Last updated: September 12, 2026
By Mike Carter, Managing Director, KORE1
Direct sourcing fills contingent roles from a private talent pool a company already owns, usually alumni, retirees, and silver-medalist applicants, while a staffing agency recruits people the company has no relationship with. Neither one is a program by itself. The decision worth your time is which requisitions go down which channel, and most buyers get further splitting the list than picking a side.
The deck always has the same slide. Agency on the left, direct sourcing on the right, and a savings figure at the bottom of the right-hand column, somewhere between 15 and 40 percent. The left column is priced. The right column is a promise.
I’ve sat through a version of that slide maybe a dozen times in the last two years. The number is rarely wrong. It’s just answering a question nobody in the room asked, which is what your rate card would look like if somebody had already solved the sourcing problem for you.
Here’s what the slide leaves out. Staffing Industry Analysts has been tracking buyer participation in direct sourcing programs since 2016, and across that entire stretch it has moved between 30 and 36 percent. Flat, in SIA’s own description. Kersten Buck, SIA’s senior vice president of global strategic solutions, wrote it up for CWS 3.0 under a headline that does the summarizing for me: a proven concept still waiting for broad adoption.
Ten years is a long time for an idea this good to stay flat.
You should know where I sit. KORE1 is a staffing agency. We run contract staffing desks across technology, accounting and finance, and engineering, so one of the two columns on that slide pays our bills. I’m still going to argue that direct sourcing belongs in most large contingent programs, and I’ll be specific about the requisitions where it should beat us.
This is written for contingent talent buyers. Program managers, procurement leads, and the HR operations people who own a contingent population and keep getting asked what they plan to do about it. If your question is really about permanent headcount, whether to build an internal recruiting function instead of paying placement fees, that’s a different set of numbers and we covered in-house recruiting versus a staffing agency on its own.
What Direct Sourcing Actually Is
SIA’s 2025 Direct Sourcing Platform Landscape report, published that November, puts it this way: a company uses its own brand and its own private candidate pool to place people inside the organization as temporary workers. Former employees. Retirees. The silver medalist sitting in your ATS from a search that came down to two finalists. The pool is private and stays private. That’s the whole idea.
Three things have to exist for it to function, and the SIA report names all three.
Candidate engagement technology comes first. Not a job board and not your careers page. A system that keeps a conversation going with people you have nothing to offer today, so that the relationship is already warm on the morning a requisition opens.
Then curation, which is where programs quietly die. Somebody has to keep the pool from rotting. Skills go stale, numbers change, and the silver medalist you loved in 2023 has been promoted twice since anyone in your company last spoke to her. Curation is a job. Most programs fund it as a feature.
Third is a compliance and employer-of-record partner, and this is the component that surprises buyers who came for the savings number. You can’t direct-source your way out of the employment relationship. Someone still runs payroll, someone still owns classification, and someone still carries the liability when a contract worker’s status gets questioned. Direct sourcing changes who recruits. It doesn’t change the fact that a contract worker has an employer.
A Number the Slide Never Includes
Back to 30 to 36 percent.
That flat line tells you more than any savings claim, because it represents a decade of buyers with real budget and executive cover trying this and then not scaling it. Marketing didn’t fail. Buck’s diagnosis is that the problem is clarity: organizations struggle to define direct sourcing’s role, its operational requirements, and its relationship to the staffing partners already sitting in the program.
I’d put it less diplomatically. These programs stall because nobody ever wrote down which requisitions belong to them.
The pattern is consistent enough to predict. A pilot launches and it works, because pilots get scoped to the roles most likely to succeed, which means recurring high-volume seats where the alumni pool runs deep. Encouraged, the program reaches for scarce skills. The pool has nobody. The hiring manager waits three weeks, gets tired of waiting, and calls the agency that filled the identical seat last year without telling procurement. Nobody ever declares a failure. Adoption simply stops moving, and ten years of that is what a flat line looks like from the inside.
Where Each Channel Actually Wins
Split by requisition type, not by philosophy. This is roughly how the line falls on the programs I see.
| Requisition Type | Direct Sourcing | Staffing Agency | Why |
|---|---|---|---|
| Seats you refill four or more times a year | Strong | Redundant | Pool depth compounds with every cycle |
| Alumni, retirees, and silver medalists | Strong | No access | An agency cannot sell you your own former employees |
| Seasonal ramps with known scope | Strong | Situational | Forecastable demand is what pools are built for |
| Scarce or net-new skills | Weak | Strong | Your pool contains nobody who has done it |
| First hire in a new metro | Weak | Strong | Employer brand does not travel ahead of you |
| Confidential backfills | Weak | Strong | You cannot market a role you cannot name |
| Starts inside two weeks | Situational | Strong | Depends entirely on whether the pool is warm today |
Read the direct sourcing column honestly and a direct sourcing program covers maybe a third of a typical contingent requisition list. That isn’t a criticism. A third is a lot of money.
The Savings Number Is Real. So Is the Cost Stack Underneath It.
Agency markup is visible, negotiated, and printed on an invoice every week. That visibility is exactly why it becomes the first target in any cost exercise, and why it gets attacked ahead of larger costs that happen to be buried.
Direct sourcing relocates that spend rather than deleting it. The engagement platform carries a license. Curation carries headcount, or it carries a services fee paid to whoever does the curating for you. The employer-of-record partner takes a percentage, normally a smaller one than an agency charges, because they’re doing less: no sourcing, no screening, no market intelligence, just employment, payroll, and the risk that comes with both.
Net it out honestly and the savings hold up on requisitions where the pool actually produces somebody. On requisitions where it does not, you’ve paid the platform, paid the curator, and then called an agency anyway.
Our contract desks average 17 days to hire on IT roles, and 92 percent of the people we place are still in the seat at twelve months. Those same numbers are the argument against using us on the wrong requisition. If a seat comes open four times a year and your pool already holds four people who have sat in it, 17 days is slow and our markup buys you nothing you could not do yourself.
An industrial automation manufacturer we support in the Midwest ran this split last year without much ceremony. They built a pool around QA technicians and technical documentation contractors, the seats they refill constantly, and pulled a real majority of that volume out of the agency channel inside two quarters. Then a controls engineer opened up, PlantPAx experience required, migrating a line off legacy PLC-5. Their pool held eleven QA people and zero controls engineers. That one came to us. It took nineteen days, which was slower than their internal target and faster than the four weeks the pool had already spent not filling it.
Both of those outcomes were correct. That’s the part worth internalizing.
The Classification Question Got Louder in February
Direct sourcing puts your company closer to the worker. Closer is the point of the model, and closer is also the exposure.
The Department of Labor reopened all of this in February. Its notice of proposed rulemaking, filed February 26, 2026 under RIN 1235-AA46, would throw out the 2024 independent contractor rule and put an economic reality test in its place, built on federal judicial precedent and hanging on two questions. How much control does the worker have over the work? And what is their opportunity for profit or loss, based on their own initiative or investment? Comments closed April 28. A final rule should land before the year is out. Read the DOL’s own notice rather than anybody’s summary of it, this one included.
The IRS runs a separate common-law control test for the same worker. Two agencies, two frameworks, no obligation to reach the same answer.
None of that makes direct sourcing risky. It makes the employer-of-record component non-negotiable, and it turns the choice of that partner into a compliance decision that finance should not be allowed to win on price alone. A buyer who stands up a talent pool, engages people directly, and then papers them as independent contractors to skip the EOR fee has not built a direct sourcing program. They have built an audit, and they have built it in the year the rules are being rewritten.
How to Split the Requisition List
Five steps, in this order, before you sign anything.
- Count the repeats. Pull twelve months of contingent requisitions and count how many distinct titles appear four or more times. That set is your direct sourcing candidate list. In most programs it is also the entire list.
- Check that the pool exists before you buy the platform. Export your silver medalists from the last three years, filter to the titles from step one, and call twenty of them. If eight will take your call, you have a pool. If two will, you have a database, and no amount of engagement technology turns one into the other.
- Price all three components together. Platform license, curation headcount or services fee, and the EOR percentage. Compare that total against your blended agency markup on the step-one titles only. Comparing it against your whole contingent spend is how programs get approved and then quietly missed.
- Write the routing rule down and hand it to hiring managers. One page. Which titles go to the pool, which go to suppliers, and who to call when the pool comes back empty after ten business days. Buck’s point about giving managers a decision framework is the least glamorous item in her list and the one I would protect first.
- Tell your suppliers what you are doing. Out loud, with the title list attached. The ones adding real value on scarce skills will not care. The ones who have been quietly billing you for placements your own alumni could have filled are going to care a great deal, and finding out which is which is worth the awkward call.
If you already run a managed service provider or a vendor management system, direct sourcing lands inside that program rather than beside it, and the governance gets easier rather than harder. We wrote up how MSP and VMS models differ if that distinction is still fuzzy on your team. And if you’re changing suppliers in the same window, stagger them, because a staffing vendor transition and a direct sourcing launch running concurrently is two migrations competing for the same exhausted hiring managers.
One more piece of vocabulary hygiene, since procurement will raise it. Direct sourcing is a channel. Contract staffing, staff augmentation, and temporary staffing are engagement models, and they describe what you are buying rather than who found the person. We pulled those three apart in a separate comparison. A directly sourced worker is still on a contract staffing arrangement; only the recruiting changed.
Questions Buyers Ask Once the Pilot Gets Real
Can we run direct sourcing without an MSP?
Yes, and plenty of mid-market programs do exactly that. What you cannot skip is the curation work, which without an MSP lands on an internal talent acquisition team that is already at capacity. The programs I have seen succeed at that size picked one job family and stayed there a full year before expanding.
What size contingent population does this start to pay for itself at?
Somewhere north of 200 contingent workers, though the headcount matters less than the repeat rate. A 400-person program spread across 300 distinct titles is a worse candidate than a 150-person program that refills the same dozen seats all year. Concentration is the variable. Count titles, not people.
Won’t our staffing suppliers fight this?
Some will, and the ones who fight hardest are telling you something useful about their book. A supplier earning its margin on scarce, hard-to-find skills loses nothing when you self-fill recurring seats. A supplier whose volume is concentrated in roles your own alumni could cover has a real problem, and it is not your job to solve it for them.
How long before a pool actually produces hires?
Two quarters, assuming you start from an ATS with genuine history in it. The first ninety days go to data cleanup nobody warned you about: duplicate records, dead email addresses, people who left the industry, and a surprising number of candidates who were rejected for reasons nobody documented and who now have exactly the experience you need. That cleanup is unglamorous and it is most of the work. Programs that skip it launch a pool that looks full and answers nothing.
Does direct sourcing reduce our co-employment risk?
It increases it, in the sense that matters. You are engaging the worker directly, controlling more of the relationship, and removing an intermediary who used to absorb some of that exposure. The employer-of-record partner is what puts the buffer back. Choose accordingly.
We tried this two years ago and it stalled. What went wrong?
Scope, nine times out of ten. The program was pointed at the whole requisition list instead of the repeating slice of it, so the pool got measured against roles it was never going to fill and the fill rate looked terrible by month four. Occasionally the answer is worse than scope, and the honest version is that the employer brand could not carry a pool because the alumni did not leave happy. That one is not a sourcing problem and no platform fixes it.
Where I Would Start
Pull the twelve-month requisition list before you take another vendor call. If four or more titles repeat at volume, direct sourcing is worth a scoped pilot on those titles and nothing else. If they do not repeat, you have a sourcing problem that no private talent pool can solve, and the honest recommendation is to fix your supplier list instead.
Most of the programs that work end up running both channels for years, which is roughly what SIA’s Buck concludes as well: direct sourcing is an alternative to traditional suppliers on a defined slice of the work, not a replacement for them. The flat adoption line since 2016 is what happens when that gets read backwards.
If you want a second read on which of your requisitions are genuinely agency work and which ones your own alumni should be covering, talk to our contract staffing team. We place across IT staffing, engineering, and accounting and finance, and we will tell you plainly which of your seats we would hand back to a pool. Those conversations have never once cost us a client.
