Spreadsheet comparisons are comforting. You line up a contractor's monthly rate against salary, taxes, and benefits, pick the smaller number, and call it strategy. A year later the "cheap" contractor needs daily supervision, or the full-time hire sits underutilized between busy seasons, and the spreadsheet looks naive. Cost matters. It is rarely the deciding variable that predicts whether the arrangement still works.
A founder outsourced bookkeeping at half the cost of a part-time hire, then spent six hours a week translating context the bookkeeper never absorbed. After bringing a part-time operator in-house for the same work, the hourly rate rose and the founder's calendar finally freed up — the spreadsheet had ignored management overhead.
The better frame: how central is this work to what makes your business distinct, how constant is the demand, and how much management attention will the arrangement consume? This article uses a decision table, then four questions to pressure-test the choice before you sign an offer letter or a statement of work.
A practical decision map
| Work profile | Lean toward | Watch-outs |
|---|---|---|
| Core differentiator, constant demand | In-house hire | Hiring slowly while quality slips; under-scoping the role |
| Core differentiator, bursty demand | Senior contractor or fractional specialist, with a path in-house later | Knowledge walking out at the end of each project |
| Peripheral, constant demand | Outsource to a stable provider, or hire if volume is huge | Overpaying a full-time generalist for narrow work |
| Peripheral, bursty demand | Outsource / project contractor | Turning every burst into a vague retainer that never ends |
| Compliance-heavy or access-sensitive | In-house or tightly contracted with clear controls | Casual freelancers on production credentials |
"Core" means work that shapes what customers buy from you or how you deliver the promise — product craft, key client relationships, proprietary process. "Peripheral" means necessary but interchangeable: routine bookkeeping within a standard playbook, generic graphic resizing, commodity IT helpdesk. Mislabeling peripheral work as core is how companies over-hire. Mislabeling core work as peripheral is how companies hollow out their advantage.
Question 1: Does this work need to live in institutional memory?
Some knowledge compounds inside a company: how a tricky client likes proposals, why the warehouse layout is weird, which edge cases break the billing rules. If the job continuously creates that kind of memory, outsourcing resets the meter every time you change vendors. You can document aggressively, but documentation rarely captures the judgment calls that make the work good.
Other work is standardized enough that a competent outsider with a checklist performs fine. Tax filing for a simple entity, scheduled social graphics from a brand kit, overflow customer email using your templates — these can live outside without starving the business of learning. Be honest about which bucket you are in. Founders sometimes call everything "strategic" because it feels important that week.
Question 2: Is demand constant or bursty?
Constant weekly load usually favors employment once volume is real: idle salary hurts, but so does paying billable contractor hours for work that never stops. Bursty load — launches, seasonality, migrations, redesigns — usually favors outsourcing. Hiring a full-time person for two intense months and ten quiet ones is how you fund resentment and busywork.
Measure demand before you invent a role. Track hours spent on the function for four to six weeks. If the honest average is twelve hours a week with spikes to forty, that is not a clean full-time seat unless you combine responsibilities carefully. Combining responsibilities only works when the skills actually overlap; "marketing plus bookkeeping" is a fantasy org chart.
Question 3: What is the true management overhead?
Outsourcing does not remove management; it changes it. You still define outcomes, review quality, resolve misunderstandings, and chase timelines. A low hourly rate with high oversight can exceed a salaried hire once you price the founder's time honestly. Conversely, a strong specialist who needs a weekly check-in can be cheaper than a junior employee who needs daily coaching you do not have time to give.
Estimate oversight hours per week for each option and multiply by what that person's time is worth to the business. Include coordination tax: time zone lag, tool access provisioning, and the meetings required because context is not shared by default. If you cannot name who will manage the vendor or the hire, you are not ready for either — you are ready for a backlog of neglected work.
Question 4: Can you run a smaller test before the big commitment?
Few decisions here need to be permanent on day one. A two-month contractor pilot with a clear deliverable teaches you about quality and chemistry before an annual contract. A hire can start with a focused mandate and a probation period that you actually take seriously. Expanding scope after proof beats betting the year on a LinkedIn profile or a polished proposal deck.
Define success metrics for the test in writing: turnaround time, defect rate, customer impact, or hours saved internally. Vague pilots produce vague conclusions, which produce another pilot.
Cost still matters — just count all of it
When you do compare money, include recruiting time, onboarding ramp, benefits, equipment, software seats, contractor markups, payment processing on invoices, and the cost of replacing someone who leaves with all the context. Also include switching costs: bringing outsourced work in-house later means re-learning; firing a bad hire means unemployment rules and morale effects that never appear on the contractor quote.
Be wary of outsourcing sales pitches that promise "seamless extension of your team" without discussing who owns priorities when their other clients get loud. Be equally wary of hiring pitches — including your own optimism — that assume a great generalist will appear at your salary band in thirty days. Labor markets do not owe you a convenient narrative.
Hybrid arrangements and how they fail
Many small businesses end up with a hybrid: an in-house lead plus outsourced execution, or in-house execution plus an outsourced specialist for peaks. Hybrids work when ownership is explicit. They fail when everyone can touch the work and nobody is accountable for the outcome, or when the external team never receives enough context to do more than cosmetic tasks.
If you hybridize, write a RACI-style note even if you hate the acronym: who decides, who does, who reviews, who gets informed. Revisit it after the first month. Ambiguity is cheaper than conflict only until the first missed deadline.
Revisit as the business changes
A function that was correctly outsourced at ten customers may belong in-house at two hundred. Work that justified a hire may shrink after you productize a service. Put a calendar reminder to re-evaluate major make-vs-buy choices annually, especially after revenue jumps or strategy shifts. Treating the first decision as permanent is how org charts fossilize around yesterday's constraints.
When you reverse a decision, plan the knowledge transfer deliberately. Bring contractors in-house with paid overlap weeks. Move in-house work outside only after documenting the playbook and confirming the vendor can run it without daily Slack from you. The transition is part of the cost; pretending otherwise is how both sides get burned.
Choose the structure that keeps critical learning close, matches the shape of demand, and stays manageable for the humans who already have full jobs. If the spreadsheet disagrees with that picture, rebuild the spreadsheet.
Signals you outsourced the wrong thing
Watch for these patterns: every deliverable needs heavy rewrite by someone internal; the vendor cannot proceed without daily questions that only an insider can answer; customers notice inconsistency because outsiders lack context; or strategic decisions stall waiting on someone who does not share your incentives. Those are signs the work was core (or at least context-heavy) and belongs closer to home — or that you hired the wrong partner and should switch before blaming the model.
The opposite failure also happens: you hire full-time for work that is really a fluctuating project, then invent internal tasks to justify the salary. If you find yourself creating work to keep someone busy, revisit outsourcing for the overflow and narrow the role.
Writing the brief that makes either path work
Whether you hire or contract, poor briefs create poor outcomes. State the outcome, constraints, examples of good and bad, decision rights, and communication cadence. For contractors, add what "done" means and how revisions work. For hires, add how success looks at 30, 60, and 90 days. Ambition without definition produces activity without progress.
Include systems access and security expectations up front. Nothing slows a new arrangement like discovering in week two that nobody provisioned tools, or that a contractor expected admin rights you will never grant. Sort that before the start date.
Culture and communication fit
An excellent specialist who only communicates in weekly PDF dumps may fail inside a team that lives in chat. A friendly generalist who thrives on hallway conversation may fail in a remote-first company. Probe working style during the trial. Chemistry is not a soft metric when the work requires tight iteration; it is operational risk.
Finally, protect your team from a revolving door of cheap contractors on core paths. Constant handoffs destroy morale for the people who remain and teach customers that your quality is optional. Pay for continuity where continuity is the product.
A one-page decision record
When you choose, write five lines somewhere durable: what work, why in-house or outsourced, what success looks like in 90 days, who manages it, and when you will revisit. That note prevents re-litigating the decision every time someone is frustrated with a contractor invoice or a slow hire. Frustration is data; it is not automatically a reason to flip the model without checking the original logic.