stated criteria, whether to continue, stop, hold or rework. Capital projects adopted the same structure for the work before construction.

Independent Project Analysis, the project benchmarking firm whose records Merrow (2011) draws on, names the three stages of the capital project front end FEL 1, business planning, FEL 2, facility planning and FEL 3, project planning, where FEL stands for front-end loading. The names describe what each stage settles:

• FEL 1, business planning, asks whether there is a business problem or opportunity worth a project, what outcome it must deliver and roughly what it would cost and earn.
• FEL 2, facility planning, asks which way of meeting that outcome is best. It generates and compares options, investigates whatever would change the choice, and ends with one option selected.
• FEL 3, project planning, defines the selected option well enough to commit to it: design developed to the point where it can be priced and tendered, the site investigated, approvals in train, a delivery strategy, a schedule and an estimate the business is prepared to fund.

Two Australian frameworks have the same shape. Infrastructure Australia’s Assessment Framework (2021) moves a proposal through four stages: defining the problem or opportunity, identifying and analysing options, developing a business case, and post-completion review. Its second stage deliberately filters a long list of options down to a short list, so that unpromising options are not carried into a costly business case. Victoria’s Gateway Review Process, run by the Department of Treasury and Finance, places independent reviews at six gates: concept and feasibility, business case, readiness for market, tender decision, readiness for service, and benefits. The first three cover ground that a private owner’s front end also covers. Northbank is a family business and needs none of the paperwork, but it borrows two ideas: narrow the options early, and have someone outside the team look at the evidence before the big commitment.

What each Northbank stage produced

This is an illustration. Northbank’s front end ran as three stages. The concept stage, FEL 1, finished before week 0 (Monday 2 November 2026) and cost A$22k, money already spent and outside every total in this book. It produced a statement of the outcome the business wanted, a sketch of four rooms and a grader beside the existing packhouse, and the Class 5 estimate of A$5,410k, all amounts in Australian dollars excluding GST. Gate 1 was the decision to spend on FEL 2 rather than approve the concept at once, the choice Chapter 1 described.

Figure 1Northbank's front end, stage by stage

1. 1FEL 1, business planningconcept, outcome statement, Class 5 estimate A$5,410k; A$22k, complete before week 0
2. 2Gate 1fund FEL 2, or approve the concept now; FEL 2 funded
3. 3FEL 2, facility planningweeks 0–10, A$96k: options compared, electrical capacity, plant condition, grader trials, Class 4 estimate A$5,960k
4. 4Gate 2, week 10option (a) selected; FEL 3 plan and budget approved
5. 5FEL 3, project planningweeks 10–22, A$152k: design development, ground, permit, fire basis, refrigeration duty, long-lead tenders, Class 3 estimate A$6,428k
6. 6Gate 3, week 22approval: delivery budget A$6,700k, long-lead packages let

FEL 2 ran for 10 weeks, from week 0 to Gate 2 on Monday 11 January 2027. A contract project engineer, engaged at A$3.0k a week, ran the stage. A refrigeration consultant compared the options and assessed the old plant; the grader supplier’s equipment was seen running and tested on Northbank’s own fruit; a surveyor measured the site; an electrical engineer checked whether the existing supply could carry the new load; and a concept designer turned the sketch into a layout for each option. The stage ended with an option selected and a Class 4 estimate of A$5,960k, which already included two items the concept had missed: an electrical supply upgrade of A$380k and A$170k of extra refrigeration capacity.

FEL 3 ran for 12 weeks, from week 10 to Gate 3 on Monday 5 April 2027. It developed the design, investigated the ground, lodged the planning permit application, set the fire engineering basis, specified the refrigeration duty, tendered the two long-lead packages and had a quantity surveyor prepare the Class 3 estimate of A$6,428k, which includes the A$248k front end.

The stage lengths were not arbitrary. They were set backwards from the approvals that take longest. The planning permit was expected to take 16 weeks, so the application was lodged in week 8, two weeks before Gate 2, once the options study had pointed to option (a), at the risk of amending it if the gate chose otherwise. That way the permit would be available in week 24, just as site works began. The distributor needed about 8 weeks to make a connection offer, so the application went in at week 12, two weeks into FEL 3, and the offer arrived in week 20, in time to be accepted at approval. A front end that runs its stages in strict sequence, waiting for each gate before starting anything in the next, can add months that no study needed. Northbank started the slow external processes early, at its own risk, and kept the decisions for the gates.

What the stages cost

FEL 2 cost A$96k, about A$9.6k a week, and FEL 3 cost A$152k, about A$12.7k a week. FEL 3 is dearer because design development, investigations and tenders cost more than comparing options. Together they make the A$248k front end, 3.9% of the A$6,428k total.

Figure 2Where the A$248k front end went

1. FEL 2 project engineer, 10 weeksA$30k · 12%
2. Options study, refrigeration consultantA$18k · 7%
3. Grader supplier visits and fruit trialsA$8k · 3%
4. Site surveyA$6k · 2%
5. Electrical capacity studyA$9k · 4%
6. Concept designA$25k · 10%
7. FEL 3 project engineer, 12 weeksA$36k · 15%
8. Design developmentA$48k · 19%
9. Geotechnical investigationA$14k · 6%
10. Planning permit application and reportsA$12k · 5%
11. Fire engineering basisA$11k · 4%
12. Refrigeration duty study and specificationA$12k · 5%
13. Tenders for two long-lead packagesA$10k · 4%
14. Class 3 estimate, quantity surveyorA$9k · 4%

The figure answers a question every owner should ask of a front-end budget: what is the money for? About a quarter of it, A$66k, is the project engineer’s time over 22 weeks. Design takes A$73k across the two stages. The rest buys specific answers: whether the supply is big enough, what the ground will carry, what the fire engineer and the building surveyor will require, what the plant must do, and what the market will charge for the long-lead equipment. Chapter 5 shows what those answers were worth.

Staging also limits what is at risk at each decision. If the business had stopped at Gate 2, it would have spent A$118k: the A$22k concept and the A$96k of FEL 2. If it had stopped at Gate 3, before approval, A$270k. Each stage costs a small fraction of what it governs.

Options at Gate 2

FEL 2 compared three options at the Class 4 level, each with a 10-year life and a 9% discount rate. At 9% over 10 years, a benefit of A$1 a year is worth A$6.418 today, the annuity factor, so each option’s net present value (NPV) is its annual benefit multiplied by 6.418, less its capital cost.

• (a) Four rooms, new plant, grader. Capital A$5,960k, net benefit A$1,400k a year. Simple payback 5,960 ÷ 1,400 = 4.26 years. NPV 1,400 × 6.418 − 5,960 = A$3,025k.
• (b) Two rooms, new plant, grader, and a five-year storage contract for the rest. Capital A$4,980k. The cold store benefit halves from A$640k to A$320k, so the net benefit is A$1,080k a year. Payback 4.61 years, NPV A$1,951k.
• (c) New plant and grader only; keep hiring storage. Capital A$3,400k, net benefit A$760k a year from the grader and energy. Payback 4.47 years, NPV A$1,477k.

Figure 3The three options at Gate 2

	Capital	Net benefit a year	Payback	NPV, 10 years at 9%
(a) Four rooms, plant, grader	A$5,960k	A$1,400k	4.26 years	A$3,025k
(b) Two rooms, plant, grader, storage contract	A$4,980k	A$1,080k	4.61 years	A$1,951k
(c) Plant and grader only	A$3,400k	A$760k	4.47 years	A$1,477k

The comparison that decides between (a) and (b) is the increment. Option (a) costs A$980k more than (b) and earns A$320k a year more, so the extra two rooms pay for themselves in about 3.1 years and add about A$1,074k of NPV. Against (c), the extra A$2,560k earns A$640k a year more, a 4.0-year payback and about A$1,548k of extra NPV. Ranking by payback alone would have placed (c) ahead of (b), which is one reason the business ranked by NPV and looked at the increments.

The choice was also tested against what FEL 3 might find. If (a)‘s cost rose to the full A$6,948k that was eventually funded, its NPV would still be about A$2,037k, above (b)‘s at its Class 4 capital, although (b) and (c) would also have grown with the same discoveries. An option that wins only if its estimate holds exactly is not much of a winner; (a) had room to spare. Options derived from the outcome, and why the choice changed in Outcome-First Projects by the Numbers covers how a good option list is generated from the outcome, which matters more than how it is ranked.

What a gate asks for

A gate needs two things: a package of evidence, and a decision made on it. This book covers the first. Northbank wrote the evidence each gate would receive before the stage began, so that the stage had a target and the gate had something definite to examine.

• Gate 2 received the options comparison with the basis for each option’s capital and benefit; the electrical capacity study and the condition assessment of the old plant; the grader trial results; the Class 4 estimate with its range; a definition score (Chapter 3); and the FEL 3 work plan and its A$152k budget.
• Gate 3 received a definition score at or below the business’s threshold of 250; the Class 3 estimate with its range and contingency basis (Chapter 4); design developed enough to price; the geotechnical report; the planning permit application lodged; the distributor’s connection offer; the fire engineering basis; tender prices for the refrigeration and grader packages; a schedule showing long-lead dates and float to the 31 January 2028 deadline; a risk register with owners; and the commissioning and pre-season trial plan.

How a gate should decide is a separate discipline. Stage gates that make real decisions: go, hold, recycle or stop explains the four decisions a gate can make and how to choose gatekeepers, and What the right to stop is worth in Stage Gates and Stopping Rules by the Numbers puts a value on staged approval. The point for the front end is narrower: a gate can only decide well if the stage before it has produced what the gate needs, and a stage can only produce that if the list was written first.

Figure 4Gate 3 evidence package

• Do: Definition score at or below 250, independently checkedChapter 3
• Do: Class 3 estimate with range and contingency basisquantity surveyor's report
• Do: Geotechnical report covering every loaded areaincluding the plant room pad
• Do: Planning permit lodged; connection offer from the distributordates for each
• Do: Fixed prices for refrigeration and grader packagesvalid to the approval date
• Do: Schedule with long-lead dates and float to week 65line-change window shown
• Ask: Stormwater detention requirementnot assessed at Gate 3; carried as a risk

The open item is real, and it costs the business later in the book.

The error: an options stage with one option

This is an illustration. The first FEL 2 work plan, drafted in the week the managing director agreed to fund the front end, contained an options study on paper only. The A$18k for the refrigeration consultant was scoped as “size the plant for four new rooms and the existing six”, and the A$25k of concept design was to develop the four-room layout. Nobody had decided to drop options (b) and (c); they had simply not been written down, because everyone already knew what they wanted.

The gap was caught in week 3 (Monday 23 November 2026), when the project engineer made the first pass of the definition checklist that Chapter 3 describes. Element A3, options considered and selection basis, asks for a documented comparison of the realistic alternatives and the reasons for the choice. Rated honestly, the work plan would leave A3 at level 4 at Gate 2, major gaps, because it would compare nothing. The project engineer raised it with the managing director the same day.

The correction cost nothing extra. The refrigeration consultant’s brief was rewritten to compare plant sizing for all three options, the concept designer drew a two-room layout as well as a four-room one, and the finance manager priced a five-year storage contract for option (b). The study confirmed option (a). It also documented why, which mattered three months later when a director asked, as directors do, whether two rooms and a storage contract would have been cheaper. The answer was in the Gate 2 paper: cheaper to build by A$980k, worse by about A$1,074k over ten years. Element A3 reached level 2 at Gate 2.

An options stage that compares nothing is not harmless. It spends the FEL 2 budget refining the first idea, leaves the gate with nothing to decide, and gives the business no defence later when the choice is questioned or when an estimate rises and someone asks whether a smaller option would now be better.

Exercise: stopping at Gate 2

Suppose FEL 2 had shown that the upgrade did not pay and the business stopped at Gate 2 in week 10. How much would it have spent? Compare that with the rush path at the same week, which approved in week 2 and spends about A$104k a week in delivery.

Worked answer. On the front-end path the business would have spent the A$22k concept and the A$96k of FEL 2, A$118k, and would owe nothing more, because nothing beyond the front end had been committed. On the rush path, week 10 is 8 weeks into a design-and-construct contract, so the business would have spent about 8 × A$104k = A$832k, plus any deposits paid on long-lead equipment, and would still owe the contractor whatever the contract provides on termination. Stopping on the front-end path costs about a seventh of stopping on the rush path at the same date, and the front-end path is also far more likely to have found the reason to stop.