There is a familiar piece of advice given to young practices: take the small jobs, they lead to bigger ones. It is true often enough to be worth repeating and vague enough to be dangerous, because it omits the part where small projects consume disproportionate time, generate disproportionate risk, and are the most common reason practices of our size have a bad year.

Both things are true. Small work is where a practice becomes good at anything, and small work is where practices quietly lose money. Being clear-eyed about the second is what makes the first sustainable.

Fee does not scale with effort

A £200,000 project does not take a tenth of the time of a £2 million one. It takes perhaps a third.

The reason is that most of the work is fixed regardless of size. There is a client to be understood, a site to be surveyed, a planning application with the same documents and the same statutory period, a building control submission, a tender process, a contract, and a construction phase requiring visits at roughly the same frequency. Only the drawing package scales, and even that scales badly, because a small building has the same number of junction types as a large one — it just has fewer of each.

Percentage fees make this arithmetic worse. A 7% fee on a small project is a smaller absolute number covering nearly the same tasks, and practices absorb the difference through unbilled time. The honest answer is fixed fees on small work, priced from a realistic assessment of hours rather than a percentage, which usually produces a figure that looks alarming as a proportion and is merely adequate as a sum.

We now decline small projects that cannot support a fixed fee at that level. This costs us work and it has been the single most useful commercial decision we have made.

The genuine returns

That said, the case for small work is strong, and it is not primarily about future large commissions.

Details get tested. On a large project a detail is drawn, built once at scale, and its performance is known in three years. On a small project the whole thing is built in eight months and you see the result almost immediately. Nearly every detail we now use at scale was first built somewhere small.

People develop faster. An architect two years qualified can run a small project end to end — client meetings, planning, tender, site — which on a large project would take a decade to accumulate. The learning rate is not comparable. Three of our current project leads were formed this way, and they are better than they would have been shadowing on something bigger.

The client relationship is direct. No project manager, no cost consultant filtering the conversation, no development director who joined last month. You talk to the person paying, and they tell you what they actually think. This is where practices learn to have difficult conversations, which is a skill that does not develop on projects mediated by three consultants.

Risk is contained. A mistake on a small project is survivable. The same mistake on a large one may not be. There is a real argument for testing anything unfamiliar — a new material, a new contract form, a new consultant — at a scale where being wrong is affordable.

Where the money actually goes

The losses on small projects are rarely in the design stages. They are in three places, and all three are manageable.

Scope creep on domestic work. Private clients spending their own money change their minds, reasonably, and each change is small enough to feel unfair to bill for. The cumulative effect is enormous. The remedy is a written change procedure from day one, applied to changes that cost twenty minutes as well as those that cost twenty hours, so that the precedent exists before it matters.

Site. Small contractors need more support, not less, and small projects are frequently built by firms without a contracts manager or a technical department. The queries arrive by phone, unrecorded, at inconvenient times. Budget more site time than the contract value suggests, and insist on written instructions even when it feels bureaucratic.

The tail. Small projects have long, thin endings: snagging that drifts, a defects period nobody has resourced, a client who calls two years later about a leak. This time is almost never in the fee. Put a defined end in the appointment and a rate for anything beyond it.

The workshop

The most useful small project we have done was sixteen craft workshops in a row of dock sheds. The construction value was modest, the fee was fixed, and the work was mostly repair rather than design.

It taught us more about reuse economics than anything at scale — specifically that the money saved by retaining a structure is best spent outside the building rather than on upgrading it beyond what tenants need. The embodied carbon came in at 96 kgCO₂e/m², which remains our lowest figure and which we have quoted in three subsequent bids for much larger projects.

It also produced no direct follow-on commission, which is the part the standard advice would not predict. The value was entirely in what the practice learned and in a set of numbers we can now put in front of a developer.

How we decide now

Three questions, in order.

Can it carry a fixed fee that reflects the real hours? If not, it will be subsidised by other projects, and someone should be able to say why that subsidy is worth it.

Is there something to learn — a material, a client type, a contract, a technique — that we cannot learn as cheaply elsewhere? Small work justified purely by hope of future commissions is a bad bet; small work justified by what it teaches usually pays for itself.

Is there a named person who will run it end to end, and is the project the right size for where they are? Small projects distributed across a team as spare-capacity work are the ones that go wrong. Given wholly to one person, they are the best training available.

If all three answers are yes, we take it. If two are yes, we discuss it. If it is only the third, we say no, and we say no more often than we used to.