MSP Staffing: How the Managed Service Provider Model Actually Works

An MSP, or managed service provider, is a third party that runs your contingent workforce program. They take in requisitions, route them to a panel of staffing suppliers, manage the timesheets and the invoicing, and own the compliance paperwork. The model is built for enterprises with a lot of contractors and not enough internal capacity to herd them. That last part is the part most vendor pages skip.

I should disclose where I am sitting. KORE1 is one of the suppliers that gets routed to under MSP programs. We win some of those reqs and we lose money to fee compression on others. So this guide is going to be honest about both sides, including the part where the model genuinely solves problems and the part where it costs you more than it saves. If you are weighing whether to stand up an MSP for your contingent workforce, you deserve to hear from someone on the receiving end of the supplier panel, not just the firms selling the model.

For context, this whole conversation sits inside a broader IT staffing program. The MSP layer is what runs on top.

MSP Staffing in 60 Seconds

An MSP is a managed third party that owns the lifecycle of your contingent workers. Requisition intake, supplier coordination, timekeeping, billing, compliance, offboarding. It is a service wrapped around a software platform called a VMS, run by a project team that usually sits partly on site. Companies typically reach for one when they have more than a hundred contractors and the spend has stopped being legible. At that scale, most buyers also need a staffing partner with enterprise staffing solutions that plug into the program cleanly.

That last sentence is the real trigger. Not headcount. Legibility.

What an MSP Actually Does, Day to Day

The clean version of the lifecycle goes like this. A hiring manager opens a req. The MSP intake team validates it, scrubs the description, and pushes it into the VMS. Suppliers on the panel receive the req at the same time, or in tiers if it is a tiered program. They submit candidates back through the platform. The MSP screens, the manager interviews, an offer goes out, the contractor gets onboarded against a standardized compliance pack. Time goes in weekly. Invoices consolidate to a single AP entry. When the engagement ends, offboarding runs through the same machinery.

That is the brochure.

The version that lives in our queue is messier, and the gap between the brochure version and the operational version is where most of the friction in the model actually lives, even though it almost never makes it into the procurement deck. Reqs sit in the VMS for two days before they get released to suppliers, often longer when the intake team is short staffed or when the requisition needs a re-write because the manager pasted in a job description that violates the program’s standardization rules. Submission caps mean we can only put forward three candidates per role even when we have eight strong ones, so the recruiter on our side ends up making a call about which candidates to hold back, which is exactly the kind of editorial filter that the original promise of vendor-neutral competition was supposed to remove from the process. The MSP recruiter assigned to the req has 60 other reqs, so feedback takes a week. By the time a contractor starts, the manager has often moved on to a different problem and the urgency that opened the req is gone. None of that is the MSP being lazy. It is the structural cost of putting a coordinator between a hiring manager and a recruiter.

The trade is real. You buy compliance and visibility. You pay in speed and intimacy.

Ardent Partners, which has been tracking this market longer than most, puts the contingent share of the average enterprise workforce at roughly 43 percent. That is a lot of people to manage with a spreadsheet. The MSP exists because the spreadsheet stopped working.

The Three MSP Program Models

Not all MSPs are built the same. There are three structural shapes, and the difference matters more than most procurement leaders are told before they sign.

Master vendor. One staffing firm sits at the front door. They fill what they can themselves and subcontract the rest to a deeper bench. Fast to set up. Easy to govern. The catch is that the master vendor has an obvious incentive to fill from inside its own bench first, and a less obvious incentive to keep the second tier hungry enough to compete but not strong enough to displace them. We have been the second tier in master vendor programs. The reqs you get are the ones the primary could not fill, which is a polite way of saying the hard ones.

Vendor neutral. The MSP runs the program but is contractually barred from filling reqs themselves. Every req goes to every supplier on the panel at the same time. In theory this maximizes competition and quality. In practice the panel still has a pecking order. It is just informal instead of contractual. Some suppliers learn the MSP recruiters by name and get a quiet edge on the reqs that matter.

Hybrid. Most large programs end up here, even when they were sold as one of the first two. A primary panel of three to five suppliers gets first look on most reqs. A deeper bench picks up the long tail. The primary panel does maybe 70 percent of the volume. Everyone else fights over the remaining 30. If you are evaluating MSPs and one of them tells you the model is “purely vendor neutral,” ask how many suppliers are in the primary panel. If the answer is fewer than ten, it is hybrid. Just call it that.

Source: KORE1 supplier panel observations and the Staffing Industry Analysts MSP Global Landscape Summary 2024.

MSP vs VMS vs RPO

Three acronyms, constantly confused, three different things.

A VMS is software. SAP Fieldglass, Beeline, Coupa Contingent Workforce. It is the database and workflow engine that holds reqs, candidates, timesheets, and invoices. Companies can technically buy a VMS without buying an MSP, and a few do. Most do not, because the software needs people running it.

An MSP is the people running it. Plus the program design, the supplier negotiations, the compliance audits, the dashboards procurement actually reads.

An RPO, recruitment process outsourcing, is the same idea applied to direct hires. Different talent pool, different velocity, different SLAs. RPO and MSP can run side by side at the same company. They do not overlap.

How MSP Fees Actually Work

This is the section the vendor pages skim through. So we will sit on it.

There are two funding models. Client funded, where you pay the MSP directly, usually as a flat percentage of program spend. Supplier funded, where the MSP charges the staffing suppliers a percentage of every bill rate, and the suppliers absorb that fee. Eighty percent of programs use percent-of-spend pricing in some form, according to Staffing Industry Analysts benchmarks. The typical range is 1.5 to 3.5 percent.

Supplier funded is the popular option because it looks free. It is not free.

Here is the math the procurement deck does not show you. Imagine a senior Java developer placed at a $135 per hour bill rate. The MSP fee is 2.5 percent. The staffing supplier owes the MSP $3.38 per hour for the life of that contract. Over a 12 month engagement at 2,000 hours, that is $6,750 the supplier never sees. The supplier did not invent that money. It came out of one of two places. Either the supplier shaved its own margin, which works for one quarter and then they stop putting their best candidates on your reqs. Or the supplier shaved the contractor’s pay rate, which is what usually happens. The contractor at $135 bill is now taking home $80 an hour instead of $84. They do not know why. They look at the same role on a direct staffing posting from a different agency, see the better take-home, and the next time their contract comes up they are gone.

This effect compounds across the panel over time, and the way it compounds is the part that does not show up on a quarterly dashboard until the fill rate has already cratered. The suppliers most willing to absorb fee compression are usually the ones with the weakest candidate pool, which means the panel quietly self-selects toward whoever is desperate enough to keep playing for shrinking margins, while the firms with the best benches stop submitting their A-list candidates and route them to direct accounts that pay full margin instead. After about 18 months the strongest supplier panels stop competing hard for tier 2 reqs because the unit economics stop working, and the contractors who do still come through are the ones who could not get placed anywhere else first. Procurement looks at the dashboard and sees stable fill rates. The dashboard does not measure who is being submitted, only how many.

So the model is not bad on its own terms, but it is also not the savings machine the slide deck promised, and the difference between those two framings is exactly the gap most enterprises do not figure out until they are 14 months into a three-year contract. If you want to sanity check rates against the open market before you sign anything, our salary benchmark assistant pulls live data on the most common contract roles, and it will at least tell you whether the pay rates inside your VMS look like the rates a candidate could get on the same role from a direct staffing engagement somewhere else.

Client funded fees, by the way, tend to produce better supplier behavior because the cost is visible and the suppliers are not absorbing it. They also tend to be slightly higher percentages, because procurement gets a clearer view and pushes back. There is a reason most enterprises pick the supplier funded model anyway. It is easier to defend in a budget meeting. Whether it is actually cheaper is a different question.

When an MSP Is the Right Call

MSPs solve a real problem. Not always the one buyers think they are solving, but a real one.

The right call usually looks like this. You have somewhere north of a hundred active contingent workers. They are spread across a dozen suppliers, none of whom talk to each other. Procurement does not know what you are spending in total because the invoices come in different formats and three different AP buckets. Compliance is a question nobody can answer in writing. There was an incident last year where a contractor’s I-9 was missing and you found out from a state audit instead of from the supplier. The CFO has started asking pointed questions in QBRs.

That is the company that needs an MSP. The pain is not headcount. The pain is opacity.

An MSP gives you a single dashboard, a single compliance posture, a single point of escalation, and a defensible answer when legal asks who is on the floor and under what terms, all consolidated into a program that one person can actually defend in front of the audit committee without sounding like they are guessing. The fees are real, but the alternative is worse, because the alternative is the audit you cannot pass and the regulatory exposure that comes with not being able to produce a clean roster of who is doing work in your buildings under what terms on what budget code.

The other clean fit is regulated work. Healthcare systems, financial services, anything with a PII or PHI exposure surface. The compliance overhead alone justifies the program.

When an MSP Is the Wrong Call

This is the section that gets posts like this one buried by the MSP vendors. So be it.

If you have fewer than about 150 contingent workers, the MSP fee will probably eat any procurement savings the program produces. The math we walked through earlier scales both ways. At low volumes the percent of spend is small in absolute dollars but the implementation cost, the supplier panel rebuild, and the speed degradation are the same as they would be at five times the volume. We have watched two companies in the 60-to-80 contractor range stand up MSPs because their parent company pushed them into it. Both unwound the program inside 18 months. Both spent more than they saved.

If your contingent work concentrates in hard-to-fill skills, the MSP model is also a worse fit than the marketing suggests. Niche stacks live on relationships. The recruiter who has been placing your Snowflake engineers for three years has a passive bench that does not respond to VMS reqs from an unfamiliar firm. Put that recruiter inside an MSP queue with a 72 hour SLA and you will lose access to the candidates you actually want. We have seen this play out in cybersecurity staffing and high-end cloud engineering placements over and over. The candidates are not in any database. They are in someone’s contact list.

If you have one or two strong staffing partners and a manageable spend, you almost certainly do not need an MSP, and the honest move is to invest in the relationship you already have rather than overlay a four-month implementation and a 2 to 3 percent fee on a system that is mostly working. You need better invoicing discipline and a quarterly review. That is a smaller fix.

If your contingent program is greenfield with no baseline data, the MSP will spend the first year discovering what you do not know about your own spend, building a supplier panel from scratch, negotiating master agreements with firms they have never worked with, and producing the first three quarters of dashboards from a dataset that did not exist 90 days earlier. You are paying them to learn the company. That is a real service, but it is not a contingent workforce program in the sense that the marketing pitch describes. It is a consulting engagement with a different invoice format, and you should price it as one before you sign anything.

What Changes for Your Existing Staffing Suppliers

If you stand up an MSP, your current suppliers will not quietly disappear. Most will sign the supplier agreement, swallow the fee, and try to stay on the panel. A few will walk. The ones who walk are usually the smaller boutique firms whose unit economics cannot survive a 3 percent haircut. Sometimes those are the firms doing your hardest fills.

The suppliers who stay will adjust. Day to day, the senior recruiter you have been calling for two years now has to route everything through the VMS. The relationship is not gone. It is just slower and more formal. Pay rates for new contracts often drift down by 3 to 5 percent in the first year as suppliers absorb the fee. New candidate quality usually softens over the same window, although it can take procurement another year to notice because the dashboards do not measure what they should.

The supplier ecosystem you have today is not the supplier ecosystem you will have 18 months after the program goes live, and the speed of that turnover depends almost entirely on whether your MSP is one of the handful that takes panel health seriously or one of the many that treats supplier churn as somebody else’s problem. Plan for the turnover. The good MSPs help you plan for it. The mediocre ones do not.

How to Decide Whether You Need One

Three questions usually settle it.

First, can you, today, in less than an hour, produce an accurate report of every contractor working in your buildings, who they work for, what they cost, and when their contract ends? If yes, you do not need an MSP. If no, and the answer to that no is “we have tried and it is impossible,” you probably do.

Second, has compliance ever been a question someone outside your team had to answer for you? An auditor, a state agency, a parent company. If yes, the MSP starts to pay for itself in risk reduction alone.

Third, what is your spend? If it is under about $8 million annually in contingent labor, the math is hard. Above that, the math gets easier. Above $25 million, the math is usually a foregone conclusion.

If the answer is still ambiguous, the cheaper move is to consolidate to two or three trusted suppliers, run a quarterly business review with each of them, and stand up internal compliance controls. That replaces an MSP up to a real ceiling. We have seen companies get to roughly 200 contractors on that model before it broke. Past that, the spreadsheet stops working and the MSP conversation becomes serious.

When you want a second opinion before you sign anything, talk to a recruiter who has worked the supplier side of an MSP program. We are happy to walk you through the math, including the parts that argue against hiring us.

Common Questions

What is the difference between an MSP and a VMS?

Short answer: the VMS is the software, the MSP is the team that runs the software. You can technically buy one without the other, but very few enterprises do, and the ones that buy a VMS without an MSP usually end up hiring an MSP within two years anyway because nobody internally has time to run the program.

How much does an MSP cost?

1.5 to 3.5 percent of program spend is the standard range, billed either to you directly or to your suppliers as a margin haircut. Implementation fees are separate and usually run six figures for an enterprise rollout. The fee that looks small on a deck gets large when applied to a $20 million annual contingent budget. The supplier funded model is the popular pick because it looks free, but it is not free, the cost just shows up in different places on the P&L.

Do small companies need an MSP?

Usually no. Below about 150 active contractors the management fee tends to outrun the savings, and the speed and relationship costs are real. A small panel of trusted contract staffing partners and a quarterly review meeting will get most companies further than a half-built MSP program will.

How is an MSP different from an RPO?

Different talent pool, mostly. RPO covers permanent recruitment, the people who will be on your payroll. MSP covers contingent workers, the people who will be on a staffing supplier’s payroll. Same idea structurally, completely different mechanics, different vendor lists, different SLAs. The two often run alongside each other at the same company.

Will my favorite staffing supplier still work with me if I sign an MSP?

Probably. Most suppliers will sign the MSP agreement to stay on the panel, even when the math is painful, because the alternative is losing the account entirely. The relationship will change. The senior recruiter you have on speed dial will now have to push candidates through the VMS, and pay rates on new contracts will likely soften as the supplier absorbs the MSP fee. Over 18 months you will probably see candidate quality drift somewhat. The honest version is that the supplier ecosystem you have today is not the one you will have a year and a half after the program goes live, and a good MSP helps you plan for that transition instead of pretending it will not happen.

How long does an MSP implementation take?

Four to nine months for a real one. The discovery phase alone, where the MSP maps your current spend and supplier base and compliance posture, runs two to three months. Then the supplier agreements get redone, which is another two months because the legal review is ugly. Then the VMS gets configured. Anyone promising you a 90 day implementation is either rolling out a stripped-down program or planning to send you the rest of the bill later.

Can an MSP run alongside our existing direct hire process?

Yes, and that is actually the smart setup. The MSP owns the contingent layer. Your internal talent acquisition team, or your project staffing partners, owns the rest. The two stay in their lanes, the data flows separately, the compliance pictures are separate, and you still get the consolidated visibility on the contingent side. Trying to merge the two under one program is where companies make the implementation harder than it needs to be.

One More Thing Before You Sign

An MSP is a real tool that solves a real problem at the right scale. It is also an expensive tool that solves a different problem than the one most marketing decks describe. The pitch is “save money on contingent labor.” The reality is “buy visibility, compliance, and consolidation, and pay for it in fees and a slightly slower fill rate.”

If that trade is the right one for your company, sign the contract and run the program well. If you are not sure, the cheaper experiment is to consolidate suppliers first, stand up better internal reporting, and see how far that gets you before you commit to a four-month implementation and a 2.5 percent fee.

And if you want a sanity check from someone who has watched dozens of these programs go live from the supplier seat, that is a phone call we are happy to take. KORE1 wins reqs from MSP programs every week. We also lose money to the model. Both perspectives are honest, and you should hear both before you decide.