In-House Dev Team vs. White-Label Partner: A Real Cost Comparison for Agencies

The comparison agencies usually make is a salary against an hourly rate. The hourly rate looks alarming, the salary looks manageable, and the decision gets made on two numbers that are not comparable.

This is a framework for comparing them properly. It deliberately contains no benchmark figures — salaries and rates vary enormously by market, and a number from someone else's market would make your model worse, not better. Fill it in with your own.

Why the obvious comparison misleads

An annual salary divided by working hours is not what an employee costs per hour, for three reasons:

1. Salary is a fraction of employment cost. Employer taxes, benefits, equipment, software licences, and recruitment fees are real and recurring.
2. Not all paid hours are billable. Leave, holidays, sick days, training, internal meetings, and ramp-up all consume paid time.
3. You pay for capacity, not consumption. An employee costs the same in a slow month. A partner does not.

Point three is usually the one that decides it.

The full cost of an in-house hire

Work through these categories:

Direct compensation — base salary, plus bonus or commission if applicable.

Employment overhead — employer taxes and statutory contributions, healthcare or benefits, pension or retirement contributions, insurance. This is a substantial multiplier on base salary in most markets.

One-time costs, amortised — recruitment (agency fee or your own time), equipment, onboarding, and the productivity ramp before a new hire is at full output. Spread these over expected tenure, and be realistic about tenure.

Recurring non-salary costs — software licences and tooling per seat, workspace if applicable, training and professional development, management overhead. That last one is real: someone senior spends time on one-to-ones, reviews, and unblocking, and that time has a cost.

Then compute effective billable hours. Start from total working hours in a year, subtract leave, public holidays, expected sick days, training, and internal or administrative time. What remains is the denominator.

Total annual cost divided by effective billable hours gives a true internal hourly cost. It is materially higher than salary ÷ 2,080, and that gap is where the "partners are expensive" intuition comes from.

The cost of a white-label partner

Simpler, but not free of overhead:

The rate itself — hourly, or fixed per project.

Your management time — briefing, reviewing, coordinating. Lower than managing an employee, but not zero. Cost it at your own hourly rate.

Scoping and communication overhead — meaningful early in a relationship, and it declines as the partner learns your standards and stack.

Change and rework risk — with poor specs, some rework gets billed. Good scoping reduces this; nothing eliminates it.

The critical property: this cost scales with usage. No projects, no cost.

The variable that dominates: utilisation

Everything turns on how much work you can reliably feed an engineer.

An in-house hire is cheaper per hour only at high, sustained utilisation. The internal hourly cost calculated above assumes you fill those hours with billable client work. At lower utilisation, the same annual cost spreads over fewer productive hours and the effective rate rises sharply.

There is a break-even point — the monthly volume of work above which in-house is cheaper. Calculate it for your own numbers:

• Below it, a partner is cheaper and carries less risk.
• Above it, in-house is cheaper if the volume is genuinely sustained.

Then ask the harder question honestly: can you forecast that volume twelve months out? Most agencies overestimate here, because they extrapolate from a good quarter. A hire is a twelve-month-plus commitment made against a pipeline that is rarely visible that far ahead.

Costs that do not appear in either column

Risk and flexibility. A partner engagement can usually end with weeks of notice. An employee cannot — and redundancy carries financial cost, legal process, and a real effect on the team that remains.

Opportunity cost of saying no. If a lack of capacity means declining work, the lost margin belongs in the comparison. This is frequently larger than the rate difference and almost always omitted.

Time to productive output. Recruiting and onboarding a good engineer takes months. A partner starts against a scoped brief in a fraction of that. If a client deadline is real, that gap has a price.

Capability breadth. One hire is one skill set. A partner can typically supply backend one month and mobile the next. Matching that in-house means several hires.

Knowledge retention. This favours in-house genuinely. Institutional knowledge stays; with a partner, some walks out at the end of the engagement. Good documentation and code ownership narrow the gap but do not close it.

Key-person risk. A single in-house engineer who leaves takes context with them and leaves you with nothing until you rehire. A partner absorbs their own turnover.

The hybrid most agencies land on

In practice, the durable arrangement is usually:

• In-house for the work that is continuous, core to your differentiation, and requires deep institutional context.
• Partner for peaks, for skills outside your core, and for testing service lines before committing salary to them.

This is not a compromise. It matches fixed cost to predictable demand and variable cost to variable demand, which is the correct structure for a business with lumpy pipeline.

How to actually decide

1. Calculate your true internal hourly cost using every category above — not salary ÷ hours.
2. Get a real rate from a partner for a real scope, not a rate card.
3. Find the break-even monthly volume.
4. Look at your actual last twelve months of demand, not your best quarter.
5. Ask whether you can commit to that volume for the next twelve months.
6. If yes, and the skill is core, hire. If the volume is uncertain or the skill is adjacent, use a partner — and revisit once demand proves itself.

The mistake is not choosing wrongly. It is choosing on a comparison of two numbers that were never measuring the same thing.

If a partner looks like the right call, the questions to ask before signing one are worth reading first.