The comparison that decides nothing
A director gets approval for a technical role, opens a spreadsheet, and puts two numbers side by side: a salary divided by 2,080 hours, and an outside firm's hourly rate. The rate is two to three times higher. The meeting ends there. That comparison feels rigorous and it is nearly useless, because the two paths do not buy the same thing. One buys a person who will still be there in three years and whose cost continues whether or not there is work in front of them. The other buys a defined result on a schedule that stops when you stop it. Pricing those against each other per hour is like comparing a mortgage payment to a hotel bill and concluding that hotels are expensive.
The useful version of this decision takes about an afternoon and turns on two questions that most staffing conversations never reach. How long will this work exist? And is it the thing you sell? Everything else — rate, loaded cost, notice period, tax treatment — is downstream of those two answers, and when the answers are clear the money question usually resolves itself.
What follows is written from the seller's side of the table, which is worth saying out loud. We are an engineering firm that gets hired on contract. That means the sections below where contracting is the wrong answer are the ones worth reading most carefully, because they are the ones we have an incentive to leave out.
You are probably here because
- A role has been open for four months and the work behind it has not moved
- Finance approved project budget but not headcount, and you are trying to work out what that actually permits
- You hired for this once, the person left inside a year, and you are unsure whether to repeat it
- Someone has been invoicing as a contractor for two years and sits in your standups every morning
The first three are answered by the duration-and-centrality test below. The fourth is not a staffing question at all — it is the classification section, and it is the one with legal consequences.
The two questions, in order
How long will this work exist? Not the project. The work. A migration off a legacy system is eighteen months and then it is over forever. Running the platform that replaces it is permanent. Building a forecasting model is a quarter of work; maintaining it, retraining it, and defending its numbers to a finance committee is a job that exists as long as the company does. Write down the honest number of months. Most requests we see for a permanent technical hire are attached to work with a life of nine to eighteen months, and nobody has said so out loud because the requisition form has no field for it.
Is it the thing you sell? There is a real distinction between capability that is core to the product or the mission and capability that supports it. If your company sells software, the people who write that software are core, full stop. If your company sells insurance and needs a data platform so the actuaries can work faster, the platform is supporting. Supporting is not lesser — it can be the highest-return work in the building — but it does not have to live inside the payroll to be under your control.
Those two answers place you in one of four boxes, and each box has a default. Defaults can be overridden, but you should be able to say why.
| Core to what you sell | Supporting the business | |
|---|---|---|
| Permanent work No end date you can name | Hire. Employment is the cheapest way to hold context over five years, and no contract structure changes that | Hire, or buy it as a service. A managed platform with a real support agreement often beats both a hire and a contract here |
| Bounded work Ends on a date you can name | Contract the build, keep the design. Outside hands are fine; the architecture decisions and the acceptance criteria stay with your people | Contract it. This is the clearest case there is, and it is the one most often turned into a hire because a requisition was the available tool |
The lower-left box is the one that causes the most trouble. Companies read "core" and conclude "employee," and then hire three people for an eighteen-month build and spend month nineteen finding somewhere to put them. Core does not mean permanent. It means you cannot outsource the judgment. You can absolutely outsource the typing.
What the employee actually costs
Salary divided by 2,080 is wrong before you add a single line item, and the size of the error is measurable. The Bureau of Labor Statistics publishes employer compensation costs directly in its Employer Costs for Employee Compensation series. For March 2026, private-industry employers paid an average of $32.60 per hour worked in wages and salaries and $14.01 per hour worked in benefits; wages were 69.9 percent of total compensation and benefits were 30.1 percent.
The dollar figures are an economy-wide average and say nothing about what a senior engineer earns. The ratio is the part that transfers. If you have budgeted a role at its salary and stopped, you have budgeted roughly seven-tenths of the employer's real compensation cost. Then add the things that sit outside compensation entirely: recruiting fees, hardware, software seats, the training budget, the office cost if you have one, and the senior hours spent interviewing and onboarding rather than building.
The number almost nobody computes is the cost of the seat staying open. Work that is waiting is worth something per week, and that value is being forgone right now. If a hire that starts in month five is compared against a contract that starts in three weeks, four months of not-happening belongs on the same page as the salary. Leave it off and the cheaper-looking option is frequently the one that costs more.
What actually decides the call — our weighting
Weights sum to 100. Our starting point when we are asked to help with this decision, not a measurement. The bottom row is where most of the meeting goes.
Reversibility is worth paying for
The two options are not symmetrical when you turn out to be wrong, and that asymmetry is priced into the rate whether or not anyone names it. Ending a contract means serving notice and paying to the end of it. Ending an employment relationship means severance, a difficult conversation, a hole in a team, and in some jurisdictions a process measured in months. Making a wrong hire is expensive. Making a wrong contract is an inconvenience.
That matters most exactly when you are least sure. If you cannot describe what the person will do in their first month in one sentence, your uncertainty is high, and buying the reversible option at a premium is the rational purchase. The premium is the price of the option, not a markup you failed to negotiate away.
The inverse holds too, and it is the strongest argument against contracting anything permanent. Once you are certain the work is permanent and central, the reversible option is the wrong purchase and you are paying an option premium every month for flexibility you will never exercise. That is how a two-year staff-augmentation contract quietly becomes the most expensive way to have employed somebody.
The constraint nobody budgets: management attention
Both paths consume the time of the person they report to, and that time is usually the actual bottleneck. A new employee needs onboarding, direction, feedback, and someone to answer questions for months. An outside team needs a decision-maker who can answer questions inside a day and accept or reject work against written criteria. Neither runs unattended.
The difference is the shape of the load. Employment front-loads it heavily and then tapers to a low ongoing cost. A well-scoped contract concentrates it into the scoping and the acceptance, with a lighter middle. If the sponsor has no capacity at all in the next month, a hire will fail for reasons that have nothing to do with the candidate, and a contract will fail for reasons that have nothing to do with the firm.
The honest test: ask who will spend five hours a week on this person for the first two months, get a name, and check that name's calendar. If there is no name, fix that before spending anything. This is the single most common reason both paths fail, and it is invisible in every cost comparison we have ever been shown.
A third option that is usually skipped
Between "hire someone" and "contract the build" sits a bounded assessment: a small, paid, two-to-four-week engagement whose only deliverable is a written recommendation with an accountable author. It costs a fraction of either path and it frequently changes which path you take — most often by revealing that the work is smaller than assumed, or that the real blocker is a decision rather than capacity. It also shows you what good looks like in a specialty you cannot yet interview for, which is worth more than a reference call.
The line you cannot paper over
If the plan is to hire a person, call them a contractor, and then manage them exactly like an employee, that is not a staffing strategy. It is a classification question with tax consequences, and it is decided by facts rather than by what the agreement says.
Under the common-law rules the Internal Revenue Service applies, a worker is an employee if the business has the right to control and direct what will be done and how it will be done. The evidence falls into three categories: behavioral control — whether the business directs how the work is performed, through instruction, training, or other means; financial control — how the worker is paid, whether expenses are reimbursed, who supplies the tools; and the type of relationship — written contracts, employee-type benefits, whether the relationship is expected to continue, and whether the work is a key aspect of the business. A firm or a worker who wants a determination can file Form SS-8, and the IRS notes it may take at least six months to get one.
Read that list against the arrangement you are contemplating. An individual working your hours, on your equipment, under your daily direction, indefinitely, on work that is central to your business, has most of the markers of employment regardless of the invoice. Federal and state labor authorities apply their own tests as well, and those have moved more than once in recent years, so this is a question for your counsel rather than for a staffing conversation. The structural point stands on its own: contracting works when you are buying a scoped result from a firm that directs its own people. It stops working the moment you are buying a person and directing them yourself.
Contract-to-hire, and how it usually goes
Contract-to-hire is a genuinely good structure for the case where you cannot yet evaluate the specialty, and it disappoints when it is used as a way to avoid deciding. Three things are worth knowing before you write one.
The conversion fee is negotiated at the start or not at all. Agree the buyout, or the fee-free conversion date, in the original agreement. Raising it at month five, when both sides know the person is working out, is the worst possible time to have the conversation and you will lose it.
The strongest candidates convert least often. A senior engineer running an independent practice usually prefers that practice. If your plan requires converting them, say so on day one and find out. Discovering it in month six costs you the ramp twice.
An individual contractor and an engineering firm are different purchases. You can convert a person. You cannot convert a firm, and asking to hire its engineers away is a conversation that ends most relationships. If your intent is eventual employment, contract with an individual or with a firm that offers a stated conversion path, and put it in writing.
The sequence that usually works: contract the first, hire the second
For a company standing up a technical capability it has never had, the ordering that fails least often is not "hire a senior person and let them build the team." It is to contract the first instance of the work, require documentation as a named deliverable rather than a courtesy, and use what you learn to write an accurate job description for the permanent role.
The reason is unglamorous. Before the first system exists, nobody in the building knows what the job actually is, so the requisition gets written from a template and describes four careers in one person. After the first system exists, you know which parts consumed the schedule, which decisions were hard, and what the person maintaining it will really spend their days on. Job descriptions written after that are dramatically more accurate, and the searches close faster because the role is a real job rather than a wish list.
The condition on this working is documentation. A working system delivered by an outside team, with no runbook and no rehearsed handoff, leaves you dependent on the firm that built it — and firms that are comfortable with that outcome exist. Require the runbook, the dependency list, the access inventory, and a handoff rehearsal held while the engagement is still healthy, all as named deliverables with acceptance criteria. That single requirement converts contracting from a dependency risk into a knowledge transfer.
What each option is genuinely bad at
- Contracting is bad at open-ended work where nobody can write down what done means — the scope moves, the price does not, and both sides end up unhappy
- Contracting is bad at accumulating institutional memory unless documentation is a contractual deliverable rather than a courtesy at the end
- Contracting is bad at anything requiring daily judgment about your business that only comes from being in the room for a year
- Hiring is bad at deadlines inside the hiring cycle — a role filled in month five cannot rescue a milestone in month three, no matter how strong the candidate
- Hiring is bad at specialties you cannot evaluate, because a company that cannot interview for a skill will select on confidence rather than competence
- Hiring is bad at one-time work, and the failure shows up as a capable person with nothing to do in year two and no graceful way out
- Both are bad without a named sponsor who has real time on the calendar to direct the work
A six-week path to a defensible decision
Staffing decision sequence
Step five is the one that gets cut and the one that pays. Running both paths in parallel costs a few hours of somebody's week and gives you real information: what the market will actually pay for this profile, how long the search will really take, and what an outside firm proposes to build. Running them sequentially — four months of searching, then a contract discussion in a panic — is how organizations end up with the option they had time for rather than the one they wanted.
Before you open a requisition
- The work is written as outcomes with an honest end date, not as a job title
- One defensible number exists for what a month of delay costs
- The loaded cost includes the employer's benefit burden, not just salary
- The capability has been classified as core or supporting, in writing
- A named sponsor has confirmed weekly time on their calendar
- You can state what this person does in month one, in one sentence
- If contracting, documentation and a handoff rehearsal are named deliverables
- If contract-to-hire, the conversion terms are written before anyone starts
- Nobody is being engaged as a contractor to do an employee's job under your direction
- Both paths were priced in the same week, not four months apart
Bottom line
Duration and centrality decide it. Permanent and core is a hire, and the sooner the better. Bounded and supporting is a contract, and the hourly rate is not the interesting part of that decision. The two mixed boxes are where judgment is actually required, and the way through them is to separate the judgment from the hands — keep the design decisions and the acceptance criteria inside, and be relaxed about where the implementation comes from. Then check the two things that sink both options regardless of which you pick: whether a real deadline falls inside your hiring cycle, and whether anybody has the calendar space to direct the person once they arrive. Most staffing decisions we are asked to help with are lost on those two, long before anyone gets to the rate.
Frequently asked questions
Per hour, almost always. Per outcome, frequently not. The comparison has to include the employer's benefit burden, which BLS measured at 30.1 percent of total compensation for private-industry workers in March 2026, plus recruiting, equipment, software, and the senior hours spent hiring and onboarding. Then it has to include the cost of the work not happening during a search. Once those are on the page, bounded work usually comes out cheaper on contract and permanent work usually comes out cheaper on payroll.
There is no universal threshold, but the practical break tends to land around the point where the ramp is a small fraction of the total: roughly eighteen to twenty-four months of continuous, similar work. Below that, the hiring cycle and the onboarding consume too much of the value. Above that, employment is the cheapest way to hold context and the option premium on flexibility stops being worth paying.
Yes, and it is a good structure when you cannot yet evaluate the specialty. Two conditions make it work: agree the conversion fee or the fee-free date in the original agreement rather than at month five, and state your intent to convert on day one so the person can tell you whether they want that. Note that you can convert an individual, but asking to hire the engineers away from a firm you contracted is a different and usually relationship-ending conversation.
Worker classification is determined by facts, not by the label on the agreement. The IRS applies common-law rules that weigh behavioral control, financial control, and the type of relationship — including whether the work is a key aspect of the business and whether the relationship is expected to continue. Either party can request a determination on Form SS-8, which the IRS says may take at least six months. Because federal and state tests differ and have changed, this belongs with your counsel rather than in a staffing plan.
It settles what you are able to do this quarter, which is not the same as what you should do. If the work is genuinely permanent and central, contracting it under a freeze is a workaround with a running cost, and the right move is to do the bounded piece now while making the case for the role with evidence from that work. Treating a budget constraint as a strategic answer is how organizations end up with multi-year staff augmentation that costs more than the role they were not allowed to open.