What an Idle Week Actually Costs: the Bench, in Pounds

Every agency owner can quote what their people cost when they're working — it's in every estimate. Almost none can quote what their people cost when they're not working, even though it is the same number. The salary doesn't pause between projects. The £75-an-hour developer is £75 an hour on the Tuesday they ship a release and £75 an hour on the Tuesday they refresh the internal wiki waiting for the next statement of work to sign.

I ran delivery for years before I could put a pound figure on our bench, and the reason was not negligence — it was that no system I ran had anywhere to put the number. Billable hours had project codes. Bench hours had a shrug. This piece is about what the shrug costs, where it hides, and the one change that makes it manageable.

The arithmetic nobody runs

Take the standard team from the utilisation discussion: six people, blended fully-loaded cost of £75 an hour — roughly £216,000 a quarter whether they deliver anything or not, because fully-loaded cost is a fact about employment, not about activity.

Now give them a gap. Project A ships at the end of March; Project B's contract, "days away" for three weeks, signs mid-April. Two idle weeks for six people:

6 people × ~37.5 hours × 2 weeks × £75 ≈ £33,750 — call it £34k if the whole team is dark, £17k if half of them caught overflow work elsewhere. Per idle fortnight. At an agency running 30–40 people, three or four such gaps a year across different teams is entirely ordinary — and entirely uncounted. The annual bench bill at a mid-sized agency, when we finally measured ours properly, was comfortably into six figures: the cost of a decent salesperson, spent annually on paying skilled people to wait.

None of this is an argument against ever having a bench. It is an argument against not knowing its price.

Where the cost hides

The bench is invisible for a structural reason: cost reporting at agencies is organised by project, and the bench's defining feature is that it has no project. So the hours either go unlogged entirely, or land in a catch-all "internal" bucket, and the payroll cost behind them drains into the general overhead figure.

That routing does two kinds of damage. First, the bench cost disappears as information — nobody reviews a line that doesn't exist, so a growing pipeline problem can hide inside "overhead crept up a bit" for years. Second, and more corrosively, when it dissolves into overhead it gets allocated onto live projects — so every healthy, well-priced project in the building reports a slightly worse margin because a different team sat idle in April. The projects look worse; the pipeline looks blameless; and leadership goes hunting for a delivery problem when what they have is a sales problem.

The diagnostic tell is exactly the inverse of the busy-but-thin pattern from the seven signs: margins on individual projects look mysteriously mediocre while every project team seems disciplined. When project-level numbers and project-level behaviour disagree, the explanation is usually a cost that belongs to no project — and the bench is the biggest one there is.

The panic response costs more than the bench

Once an owner feels the bench (usually via the bank balance rather than a report), the instinct is to fill it with anything billable: a discounted project, a mate's-rates favour, a thin retainer taken "to keep the team warm". Sometimes that is right. Often it costs more than the idleness it cured, through two mechanisms the panic never prices.

Occupancy. The discounted project consumes the team for eight or ten weeks. When properly priced work lands in week three, the team is busy — on the cheap work. The real cost of the discount is not the thin margin on the filler; it is the healthy margin on the work you couldn't take. An empty bench can accept the good project tomorrow; a badly filled one cannot.

The rate memory. Discounts given under bench pressure have a way of outliving the bench: the client renews, the realised rate on the relationship is now anchored low, and a two-week capacity problem has become a two-year pricing problem.

The honest comparison, run with real numbers, frequently lands the other way from instinct: a fortnight of priced-in bench (£17–34k, once, visible) against a quarter of discounted occupancy (thin margin, opportunity cost, rate damage — recurring). Sometimes the filler work still wins. But it should win a calculation, not a panic.

What to actually do

Three practices turn the bench from a shrug into a managed number.

Name it and cost it. Bench hours get logged — to a real internal code, not scattered into project slack — and costed at each person's real rate like any other hour. The point is not surveillance; it is that an idle week should produce a visible £17k line that says pipeline gap, week 14, instead of a 2% mystery degradation smeared across every client project. What gets a line gets a conversation.

Read it as a sales signal. A structural bench is almost never a delivery problem. It is the pipeline arriving late, or lumpy, or at the wrong shape for the team you keep. Priced in pounds per quarter, the bench becomes the business case for the pipeline fix — another salesperson, a partnerships channel, smaller-but-steadier engagements — with an ROI you can actually compute, because you finally know what the idleness costs.

Spend it deliberately when you keep it. Some bench is worth carrying: it is what lets you say yes quickly, and it is often cheaper than the contractor premium you'd otherwise pay for every spike. But kept bench should be spent like the money it is — on the internal tooling, the case study, the training that compounds — not merely endured. An idle week that produces the capability you sell next quarter was an investment. An idle week that produces nothing was just payroll on fire, slowly.

Seeing it continuously

All three practices share a prerequisite: knowing, this week, who is on the bench and what it is costing — not discovering it at the quarter's post-mortem, when the number arrives too late to act on.

That visibility falls out of the same data that runs everything else in this series. Saldo costs every logged hour at the person's real rate continuously — which means the hours not landing on client projects are just as visible as the ones that are, priced in the same pounds, while the gap is still fillable. The 15-minute demo runs on your own Jira; if there is a bench hiding in your overhead right now, it will have a number on it before the call ends.

The shorter version: the bench is not free time, it is unpriced time — £17,000 a fortnight for a six-person team, dissolved into overhead where it blames your best projects for a pipeline that ran dry. Put a code on it, put a price on it, and it stops being a shrug and starts being the clearest sales signal the agency produces.