Two proposals are open on your screen for the same problem. One is from a recruitment agency: 20% of first-year salary, paid when the person starts, 90-day replacement guarantee. The other is from a firm calling itself a staff augmentation partner: $85 an hour per engineer, three-month minimum, no placement fee at all. A third vendor emailed this morning offering “managed capacity,” which nobody has defined. You need four more engineers by Q1 and you are trying to work out whether these people are selling the same thing with different paperwork.
They are not. Staffing and staff augmentation differ on four things: who directs the day-to-day work, who legally employs the worker, who carries which risk, and how the vendor earns its margin. Everything else, including the confusion, follows from those four.
The short version. Permanent placement sells you a person you then employ, and the vendor is done. Contingent staffing and staff augmentation sell you a person the vendor keeps employing while you direct the work, so the vendor earns on every hour. Project outsourcing sells an outcome and keeps both the people and the knowledge. If you want capability that compounds inside your own company, augmentation is the model that can deliver it, but only under conditions most buyers never set up.
The words, defined precisely
Most of the argument online is people using one word for several different products. This grid separates them.
| Model | Directs the work | Employs the worker | Carries the risk | How you pay |
|---|---|---|---|---|
| Permanent placement | You, day one | You, day one | You, once the guarantee lapses | One-time fee, a percentage of first-year salary |
| Contingent staffing / temp | You | The staffing firm | Firm: payroll and benefits. You: output and joint-employment exposure | Hourly bill rate = pay rate + markup |
| Contract-to-hire | You | The firm, then you | Shared, shifting to you at conversion | Hourly, plus a conversion fee that declines with tenure |
| Staff augmentation | You, inside your team and process | The vendor or its subcontractor | Vendor: employment admin. You: direction and joint-employment exposure | Hourly, or fixed monthly per head |
| Managed capacity / dedicated team | The vendor’s lead, against your priorities | The vendor | Shared: you own the roadmap, they own delivery mechanics | Fixed monthly per head or per pod |
| Managed service / project outsourcing | The vendor, entirely | The vendor | Vendor carries delivery risk against a defined scope or SLA | Fixed price, or fixed monthly for an outcome |
| RPO | You set strategy, vendor runs the function | Vendor employs recruiters, you employ the hires | Vendor carries process and throughput risk | Monthly fee, per-hire fee, or hybrid |
| EOR | You | The EOR, as legal employer where you have no entity | EOR carries statutory employer duties | Monthly per-employee fee on top of salary and statutory costs |
| C2C (corp-to-corp) | Varies | Nobody, in the W-2 sense: the worker’s own company contracts in | Classification risk concentrates here | Hourly or daily, invoiced company to company |
Two clarifications that save arguments. C2C is a contracting structure, not a service model. It describes how money moves, not who directs the work, and it sits underneath staffing, augmentation and projects alike. It is also where classification exposure is densest, because it rests on the worker being a genuine business rather than somebody’s employee in a costume. And staff augmentation is not a synonym for offshore contracting, though vendor marketing welded the words together. Augmentation describes a relationship: their employee, your direction, your process. Geography is a separate decision.
How each model makes money, and what that makes the vendor do
You are not choosing a model. You are choosing an incentive, and then living with it.
Permanent placement: a percentage of first-year salary, once. Fees in the 15% to 25% band are what the industry press and most agency contracts show, but no authority publishes them and they are negotiable, so treat that range as a convention rather than a fact. The incentive is honest and narrow: close the hire, at a good salary, because the fee scales with the salary. Once the replacement guarantee lapses, typically after 30 to 90 days, the agency’s economic interest in that person is zero. Not cynicism. Arithmetic. If you want an agency to care about month nine, pay for month nine.
Contingent staffing and hourly augmentation: a markup on the pay rate. The vendor pays the worker, absorbs employer burden (the employer half of FICA at 7.65%, unemployment insurance, workers’ compensation, any benefits), and bills you more. The incentive is hours, which aligns the vendor with retention and misaligns it with efficiency, because nobody there gets richer by finishing early. Convert someone onto your payroll and the vendor loses an annuity. That is the entire reason conversion fees exist, and knowing it changes how you negotiate them.
Fixed monthly per head: the spread between your price and their loaded cost. Easier to budget, ends the timesheet argument. Two risks: silent substitution of the person you interviewed, and the vendor moving your strongest engineer to a higher-margin account once your engagement looks stable. Name individuals in the contract, require written notice before any change.
Project outsourcing: fixed price minus actual delivery cost. Margin comes from the fewest and cheapest hours that will pass acceptance, which pushes toward junior staffing and toward treating every change of mind as a change order. Fine when scope is genuinely fixed. Miserable when it is not. RPO sits between: pure per-hire behaves like contingency at scale, pure monthly can drift away from results, and hybrid with time-to-fill SLAs is where workable contracts land.
Six criteria that actually decide it
- Duration. Under six months, hiring is the wrong instrument; you will spend the engagement recruiting. Three years or more on core work, and any contingent model is a slow way to avoid a decision.
- Defined or evolving work. A migration with known endpoints can be a fixed-scope project. A product area whose priorities move monthly should never be, because every change of direction becomes a negotiation instead of a standup conversation.
- Management capacity. The most common and least discussed failure. One competent lead directs roughly five to seven people well, and adding a sixth engineer under a lead already running seven reduces throughput. No spare direction capacity means you want managed capacity, where the vendor supplies the lead too.
- Speed. A US direct hire runs weeks to months from open requisition to first day, plus a notice period. Augmentation runs days, because the people already exist on someone’s payroll. A six-week deadline has already chosen your model.
- Whether the knowledge must stay in-house. The whole answer to the growth question, below.
- Compliance exposure. No entity in the country your candidate lives in is an EOR question, not a staffing question. Where workers touch regulated data or production credentials, screening and access design matter more than the engagement model does.
Co-employment risk when you direct someone else’s employee
General information, not legal advice. These rules are jurisdiction-specific, several moved during 2026, and any real arrangement should go past employment counsel.
“Co-employment” is not a single legal status you either have or avoid. It is shorthand for separate exposures under different statutes, each with its own test. Three matter.
Worker classification, under the IRS common-law test. The IRS position is blunt: “anyone who performs services for you is your employee if you can control what will be done and how it will be done,” and “what matters is that you have the right to control the details of how the services are performed,” whether or not you exercise it. Three categories get weighed: behavioral control, financial control, and type of relationship. Publication 15-A sets it out; Form SS-8 exists if you want an official determination. Notice the trap: directing the details of how work is done is exactly what staff augmentation is for, which is why the model belongs on top of a properly employed worker, the vendor’s W-2 or an EOR, rather than a thin independent contractor arrangement.
Joint employment under the National Labor Relations Act. The NLRB withdrew the 2023 standard, which a federal court had vacated, and reinstated the 2020 rule effective 27 February 2026. Under it you are a joint employer only if you “share or codetermine” the employees’ essential terms and conditions of employment and exercise “substantial direct and immediate control” over one or more of eight listed terms: wages, benefits, hours of work, hiring, discharge, discipline, supervision, and direction. Control is not substantial if exercised only on a “sporadic, isolated, or de minimis basis.” That is a higher bar than the vacated 2023 version set, and it is the standard in force as this is written.
Joint employment under wage and hour law. On 22 April 2026 the Department of Labor proposed a rule covering joint employer status under the FLSA, FMLA and the Migrant and Seasonal Agricultural Worker Protection Act, comments closing 22 June 2026. It proposes four factors for vertical joint employment, the staffing-client situation: whether the potential joint employer hires or fires the worker, substantially supervises and controls their schedule or conditions of employment, determines their rate and method of pay, and maintains their employment records. It is a proposal, not a final rule. The consequence is not in dispute: joint employers are jointly and severally liable for unpaid minimum wage and overtime, so a vendor’s payroll failure lands on your balance sheet too.
These are different questions with different answers, which is what people muddle. Being a joint employer for wage-hour purposes does not make the worker your common-law employee for tax purposes, and a worker correctly employed by a staffing firm can still make you a joint employer under the FLSA.
The controls that reduce exposure are boring and effective. Route pay rates, discipline, formal performance management and termination decisions through the vendor. Keep contingent workers out of your review cycle, bonus plan and benefits enrolment. Set assignment end dates and observe them. Verify the vendor genuinely runs payroll rather than passing through C2C invoices you assumed were W-2. State law adds a layer, California especially.
The knowledge problem, which is the actual growth question
After the engagement ends, where does the knowledge live?
With project outsourcing it leaves with the vendor. You receive an artefact. The reasoning behind it, the approaches tried and abandoned, the undocumented quirk in the payment provider’s API, all of that sits in the heads of people now working on someone else’s account. You own the code and none of the context, which is why the next vendor’s estimate to change it is always higher than you expected.
Staff augmentation can keep the knowledge, and that is its genuine structural advantage over both alternatives. “Can” is load-bearing. Four things decide it:
- The work happens in your systems. Your repository, issue tracker, cloud account, incident channel, documentation. Not theirs with a periodic export. This single decision does more than the rest combined.
- Decisions get written down where you keep things. A short decision record or ticket comment for anything non-obvious. The test: someone joining in eight months can reconstruct why, without asking anyone who has left.
- An employee owns every area. Augmented people contribute to areas a named employee is accountable for. If nobody on your payroll owns it, you have outsourced it at augmentation prices.
- Pair on first instances, and keep your own people on call. The first two times any recurring task happens, an employee does it alongside the contractor. On-call is where real system knowledge accumulates, so never let the vendor absorb it permanently.
Contract the handover while everyone is still pleased with each other: defined transition period, agreed rate, named deliverables. And the honest caveat, augmentation retains knowledge only if you have employees to retain it in. A team that is 80% contracted has project outsourcing’s knowledge-flight problem with a worse cost structure.
What each costs over one, two and three years
Take a role you would pay $150,000 in base salary. For scale, the Bureau of Labor Statistics puts the median annual wage for software developers at $135,980 as of May 2025, so this is a solid mid-to-senior engineer, not a star. Employing that person costs more than the salary: the BLS Employer Costs for Employee Compensation release for June 2026 shows benefits at 30.0% of total compensation for private industry workers ($14.07 of $46.89 per hour worked). Gross $150,000 up on that basis and loaded cost is about $214,300 a year, before recruiting, equipment or manager time. That is an all-industry average; tech benefits often run richer.
- Direct hire through an agency. Year 1: a $30,000 fee plus $214,300 loaded, so $244,300. Cumulative year 2: $458,600. Year 3: $672,900. Raises push this up; the fee never repeats.
- Staff augmentation at $85 an hour. At 2,080 hours, $176,800 a year, with no placement fee, no benefits and no severance. Add 3% to 5% annual rate escalation and three years lands near $552,000. Cheaper in cash every year.
- Contract-to-hire. Six months at $85 is $88,400, then a conversion fee (say $15,000 on a declining scale), then six months employed at $107,150. Year 1 is $210,550, the cheapest first year of the three. Years 2 and 3 revert to $214,300, for a three-year total near $639,000.
The number worth extracting is the break-even bill rate. Divide the loaded annual cost of the employee by 2,080 and you get the hourly rate at which augmentation stops being cheaper. Here that is about $103 an hour. Below it, augmentation wins on cash every year; above it, the employee wins from year two, and for a twelve-month need the break-even climbs to roughly $117 once the placement fee is amortised.
Cash is not the whole comparison. At $85 an hour you pay less and get less: no claim on the person’s future, no retention mechanism beyond the rate, and a price that resets at every renewal. The employee costs more and can still be there in year six, holding the context that makes year six cheap.
What to ask each vendor type
A permanent placement agency: is the guarantee a replacement or a refund? What is the fee on a candidate we were already talking to? Who else are you presenting this shortlist to?
A staffing or augmentation vendor: is this person your W-2 employee, an EOR employee, or a C2C subcontractor, and can we see the paperwork? Does the quoted rate include employer burden? What is the conversion fee at 3, 6 and 12 months, and when does it reach zero? Will the people we interview be the people who work? Is there a window at the start of an assignment in which we can reject someone unbilled? Show me your indemnity language for wage-and-hour claims.
A project or managed service vendor: what is explicitly out of scope? What is the change-order process and rate? Who owns the IP, and does the assignment survive a payment dispute? What does handover include, and is it priced in the contract today?
So which one supports long-term growth?
The either/or framing is the mistake. Companies that grow well run both deliberately: a core of employees who own the domains, a flex layer sized to how uncertain demand is, and project outsourcing kept for bounded work far from the product.
The question still deserves a direct answer. Staff augmentation supports long-term growth better than permanent placement alone or project outsourcing, under three conditions: employees who own each area, spare management capacity to direct the work, and the discipline to run everything inside your own systems. Fail one and augmentation degrades into expensive project outsourcing with extra compliance risk. Permanent placement supports growth more bluntly and more durably, by converting into headcount you own, and its case strengthens the longer the need lasts and the closer the work sits to what you sell.
If you are selling rather than buying, the same arithmetic points somewhere specific. A $30,000 perm fee is high-margin, lumpy and gone; the account resets to zero and you hunt the next requisition. Put that person on contract at $85 against a $60 pay rate and the spread is $25 an hour, roughly $18 after employer burden, about $36,000 across a year of assignment, and it recurs while they stay. Those figures are only as good as your own rate card, but the shape holds: contract and augmentation build enterprise value where placement builds cash flow. The catch is working capital, because you pay the contractor weekly while the client pays in 45 to 60 days. Contract staffing is a financing business wearing a recruiting business’s clothes. The market is big enough to plan around: the American Staffing Association, reporting member-supplied data, counts about 2.2 million temporary and contract employees working for US staffing firms in an average week in 2024, and Staffing Industry Analysts forecasts the US market at $183.1 billion in 2026.
AB7 Solutions runs both sides of this comparison, which is why we will tell you when you have picked the wrong one: contract staffing and staff augmentation with named people working inside your repository, your tracker and your definition of done, alongside permanent recruitment, RPO and C2C when the right answer is headcount you own. If you have a bill rate and a placement fee in front of you and want someone to run the break-even, read the conversion clause and check whether the worker is genuinely W-2, call +1 321 341 7733, write to ab@ab7solutions.com or director@ab7solutions.com, or start at www.ab7solutions.com.
Sources: IRS, Employee (common-law employee) and Independent contractor (self-employed) or employee?; National Labor Relations Board, Withdrawal of 2023 Standard for Determining Joint Employer Status, effective 27 February 2026; U.S. Department of Labor, NPRM: Joint Employer Status Under the FLSA, FMLA, and MSPA, 22 April 2026 (proposed; comments closed 22 June 2026); U.S. Bureau of Labor Statistics, Employer Costs for Employee Compensation, June 2026 and Occupational Outlook Handbook, Software Developers; American Staffing Association, Staffing Industry Statistics (member-reported); Staffing Industry Analysts, US Staffing Industry Forecast, September 2026 Update.
Related reading
Related reading
Related AB7 service: remote IT and technology staffing.