You just won work you cannot staff.
Maybe a client asked for engineering depth you do not have on payroll. Maybe three projects landed in the same month. Maybe someone quit and the delivery calendar did not care. The next decision feels binary: hire, or find a partner who can ship under your brand.
It is not binary. It is a capital allocation problem with a calendar attached.
What "build the team" actually costs
The number people put in the spreadsheet is salary. The number that hits the business is first-year total cost.
BLS puts median software developer pay at about $133,080 (May 2024). Benefits and employer payroll taxes typically add roughly 30% of total employer cost, so a $130K base is closer to $175K out the door before that person is useful. Staffing guides for 2026 put mid-level technical seats in the $150K to $215K first-year band once recruiting, benefits, tools, and management time land. Senior seats climb past $200K all-in.
So $125K is not a scare figure. It is a floor for a real technical specialist in the US.
Then there is time. Engineering hire cycles often sit in the 40 to 55 day range. Senior offers average around 47 days. Notice periods and counter-offers can add more. Ramp to full productivity for software roles commonly runs 6 to 12 months. Stack hire lag plus start lag plus ramp and you land in a familiar band: about 8 to 15 months before that seat is net-positive against a peer who already knows your stack and your clients.
ManpowerGroup's 2025 survey found 76% of US employers still struggle to find needed skills. Gem's recruiting data shows average time-to-hire rose about 24% from 2021 to 2024. Empty seats are not free. Overflow lands on the people who stayed, and the first few months of a new hire often buy orientation, not full billable capacity.
A worked example you can plug your own numbers into
Take a mid-level specialist at $140K fully loaded, plus $20K to recruit and onboard.
For nine months they are partially productive. Call it 50% effective while they learn clients, tools, and your delivery process. You still paid for nine months at full load.
Rough math:
- Recruiting and setup: $20,000
- Nine months of loaded pay: about $105,000
- Useful output during ramp: roughly half of that pay period
- Net cash out before the seat feels "real": well past $125K, with most of a year gone
Now price the failure mode finance rarely models. Replacing a bad technical hire often costs 50% to 150% of annual salary once you re-search, absorb lost output, and unwind weak work. Digital marketing specialist tenure in agency contexts is sometimes cited around 2.3 years, which means you may re-buy the same capability on a short cycle.
If demand is steady and utilization stays high, ownership can still win. If demand is spiky, you just bought a fixed cost for a variable problem.
The utilization trap
SPI Research reported professional services billable utilization at 66.4% in 2025, a record low after 68.9% in 2024. Many firms treat 70%+ as a healthy floor. Agency production targets often sit higher still.
At soft utilization, permanent headcount looks smart on the org chart and heavy on the P&L. You are paying for idle capacity. Partner capacity turns fixed payroll into variable cost of goods sold. That matters more in soft years than in boom years, and 2025-2026 has been soft for a lot of shops.
This is a timing argument as much as a philosophy. When every seat is full and booked out, hire. When work arrives in waves, partner for the wave.
What partnering is (and is not)
White-label capacity is not free labor. A mid-size agency pattern reported in industry writeups: white-labeling SEO instead of hiring a three-person team cut fixed monthly cost by about $22,000 while keeping similar margins on client work. The win was cash-flow flexibility and less idle payroll in troughs.
Partnering is also not "always cheaper quality." Low rate cards look decisive until rework, timezone lag, and weak specs show up on the invoice. Cheap partners destroy brands. Good partners buy time and depth when your PM and QA stay strong at the edge.
There is an opposite failure mode too. Partner for everything and you become a reseller with a logo. Clients hire agencies for judgment, relationships, and the work only you do well. If that layer leaves the building, so does the reason they picked you.
The real question: what must you own?
Treat each capability as a portfolio line, not a moral stance.
For development, SEO, video, security, analytics, or whatever just walked in the door, ask:
Is this core to how clients choose us? If yes, build or keep tight control. If it is adjacent or commodity execution, partner is fine.
Do we have steady demand for at least one full-time person year-round at 75%+ utilization? If no, you are hiring an integer person for a fractional problem. Partners and fractional specialists solve that cleanly.
Can we recruit and retain this skill in 90 days? If the market says no, the calendar already voted.
Does failure damage brand, compliance, security, or IP? Own the risk surface. Partner only with strong SLAs, clear ownership, and QA you control.
Does the client need this in weeks or can they wait 8 to 15 months? Client timelines do not care about your hiring pipeline.
Do platform volume and tool economics favor a specialist who lives in that stack full-time? In-house is not set-and-forget. Platforms change. Skills half-lives get shorter. A partner who does that work every day often absorbs re-training cost that would otherwise hit your payroll.
Simple rule for agencies and MSPs under about $20M: hire for relationship, strategy, and differentiation. Partner for specialized execution and overflow work. Build deep in-house only when utilization and retention math are proven, not hoped for.
The year-two spreadsheet that never arrives
A common model: hire now, absorb partner fees for a while, break even in year two or three. That model assumes stable team, stable platforms, and stable demand. Specialists get recruited. Platforms change. Client mix shifts. The hole left by a departed specialist can last another three to six months while clients still expect delivery.
Building capacity is a multi-quarter capital project. A partner arrangement is a capacity product. Smart operators mix both on purpose.
We run white-label engineering for agencies under NDA. Your brand, our work. We stay invisible. First conversation is free. No commitment. If you are staring at a hire requisition and a delivery date on the same whiteboard, talk to us.