I want to be fair about this, because the lazy version of this argument is wrong and everyone in the industry knows it. Agency engineers are not slower than in-house engineers. Many are considerably better, because they've seen forty codebases instead of one.
And yet the same work that takes a small team three weeks routinely takes an agency six months. That gap isn't effort. It's structure.
Where the time actually goes
The discovery phase that discovers what you already told them
Four to six weeks of workshops, personas and journey maps, ending in a deck that restates your brief in nicer typography. It exists for two reasons: it's billable, and it protects the agency from being blamed for a wrong direction later.
Some discovery is genuinely necessary — you cannot build what nobody has defined. But the ratio is off. If discovery costs more than the first shippable increment, you're funding an insurance policy, not a product.
The staffing gap
You sign in March. The team is free in May. Meanwhile you're assigned a partial team, or a strong lead who is genuinely on your project — for one day a week.
This is the least discussed and most expensive line item, because it never appears as a line item.
The pitch team and the delivery team are different people
You bought the person in the room. You got whoever was available. That isn't fraud; agencies are staffing businesses and the senior people sell. But the judgement you evaluated is not the judgement building your product.
Handoffs
Strategy writes a brief for design. Design produces files for engineering. Engineering asks the questions design didn't consider, and design is now on another account. Every handoff loses context and adds a queue, and queues are where calendar time hides.
Approval theatre
Three rounds of feedback per deliverable, each with a week of lead time, most of it spent waiting on someone's calendar rather than on anyone's work.
Add it up
Six weeks discovery. Four weeks staffing gap. Six weeks design with two review cycles. Ten weeks build, interrupted by three handoffs. Four weeks of QA and stakeholder review. Twenty-six weeks — and the actual engineering was maybe fifteen days of focused senior work.
Nobody was lazy. Nobody padded a timesheet. The structure produced the number.
What the structure is actually buying you
In fairness: at a certain scale it buys real things. Continuity when someone leaves. Capacity to run five workstreams at once. Someone to sue. A brand your board recognises. For a 200-person company replacing a core system, that's worth paying for.
For a founder trying to find out whether an idea works, it's paying for insurance against risks you don't have.
How to buy around it
- Ask who writes the code, by name, before you sign. Then ask what percentage of their week you get. The answer to the second question is the honest one.
- Cap discovery explicitly. "One week, then we build something." A good partner will push back with specifics if that's genuinely too short — and that pushback tells you something useful.
- Buy short fixed windows. Two or three weeks with a shippable outcome at the end. Structure can't hide inside a two-week window; there's nowhere for it to go.
- Insist on working software early, not artefacts. A deck is not progress. A build on your phone is.
- Reduce approvers to one. Every extra person with veto power adds a week, whatever the contract says.
The honest counter-argument
A small senior team is faster and cheaper, and it has a real weakness: bus factor. If it's one or two people and one gets ill, you stop. An agency absorbs that; an individual doesn't.
That's a genuine trade, and the right answer depends on whether you're validating something or running something. Just make it a decision rather than a default — and while you're at it, decide which contract shape matches how well you actually understand the work.