When a Technical Contractor Isn’t Working Out: Replace, Coach, or Convert

Last updated: September 14, 2026

By Jennifer Burdick, Recruiting Manager, KORE1

When a technical contractor isn’t working out, spend a few days finding out whether the setup, the person, or the scope is failing, then coach, replace, or convert to match. A setup problem usually clears within two weeks of honest coaching. A person problem is fastest to fix through the staffing agency, often inside a week. A scope problem needs a different engagement, because no contractor would have succeeded in the seat as it was written.

The call usually comes in week three. A hiring manager tells me the contractor we placed just is not working out and wants to know how fast we can send someone else.

I ask for a few days first. Not because replacing people is hard for us. It is not. I will get to how quickly it can happen. I ask because a good share of these calls turn out not to be about the contractor at all.

A freight-tech company in Torrance called us in March about a contract data engineer on its Databricks platform. Three weeks in, the manager said, and not one merged pull request. He was right about that. What he did not know was that the contractor had filed an access request for the production Unity Catalog on his first morning and was still waiting on it eleven business days later, because the request sat in a queue owned by a platform team two time zones away. So he had built his first two pipelines in a personal sandbox, where nobody could see them. Good pipelines, too. At a $118 bill rate, the company had paid roughly $10,400 for work that existed and could not be merged.

The access came through on a Tuesday. His first pull request merged that Thursday. He stayed nine months.

Most of what gets written about a contractor who is not working out is about construction subcontractors, or about employees, and the advice follows the employee playbook. Document everything. Issue a warning. Build a performance improvement plan. Terminate if nothing improves. For a technical contractor placed through an agency, that sequence is wrong in two directions. The formal performance process is not yours to run, since the contractor’s legal employer is the agency. And the exit is already built into the arrangement, so the expensive mistake is rarely waiting too long to replace someone. More often it is replacing a person when the real problem was the seat.

I lead recruiting and delivery on our contract staffing searches, which gives me an interest in how this comes out either way. A replacement costs us a second search. A contractor you keep keeps billing. Neither answer is free of our incentives, and you should know that before you take my read on which one you need.

“Not Working Out” Is Three Different Problems

A technical contractor is not working out when their output, pace, or collaboration falls well short of what the statement of work assumed. The cause nearly always sits in the setup around the contractor, in the contractor themselves, or in the scope they were brought in to deliver. Each needs a different fix.

The symptoms overlap, which is why managers mix them up. Here is how they usually sort on the contract seats we fill for engineering, data, and infrastructure teams.

Row one is the most common by a distance. Row four is the rarest. It is also the urgent one.

Check the Setup Before You Judge the Person

Contractors walk in without the thing your permanent engineers spent years collecting, which is knowing who to ask.

That gap is bigger than it sounds. In the 2024 Stack Overflow Developer Survey, 61% of respondents said they spend more than 30 minutes a day searching for answers or solutions at work, and 30% said knowledge silos affect their productivity ten or more times a week. Those are mostly people who already know the codebase and the org chart. Veterans, in other words. A contractor in week two has neither. Every unanswered question costs them more than it costs your team, and most of them will not tell you, because a contractor who complains about the environment in week two sounds like a contractor making excuses.

Expectations are the other half. Then there is Gallup. Only 46% of U.S. employees clearly know what is expected of them at work, per its January 2025 engagement report, and the figure stood at 56% as recently as March 2020. Those are employees with managers, job descriptions, and annual reviews. A contractor gets a requisition and a kickoff call. Sometimes not the call. If your own people are split roughly down the middle on what good looks like, it is fair to ask whether the contractor ever heard it said out loud. The same gap often starts before anyone is hired, in a job description nobody questioned, which is the subject of Samantha Litman’s piece on questioning every line of a job description.

So before anyone decides anything, pull three things. The access requests the contractor filed and when each one was granted. The first three tickets they were given, with dates. And the review thread on whatever they have submitted so far.

Then ask the contractor, privately, what is in their way. Usually it is specific. A missing role in AWS IAM. A staging database nobody has refreshed since spring. A lead engineer who went on leave the day they started. If the answer is vague, or everything is supposedly fine, that tells you something too.

And ask the agency to call them. Contractors tell their recruiter things they will not tell a client, especially in the first month, because the recruiter is the one person in the arrangement whose job includes finding them the next assignment. We make that call the same day a client raises a concern. Every time. It rarely takes twenty minutes, and more often than you would expect the contractor has been waiting for somebody to ask.

Your onboarding checklist for contractors is the other place to look. If the access and hardware steps on it ran late, the first two weeks of output will show it.

Coach the Work, and Leave the Employment Side to the Agency

Coaching a contractor looks different from managing an employee. Not only in style.

You direct the work. You decide what gets built, set the priorities, review the output, and tell the contractor directly when something is not good enough. That is all yours. Do not water it down. What is not yours is the employment relationship itself, so formal warnings, performance improvement plans, pay decisions, and discipline belong to the staffing firm that employs the contractor. Keeping those lines clean protects you in co-employment arguments, and it also simply works better, because a written warning from a client to somebody else’s employee lands strangely and ends up forwarded to the agency anyway. If the contractor is on a C2C or 1099 arrangement instead of an agency W-2, those lines move, and how W-2, C2C, and 1099 contractors differ is worth reading before you coach anyone.

The reset we suggest fits in ten business days.

1. Write two or three sentences describing what finished work should exist by the end of next week, whether that is a merged feature, a pipeline running on schedule, or a closed set of tickets.
2. Clear every access or environment blocker the contractor names within 48 hours, and put a name next to each one.
3. Assign one engineer the contractor can interrupt without apology, and tell that engineer it is part of their week.
4. Tell the agency’s account manager what you have seen and what you asked for, so the contractor hears one consistent message from both sides.
5. Review the output together on day ten and decide whether to keep the contractor, replace them, or change the engagement.

Step three matters more than it looks. The named engineer will groan. Let them. In practice the questions start arriving in batches once there is a named person, which costs that engineer less than a steady drip of half-answered messages in a team channel.

Coaching has limits. I would rather say so plainly. Ten business days can fix access, priorities, a misunderstanding about what done means, and a lot of first-month nerves. It will not turn a mid-level React developer into the senior engineer the seat needed. It will not give someone five years of Kubernetes operations experience they did not have. If the day-ten review reads like the day-one review, the problem was never the setup.

Replace a Person Problem Quickly

Once you are sure it is the person, move. This is the case the contract model was built for.

An agency contractor is employed by the staffing firm, so ending the assignment does not put a termination process, severance, or an unemployment claim on your side of the table. Most agreements allow it on short notice. One caution on the reason you give. The EEOC’s guidance on contingent workers says a covered client can be liable as a joint employer when it turns away agency workers for discriminatory reasons, so keep the explanation about the work. KORE1 backs its placements with a replacement guarantee, and for a seat with a clear scope we can usually have new candidates in front of you within a few business days. Not every agency works that way, so read the replacement and termination language in your own agreement before you need it.

Fast is not free, though.

Matthew Bidwell at Wharton came at this from the permanent side. He had personnel records from the U.S. investment banking arm of a financial firm, 2003 through 2009, and he compared people hired from outside with people promoted from within. The outside hires were paid about 18% to 20% more. For their first two years on the job, they got significantly lower performance evaluations. Bankers, not engineers. The mechanism carries over anyway. Someone new to your systems starts without the context the last person spent weeks picking up, and a second contractor pays that toll all over again.

So make the second start cheaper than the first. Keep whatever the outgoing contractor figured out, even if their code is getting thrown away. The access list, the environment quirks, the three services nobody documented. Hand it to the replacement on day one.

Tell the agency exactly what was missing, too. “Not a fit” gets you another candidate with the same gap. “Strong on Python, but had never run Airflow in production and could not debug a failed DAG without help” gets you a different shortlist, and it is the single most useful sentence a client can give us on a replacement.

Some person problems have nothing to do with skill. A SaaS company in Sorrento Valley brought us in to replace a contract site reliability engineer placed by another firm. He had missed two pages on the EKS on-call rotation in three weeks, answered messages in a four-hour window each afternoon, and billed forty hours every week. When the client finally asked him directly, he admitted it. He was working a second full-time contract for another company. The assignment ended that Friday. Our engineer started the following Wednesday, and the handoff notes he received were the two pages the client wrote while it was still angry.

When the same staffing vendor sends you two contractors in a row who do not work out, stop treating it as two person problems. It is probably a screening problem, and a third submittal from the same desk will not solve it. Our page on switching staffing vendors mid-engagement walks through moving the contractors who are working to a new supplier on a dated plan.

The rarest case gets the fastest response. If the contractor in your standups does not sound like the person you interviewed, or cannot talk through work they supposedly did last week, get the hiring manager on a camera-on call and phone the agency the same day. Grant no additional access in the meantime. The identity checks that belong in the interview itself are part of our contract developer interview process.

Convert the Engagement When the Seat Was the Problem

Some contractors fail because nobody could have succeeded at the job as it was written. Replacing them just repeats the failure on a new invoice.

A credit union in Riverside brought on a contract Power BI developer last year to build executive dashboards for a new lending product, due in eight weeks. By week four there were no dashboards, and the VP told us she was not working out. She had spent those four weeks building the data model the dashboards were supposed to sit on, because it did not exist. Nobody had checked. Loan data lived in three systems that disagreed about what counted as an active account, and she had been quietly reconciling them in SQL Server. The requisition asked for a report developer. The work needed an analytics engineer. We rewrote the scope around the data model, adjusted the rate to fit the harder work, and a second, less expensive contractor picked up the dashboards six weeks later. Both engagements finished on the dates they promised.

Converting the engagement usually means one of three things.

The simplest is converting the level. A seat written as mid-level that turns out to need senior judgment gets re-papered at a senior rate, and sometimes the current contractor is the right person at that level and sometimes they are not. Converting the model is bigger. Staff augmentation assumes you will direct the work day to day. If the real need is a defined deliverable with someone else accountable for the plan, a project-based staffing engagement fits better than another contractor reporting to a manager who has no time to manage them.

Resist adding people. When a contract seat falls behind, the instinct is to keep the struggling contractor and bring in a second one to catch up. Fred Brooks named that trap in The Mythical Man-Month back in 1975, writing that adding manpower to a late software project makes it later. More than fifty years on, it holds for a two-person data team about as well as it did for IBM’s OS/360. Fix the scope first. Add people after.

The last conversion is the permanent one. Sometimes a contractor struggles as a contractor because the team has started treating the seat as permanent. They own a system now, they get paged, other teams route questions to them, and the end date means nothing to anyone. That seat needs an employee. If the contractor is good and wants a salary, convert them. KORE1’s conversion fee drops from 20% of first-year base salary to 15% after 800 billable hours, and to zero after 1,040. It also helps to know how often contractors actually convert across the market before you plan around it. If the contractor would rather keep contracting, start a direct hire search now and let them hold the seat until the permanent person arrives.

None of this is the same call as a contract reaching its end date with the work and the person both fine. That one is about timing, and Tom Kenaley’s guide to extending, converting, or releasing a contractor at the end date covers it.

The Bill for Waiting

Indecision has a bill rate.

A mid-to-senior contract engineer billing $110 an hour costs $4,400 a week at forty hours. Three weeks of “let’s see how it goes” without a diagnosis or a plan comes to $13,200, before you count the review time your own engineers spend on work that comes back, or the deadline the whole arrangement was supposed to protect. The ten-day reset above costs exactly the same per week. The difference is the ending.

Waiting costs the contractor too. They notice. Most contractors can tell when a client has gone quiet. The good ones start returning recruiter calls, and if the real problem was your staging environment, you can lose the person right as they were about to start delivering. Keeping the good ones is a separate discipline, and our guide to holding on to contractors you want to keep covers it.

Say something to the contractor, and say it soon. A short conversation that sounds like “The first two weeks have not gone the way either of us wanted. Here is what I need to see by the 24th, and here is what I am fixing on my side” does more than a month of polite silence. Most experienced contractors have heard some version of it before. The professional ones respect it.

What Managers Ask When a Contract Starts Badly

Can we end a contractor’s assignment this week?

In most agency agreements you can, with whatever notice the agreement specifies, which for contract assignments is often a few business days rather than weeks.

Check the termination clause first, because some project-based agreements carry minimum hours. Then call the agency before you tell the contractor. The agency is the employer. It should deliver the news, and it needs the lead time to start on a replacement if you want one.

Do we write a performance improvement plan for a contractor?

No, because a formal PIP belongs to the contractor’s employer of record, which is the staffing agency and not your company.

Give direct, specific feedback on the work, which is entirely yours to give. Route warnings, formal plans, and any documentation of conduct through the agency’s account manager. A written plan from a client tends to blur the employment line you are paying the agency to hold.

How long should a struggling contractor get before we decide?

Ten business days after the setup problems are fixed is enough to see whether the work changes.

Start the clock when access and priorities are actually sorted, not on the day you first got frustrated. If the review comments on day ten look like the ones from day one, you have your answer. For the availability and identity problems in the table above, do not wait ten days at all.

Who pays for the hours a contractor worked before things fell apart?

Usually you do, since hours worked on an agency assignment are billable whether or not the work turned out to be usable.

Some agencies offer credits or a no-cost replacement period under their guarantee, and the terms vary a lot from firm to firm. Read yours before the invoice arrives rather than after. Hours spent waiting on access are billable too, which is one more reason to diagnose fast.

Is it the contractor or our onboarding, and how can we tell?

Look at the access tickets before you look at the code.

If the contractor was blocked for more than a few days on permissions, environments, or an unanswered question, the first two weeks of output say more about your setup than about them. A contractor who had everything by day two and still has nothing merged by day ten is a different story. That is a person problem.

The contractor is strong, just not in this role. Can we move them?

Moving a good contractor to a better-matched seat is usually faster than starting a new search, as long as the agency re-papers the assignment with the new scope and rate.

We did this for a manufacturer in Chino last winter, moving a contract .NET developer off a customer portal rebuild she was struggling with and onto an MES integration for the plant floor that another team had been trying to staff for two months. She was far more useful there, and she stayed through go-live. Have the manager of the new seat interview her for it, even briefly, so the move is a real decision and not a quiet reassignment.

Make the Call, Then Say It Out Loud

Find the cause before you pick the fix. Check the access and the first tickets, ask the contractor, and have the agency ask too. Then coach for ten business days if the setup was the problem, replace quickly if it was the person, and change the engagement if it was the seat. Whichever it turns out to be, tell the contractor what you decided and why.

KORE1 has filled contract and permanent technology seats since 2005, and our twelve-month retention sits at 92%. Some of that is screening. A lot of it is catching the rough starts early and fixing them before anyone gives up on the engagement. If you have a contractor three weeks in and struggling, walk through the situation with our contract team before you decide. Sometimes the right answer is a different contractor. Often it is a staging database.