Estimating and keeping to budget

In The Heir’s Challenge four cousins organise an exhibition without ever estimating what it will cost. The gallery owner’s contract arrives in week nine, when it is too late to change course. It is an invented story, but anyone who works in an agency has seen it at least once: a project sold well, worked on with passion, and closed at a loss without anyone noticing before the end.

Cost estimation is not an accounting exercise. It is the tool you use to decide whether a project can be done, how, and with what safety margins.

The right words

Half the misunderstandings about numbers come from words used differently.

• Budget: the amount the client makes available.
• Estimate: the forecast of what the work will cost.
• Sold: the amount actually agreed in the contract.
• Costs: internal (people’s time) plus external (suppliers, licences, production).
• Contingency: a buffer, in money or days, for the unexpected. It should be declared, not hidden in the other items.
• Margin: (sold − costs) / sold.
• Internal hourly cost: what an hour of someone’s time really costs, including their share of overheads. It is calculated from the annual company cost divided by the hours that are actually productive, not by calendar hours.

Example	Amount
Sold	50,000
Internal costs	28,000
External costs	9,000
Total costs	37,000
Margin	13,000 (26%)

Two kinds of estimate

The order-of-magnitude estimate is for the beginning, to understand whether a project is feasible: it is given as a range (“between 40 and 60 thousand”) and helps client and supplier align expectations before investing time.

The detailed estimate breaks the project into phases and deliverables, each with its own cost items. Until the Statement of Work is final, there will be several versions of it, and that is normal.

Five techniques

1. Top-down. Start from the available budget and divide it across the phases, asking whether each part is enough. It relies on experience and is very useful at the start, to check with the client that expectations are realistic and to give the team a clear mandate.
2. Analogous. Look at what a similar project cost. Quick, but not very precise: the risk is comparing things that only look alike on the surface.
3. Parametric. Like the analogous one, but based on measured historical data: what a page, a screen, a minute of video costs on average.
4. Three-point. For each task, estimate the best case (O), the most likely (M) and the worst (P). The most common formula weights the likely case more: (O + 4M + P) / 6. It forces you to think about uncertainty instead of hiding it.
5. Bottom-up. List every single task, estimate them one by one and add them up. It is the most precise and the slowest. It is the one to base the Statement of Work on.

In practice they are combined: top-down to see whether it makes sense, bottom-up to know what it really costs, three-point where uncertainty is high.

The items everyone forgets

Production tasks are almost always estimated well. What is usually missing is everything else:

• project management and coordination time;
• meetings, starting with the kick-off;
• the client’s review rounds, and the work they generate;
• testing, fixes and release;
• client training and documentation;
• the warranty period after delivery;
• contingency, declared.

Every estimate then depends on the type of project, its duration, the seniority of the team, the technical resources and the production costs. And it sits inside the triple constraint: scope, time and cost are linked. Change one and the others change too.

And in agile projects?

Estimation is needed there too, but it changes shape. You work in fixed-price packages, each re-estimated in the light of what you have learned; in ranges (“12 to 16 weeks for this scope”); with extra budget slots for features that emerge along the way, and the option of stopping early if the essentials have already been delivered.

Keeping to budget

An estimate is valid on the day it is signed. From then on what matters is the comparison, every week, between what you expected and what you are actually spending, in hours and in external costs.

The key question is not “how much have we spent?” but “how much will we have spent at the end, if we carry on like this?”. That is the estimate at completion: the cost already incurred plus the estimate of the work that remains. It needs redoing regularly, reviewing the remaining tasks with the people who will do them.

Earned Value, in brief

It is the most widely used method to see whether you are on track, and it relies on three numbers:

• PV (Planned Value): the value of the work you should have completed by today;
• EV (Earned Value): the value of the work actually completed;
• AC (Actual Cost): what you spent to do it.

From these come two indices: CPI = EV / AC (below 1 you are spending more than planned) and SPI = EV / PV (below 1 you are behind schedule).

An example. Budget 40,000. Halfway through, you should have completed 50% (PV 20,000); you have completed 40% (EV 16,000) and spent 20,000 (AC). CPI and SPI are both 0.8: you are late and spending more than planned. If nothing changes, the project will cost about 40,000 / 0.8 = 50,000. You know it halfway, not at the end: that is the whole point.

Say it early

An overrun reported early, with the numbers and a proposal, is a problem to solve together. The same overrun discovered at the end is a crisis. Every change of scope should come with its estimate, before it is done: that is how the client chooses knowingly, and how the margin doesn’t drain away in silence.