Agencies usually compare these two options badly, by putting a developer's salary next to a partner's project quote. Those are not comparable numbers, and the comparison almost always flatters the hire.
The numbers below are illustrative. They are not our rates and they are not drawn from any specific agency's accounts — they are a worked example using round figures so you can substitute your own. The point is the structure of the calculation, which is where the mistake usually lives.
What an in-house developer actually costs
Start with base salary. The US Bureau of Labor Statistics publishes current pay data for web developers in its Occupational Outlook Handbook, and the band is wide once you account for market and specialism — check it against your own city rather than taking a national figure. For this example, take a round $90,000 and substitute your own number.
Then add what employment costs on top of salary. Payroll taxes, benefits, insurance, equipment, software licences, workspace and training typically bring the loaded cost to somewhere around 1.25 to 1.4 times base. Using 1.3:
| Cost line | Illustrative amount | Notes |
|---|---|---|
| Base salary | $90,000 | Substitute your own market rate |
| Loaded cost at 1.3x | ~$117,000 | Payroll taxes, benefits, insurance, equipment, workspace |
| Recruitment | $13,500 – $18,000 | If using an agency at 15–20% of first-year salary |
| Onboarding | 1 – 3 months | Partial output during ramp-up |
Then divide by productive hours, not paid hours
This is the step that gets skipped, and it changes the answer more than anything else.
A full-time year is roughly 2,080 hours. Subtract holiday, public holidays and sick leave and you are near 1,850. Then subtract the parts of the week that are not billable delivery: internal meetings, admin, code review, learning, helping account managers scope work, and the gaps between projects. In agency environments, 60 to 70 percent utilisation is a realistic band for a developer who is also expected to participate in the business.
At 65 percent, that is about 1,200 productive hours. Against a loaded cost of $117,000, the effective cost is roughly $97 per productive hour — before recruitment and before onboarding.
Most agencies that believe they are paying $43 an hour for a developer, on the naive salary-divided-by-2,080 basis, are actually paying somewhere north of double that.
The cost that does not appear on any invoice
The in-house cost is fixed. It is identical in a month with four projects and a month with none. For an agency whose technical work is genuinely uneven — which is most of them — that is the real financial risk, and it is invisible in an hourly comparison.
There is a second, subtler version of the same problem: a single developer is a single point of failure. They take holiday. They get ill. They eventually leave, usually with more knowledge in their head than in your documentation. An agency whose delivery capability is one person is carrying a risk that does not show up until the day it does.
And one hire covers one specialism. The work that actually arrives spans front end, back end, integrations, ecommerce platforms and increasingly AI features. A single developer will be genuinely good at some of that and out of their depth in the rest.
What a partner costs, structurally
A partner's cost is variable. You pay per project, you know the figure before you quote your client, and in a month with no technical work the cost is zero. There is no recruitment, no onboarding period, no bench time, no payroll underneath it.
The honest trade-off is that per-hour, a partner is usually more expensive than a fully utilised in-house developer. If you can genuinely keep someone at 80 percent utilisation on delivery work all year, in-house wins on unit cost and you should hire. The question is whether that utilisation is real or aspirational.
Where the crossover sits
Roughly, and depending heavily on your market:
| Your situation | Usually cheaper | Why |
|---|---|---|
| 25–30+ hrs/week of development you can hire for | In-house | Utilisation is high enough to beat a partner on unit cost |
| Under ~20 hrs/week, or several specialisms | A partner | You stop paying for idle capacity, and get breadth one hire cannot cover |
| Uneven month to month | A partner | Cost follows the work instead of the calendar |
| Growing but unpredictable | A partner, then hire | Hire once volume is provably stable, not seasonally busy |
The mistake to avoid is hiring off the back of one exceptional quarter. Salary is a twelve-month commitment made on the strength of three good months, and the quiet quarter arrives eventually.
The hybrid most agencies end up with
In practice, the agencies that handle this well are rarely purely one or the other. They hire for what is constant and predictable — often a single generalist who handles small changes, client requests and coordination — and use a partner for project peaks, specialisms outside the hire's range, and anything with a deadline that cannot move.
That arrangement also removes the single-point-of-failure problem, because the internal person is no longer the only route to delivery.
One category worth pricing separately is ongoing support. Maintenance is predictable, recurring and easy to resell, and it is the part of an in-house hire's week that most often gets squeezed out by project work — our own maintenance plans are published with their prices and included hours so you can compare that piece directly against an internal cost.
If you want to see how the partner side works before running your own numbers, our agency model covers the engagement shapes, and white label development for agencies sets out what delivery under your own brand actually involves.
