Staffing Vendor Consolidation: How Many Suppliers Do You Actually Need?

Last updated: September 12, 2026

By Jennifer Burdick, Recruiting Manager, KORE1

Most mid-market contingent programs run well on three to six staffing suppliers per job family, not three to six across the whole company. The number is set by how much requisition flow each supplier gets, not by headcount or total spend.

Fourteen years of recruiting, thirteen of them at KORE1, where I run delivery across tech, engineering, finance, HR, and operations. Consolidation puts me on both sides of the table. We survive some of these and we get cut from others, and I have sat in the debrief for both outcomes more times than I would like.

A medical device manufacturer in Orange County went from eleven staffing suppliers to three in the spring of 2025. Good program, run by a competent sourcing lead, with a real business case behind it. Average time to fill went from 24 days to 41 over the following two quarters. Nothing was wrong with the three firms they kept. They had simply picked three that all sold the same thing, and the company had four job families, one of which nobody kept a supplier for. Nobody noticed for seven months.

Almost everything published on this subject treats consolidation as an arithmetic problem. Count your vendors, cut the bottom half, book the savings. The academic literature on supply base reduction has said something more uncomfortable than that for twenty years, and almost none of it has made the trip into staffing procurement.

Where I am sitting matters here, so take it into account. We sell into these programs. KORE1 has held contract staffing panel seats since 2005, we are more often a firm that gets kept than one that gets cut, and a consolidation that goes our way is good for us commercially. We have also been the incumbent who lost a seat we should have held. Both of those shaped what follows.

Consolidation Has Three Dials. Most Programs Only Turn One.

The most useful research on this was not written about staffing at all. Thomas Choi and Daniel Krause published a paper in the Journal of Operations Management in 2006 that has quietly framed how supply chain academics think about supplier count ever since. They defined the supply base as the part of a supply network a company actively manages, then argued its complexity runs on three dimensions rather than one.

Number of suppliers. Degree of differentiation among them. Level of interrelationship between them.

Read those three again with a staffing panel in mind. The Orange County program cut the first dial by eight and never touched the second. Three suppliers, all generalist IT and finance shops selling from overlapping candidate pools in the same metro. On paper the supply base got 73 percent simpler. In practice it got narrower without getting more capable, because differentiation is what determines whether your panel can actually cover the work, and nobody had measured it.

Choi and Krause were careful about what reduction does. Their conclusion, in their words, is that “although a reduction in complexity may lead to lower transaction costs and increased supplier responsiveness, in certain circumstances it may also increase supply risk and reduce supplier innovation.” Two of the four outcomes improve. Two get worse. Every consolidation business case I have been shown in thirteen years models the first two and treats the second two as someone else’s problem.

That is not an argument against consolidating. We consolidate our own vendor base. It is an argument that the number by itself tells you almost nothing, which is inconvenient, because the number is the only part that fits in a slide.

Divide the Requisitions, Not the Spend

Here is the mechanic that actually sets your supplier count, and it is not one procurement teams typically model.

A staffing firm allocates its best recruiters to accounts that generate steady work. Not large accounts. Steady ones. A supplier receiving eleven requisitions a year from you cannot justify a named recruiter on your account, so your reqs get worked by whoever is free, which is a different person every time, none of whom knows your stack or your interview loop. A supplier receiving three or four a month has somebody who does know, and that person has candidates warm before your req opens.

The threshold where this flips, in our experience across roughly 30 metros, sits near one requisition per supplier per month within a given job family. Below it you are a transactional account no matter what the agreement says. Above it you are a relationship, and relationships are where the 92 percent twelve-month retention rate we run actually comes from.

So the calculation runs off volume, not spend. Backward from the work.

Take your annual requisition count in one job family. Divide by twelve. That is roughly how many suppliers that family can support before you dilute yourself into irrelevance with all of them. Forty software engineering reqs a year is three suppliers, comfortably. Twelve accounting reqs a year is one, possibly two if you want a backup, and running four is how you end up as everyone’s fourth priority in a market where they each have a better account down the street.

Spend leads you somewhere different and usually wrong, because a handful of expensive senior contractors can make a job family look well funded while generating almost no requisition flow.

Job family, not company. That distinction does the work. A business running 60 engineering reqs and 9 accounting reqs is a four-supplier company and a one-supplier company at the same time, and rolling those into “we use five vendors” is how the accounting roles quietly stop getting filled.

Two of the Four Things You Are Buying Get Worse

Transaction costs drop when you consolidate. That part is not in dispute, and it is usually understated in the business case rather than overstated. Fewer agreements, fewer rate schedules, fewer insurance certificates to chase, fewer invoice formats, fewer onboarding variations, one point of escalation. If your coordinator currently reconciles nine timesheet portals, cutting to three gives that person most of a day back every week, and nobody ever puts that in the savings model because it does not show up as a rate reduction.

Responsiveness improves too, for the reason in the section above. Concentrated volume buys attention.

Then the other two.

Supply risk concentrates in exactly the way the word suggests. Three suppliers means a single firm losing its lead recruiter in your vertical is a third of your coverage having a bad quarter. We watched a client lose most of their data engineering pipeline for five weeks because one recruiter at one of their two approved suppliers went on leave, and there was no third firm with enough warm volume to absorb it. The risk was invisible until the week it was not.

Supplier innovation is the one nobody believes until it bites. A specialist firm that works one narrow market brings you things a generalist structurally cannot, including candidates who are not looking, compensation intelligence from searches you never ran, and an early read when a stack is about to get expensive. Cut to a panel of generalists and that flow stops. It does not stop loudly. Your reqs still get filled, mostly, and two years later nobody can say precisely when you stopped hearing about people before they hit the market.

The fix is not keeping more suppliers. It is keeping differentiated ones, which is the second dial, and it costs nothing extra to turn.

What the Program Actually Looks Like, Quarter by Quarter

Consolidation fails on sequencing far more often than on selection. The order below is the one that works, and the expensive mistakes all come from compressing it.

Quarter one is measurement, and it is the quarter everyone tries to skip. Pull twenty-four months of requisition history and sort it by job family, not by vendor. You are looking for four numbers per family: reqs opened, reqs filled, median days to fill, and which supplier actually filled them. Most programs discover at this point that 70 percent of their spend sits with suppliers who filled under a third of the work, and that two of their “strategic” vendors have not filled anything in nine months.

Map differentiation while you are in there. For each supplier write one line on what they genuinely cover. If two lines read the same, they are one supplier for planning purposes regardless of what the spend report says. This is also the point to run whatever scoring model you use on the incumbents, which is its own exercise and deserves more room than I am giving it here.

Quarter two is the panel design and the paper. Set the count per job family from requisition volume. Then check that the survivors cover the families, because a panel that is correctly sized and incorrectly composed is the Orange County outcome. Build the rate schedule at the same time. A short vendor list with one blended rate card is worse than a long list with a detailed one, and both decisions get made in the same meeting by people who have not noticed they are two decisions.

Quarter three is transition, and the contractors are the whole game. Everything above is planning. This is the part with people in it. If you are cutting a supplier who has active contractors on site, those workers are employed by the firm you are terminating, and how you handle the transfer determines whether you keep them. We wrote up the full cutover sequence in our guide to replacing a staffing vendor without losing the contract team, and the short version is that it is an eight-step process, it starts with reading the existing agreement rather than sending a notice letter, and pay rates do not have to move.

Quarter four is the review nobody schedules. Same four numbers as quarter one. If median time to fill went up in any family, you took that family too far down or you composed it wrong. Fixing it in month ten is ordinary. Discovering it in year two is not.

One note on sequencing that costs real money. Do not run the consolidation and an MSP or VMS implementation in the same six months. An MSP is a mechanism for administering a supplier panel, not a method for choosing one, and companies that start both together end up letting the MSP inherit a panel nobody measured. Decide the panel. Then decide who administers it.

The Tail Is Not the Problem You Think It Is

Every consolidation starts by looking at the tail, that long list of suppliers billing under $50,000 a year, and the instinct is to delete all of it. Mostly correct. Not entirely. The exception is expensive.

Sort the tail by what it filled rather than what it cost. A firm that billed $34,000 last year and filled the two hardest roles you had is not tail spend. That is a specialist you have been underusing, and it is the cheapest differentiation available to you. A firm that billed $180,000 across nine easy backfills is the one to look at hard, and it will not be in the tail at all.

The American Staffing Association counts around 27,000 staffing and recruiting companies in this country, running roughly 54,000 offices between them, with about 57 percent of those firms working the temporary and contract side. Most are small and narrow. That fragmentation is why your vendor list reached fourteen without anybody ever deciding it should, and it is also why a well composed panel of four can cover more ground than a sprawling list of fourteen generalists.

Worth knowing what the market is doing while you plan this. Staffing sales ran $27.6 billion in the first quarter of 2026, with industry employment down 4.6 percent year over year, per ASA’s quarterly survey. A softer market is the easiest time to renegotiate a panel and the hardest time to judge a supplier, since fill rates improve for everyone when candidates are available. Judge on the last two years, not the last two quarters.

Questions That Come Up Once Procurement Owns the Number

Leadership handed us a target of three vendors. Is that defensible?

Three is defensible for one job family and rarely for a whole company. Ask which families the three are meant to cover, and if the answer is all of them, the target was set on spend rather than on work.

The conversation that usually resolves it takes ten minutes. Show the requisition count by family and let people see that engineering and accounting are different purchases with different suppliers behind them. Nobody is attached to the number three. They are attached to not managing fourteen relationships, which is a reasonable thing to want and is achievable at seven.

What actually shows up as savings?

Rate improvement from volume concentration is real but modest, typically low single digits. The larger number is administrative, and it lands in your team’s hours rather than in the rate card, which makes it harder to claim.

Be careful about the second-order effects. If time to fill lengthens by two weeks across forty roles, the cost of those vacancies will exceed anything you won on markup, and vacancy cost almost never appears in the same model as the savings. Ask for both numbers on one page.

How long should we sign the survivors for?

Twenty-four months with a twelve-month performance checkpoint. Twelve months alone is too short to earn the dedicated recruiter you are consolidating in order to get.

Write the checkpoint as a review with defined measures rather than as a termination right, because the second one changes how the supplier staffs your account from day one. A firm that expects to be there for two years puts different people on you.

Do we tell the suppliers we are cutting before or after we choose?

After you decide and before you sign anything, with at least sixty days of runway if they have contractors on site. Surprising a firm that employs fifteen of your workers creates a retention problem you will own.

Most incumbents behave well here. They have seen it before, they would rather transition cleanly than litigate a conversion fee, and a few will tell you useful things about your own program on the way out. Ask them what they would fix. Some of the sharpest feedback I have given clients came in that conversation.

The firm we are cutting employs fifteen of our contractors. Do we lose them?

They transfer, in almost every case, and their pay rate does not have to change. The mechanism is a transition agreement between the outgoing supplier, the incoming one, and you, executed before anybody notifies the workers.

What breaks this is doing it in the wrong order. If contractors hear about the change from a rumor rather than a scheduled conversation, some of them start looking, and you lose people you were trying to keep. Read the existing agreement first, since conversion and non-solicitation clauses set your options and are frequently softer than anyone assumes.

Does any of this change if we are hiring contract rather than permanent?

The volume math is the same. What changes is urgency, since contract requisitions run on shorter fuses and a thin panel shows up as a missed start date rather than as a long search.

If most of your volume is contract rather than direct hire, weight differentiation higher than you otherwise would and keep one more supplier per family than the table suggests. The models themselves get confused constantly, which we sorted out separately in our piece on contract staffing versus staff augmentation versus temporary staffing.

Count the Requisitions Before You Count the Vendors

The question in the title has an answer, and it is unsatisfying. Three to six per job family for most mid-market programs, one or two for thin families, tiered above 150 reqs a year. But the number falls out of the requisition math rather than driving it, and a panel sized correctly and composed badly performs worse than the sprawl it replaced.

Two dials, not one. Count and differentiation. The first is easy to move and easy to measure, which is why programs move it alone and then spend a year wondering where the time to fill went.

If you want a second read on your current panel, talk to our team and we will run the requisition math with you. We have been a supplier on both sides of this since 2005, across IT staffing, engineering, finance, and operations, and we will tell you when your list is already the right length.