How to Hire an ERP Project Manager: 2026 Guide

Last updated: August 11, 2026

By Robert Ardell, Co-Founder and Strategic Advisor, KORE1

Hire an ERP project manager by testing whether they can price a change order, not by counting go-lives on a resume. Base pay runs $105,000 to $190,000 in 2026 depending on scope and industry, contract rates sit between $95 and $165 an hour, and a properly scoped search closes in three to six weeks.

Resumes are bad at this. Go-lives are easy to list and almost impossible to verify. Everybody near one claims it.

So ask for something else. Ask for the log. Any candidate should be able to hand you a change log from a program they ran, redacted however they need it to be, and the last column is the one worth reading. The good ones are boring in a very specific way. Every entry carries a classification, a date, a named requester, and a written reason for the classification, which means somebody had to decide on the record whether this particular change touched scope, or cost, or the validated state, and then defend that call to a person who could overrule them. Not just what changed. Why it was the kind of change that did or did not require a signature.

That column is the job. Everything else on the resume is context.

The Job Is the Change Log

An ERP project manager owns scope, schedule, budget, and change control on the buyer’s side of a software implementation. They run the requirements gate, classify and price every change after baseline, hold the vendor to a statement of work, and decide what gets escalated to an executive sponsor. They are not the person who configures the ledger.

Worth naming who is talking before you weigh any of this. KORE1 runs an ERP implementation project manager staffing desk and we bill when somebody signs an offer, so read the recommendations with that in mind. Two of the sections below argue against hiring at all.

The distinction that matters is baseline. Before a specification is baselined, changing it is a conversation. After, it is a governed event with a paper trail. Most ERP programs never draw that line clearly, and the ones that do tend to be in regulated industries, because an auditor eventually asks.

Most change logs record an outcome. A decision got made, a box got ticked, work carried on. The useful ones record the reasoning next to the outcome, because a classification nobody can defend twelve months later is not governance. It is filing.

So read down that reasoning column and ask two things. Who decided? And were they the person who then went off and did the work?

Those should be different people. Always. On serious programs the person assessing whether a change disturbs something already signed off is deliberately not the person implementing it, and on the strict ones those roles get assigned by name before any work starts, written into the plan, auditable afterward. It sounds like bureaucracy. It is the control that stops a team from quietly grading its own homework, and a PM who has worked inside one will tell you about it without being asked.

Your Partner’s PM Is Not Your PM

Nearly every implementation partner includes a project manager in the statement of work. That person is real, competent, and paid by the firm whose scope you are trying to hold.

Nobody is lying. The conflict is ordinary, which is exactly what makes it hard to see. Your requirements document slips three weeks. The build team sits there. Now somebody has to call it, cost event or schedule event, and the partner’s PM has a real professional opinion about which one it is. That opinion also protects their margin. Both things are true at the same time and from your side of the table they look identical, which is the whole problem, and no amount of goodwill fixes it because the person making the call is the person carrying the exposure.

A client-side PM exists to make that call for you. That is it. That is the entire justification for the seat.

The same conflict shows up in how the money is structured. Good engagements do not price a whole program firm on day one, because nobody knows enough on day one to do that honestly. They price discovery firm, carry the build as a range, and convert that range to a fixed fee once the requirements are approved.

Which puts a real decision point in front of you. Somebody has to defend it. If the estimate is going to land above the range, you need to hear so before you commit to the build phase rather than after, and the person whose job it is to make that conversation happen on time is the one you are hiring. It will not be the partner’s PM. Nobody wins that argument in month seven without having set it up in month one.

Which One You Need Depends on the Week You Are In

ERP project management is not one skill applied evenly across a program. The person who is excellent at vendor selection is frequently mediocre at cutover. The cutover specialist would be miserable in twelve weeks of requirements workshops, and would say so out loud by week three. Neither of those is a character flaw. It is evidence that one title is covering four different jobs that happen to share a reporting line, and most companies keep hiring against the title anyway.

Those durations are not padding. On a mid-market NetSuite integration we have benchmarks for, discovery and risk assessment ran four to five weeks, design and build ran ten to twelve, validation ran eight to ten, and release with three weeks of hypercare took another four. Roughly twenty-two to twenty-six weeks from the day requirements were approved. The clock did not start until that approval landed.

Which is why the requirements gate is the only date on the plan that really matters. Everything downstream floats against it.

Pay Bands for 2026

Compensation for this title is unusually noisy, and the noise itself is useful information. Three sources, three answers. Glassdoor has the U.S. average at $128,312, typical range $101,518 to $163,855. ZipRecruiter reports $134,565, with the middle of the market between $108,000 and $156,000 and the ninetieth percentile up at $188,500. Then Salary.com, which comes in far higher at $158,406 and runs its range out to roughly $211,582.

A $30,000 spread between three reputable aggregators for the same title is not a data quality problem. It is the market telling you the title is doing too much work. Which job are you buying?

For wider context, the Bureau of Labor Statistics put the median wage for project management specialists at $100,750 in May 2024, with 6% growth projected through 2034 and about 78,200 openings a year. ERP sits well above that median because the failure cost is higher and the pool is thinner. If you want to sanity-check a band against your own market, our salary benchmark assistant is free and takes about a minute.

One rate benchmark from the consulting side, since it reframes what you are actually comparing. Delivery hours on a specialist NetSuite integration price around $215 an hour blended across roles. Validation specialists on the same engagement bill $275. Do that comparison. A contract PM at $150 an hour is not expensive relative to what they are governing.

Hand Them a Quote and a Red Pen

This is the interview. Everything else is warm-up.

Take a real vendor quote, redact the vendor, print it, and give it to the candidate with fifteen minutes and a pen. Ask them what is missing. Then stop talking.

What separates the strong ones is where they go first. They go to the exclusions. Not the price, which is the number a weak candidate fixates on before offering a view about whether the total seems reasonable, and that is a judgment nobody in the room can evaluate anyway. The strong ones read what the vendor carefully carved out. Then they ask who does that work instead.

Some specific things worth watching for.

• Do they find the testing line? On a serious build, validation and testing runs around 30% of total hours. That is normal for bespoke software under a formal lifecycle. Eight percent is not. A quote that light is not a leaner team. It is a smaller promise. The testing still happens, later, on your clock.
• The good ones ask what happens to the range. Fixed-fee-on-approval structures are common and correct, but a candidate who has run a few will ask when the not-to-exceed becomes firm, and who tells whom if the estimate is going to breach it.
• Data migration. Almost always excluded or thinly scoped, almost always the thing that slips cutover.
• Names, not roles. A quote that assigns the work to “a senior consultant” can be staffed with anybody who is free that month. Ask for names. Strong candidates ask which named individuals are committed, what happens if one of them leaves in month four, and whether the deliverables live with the firm or inside that person’s head. It is not paranoia. One specialist carrying a workstream is a single point of failure, and the fix is unglamorous and documented, a named alternate and work product held centrally so a replacement inherits something.
• One candidate we placed last year found the assumption buried on page fourteen that the client would return document reviews inside ten business days, and asked the hiring manager whether their quality team had ever managed that in the history of the company. They had not. He priced the slip on the spot.
• Post-go-live. If nothing in the quote covers the twice-yearly platform releases most ERP vendors ship, the validated state quietly lapses, and re-establishing it later costs more than maintaining it would have. Budget for it. On the integration we benchmark against, that maintenance ran about 120 hours a year.

Fifteen minutes with a marked-up quote tells you more than an hour of behavioral questions. Same exercise. I have watched it disqualify a candidate with four go-lives on the resume, and I have watched it promote somebody with two.

Two Questions That Sort Real Program Managers

Ask both cold, late in the conversation, once they have stopped performing.

On your last program, who was allowed to sign off that your own team’s work had passed?

Weak answers put the PM in the middle of it. They tested, they reviewed, they confirmed, everybody moved on. What you want is somebody describing a boundary they did not control. The best version of this I have heard came from a candidate who said flatly that he could not close a finding written against his own build, that his client’s quality group held that pen, and that he had spent a genuinely miserable fortnight the previous autumn waiting on exactly that. He was not complaining about it. He was describing a control working. That is the answer.

The failure mode here is not fraud. It is gentler and much more common. A team that tests its own work writes tests that exercise the paths the builder already had in mind, so the exception nobody imagined stays unimagined until a user finds it in week two of go-live. Every time. Ask how they scoped the negative testing. Ask who wrote it.

Then the second one.

When something broke badly at nine at night, who did you call, and was it a person or an inbox?

An escalation path that ends in a shared mailbox is not an escalation path. Real programs name individuals on both sides with a response window attached, and they let the buyer set the severity rather than the vendor. That last detail is worth listening for specifically. If your candidate shrugs and says the partner decided what counted as a severity one, they have never had to hold that line, because every vendor alive would rather grade the urgency of their own outage.

There is a third if the first two land well. What is the current agreed scope on a program that has run eight approved change orders? The answer is the original scope plus all eight, restated in the most recent one so nobody has to reconstruct it from a folder. If they say the original statement of work, they have never watched scope drift happen. Scope drift is quiet. It arrives in eight small changes that were each individually reasonable, and it compounds.

What Changed in 2026

Something genuinely new belongs on the req this year, and most ERP job descriptions have not caught up to it.

Implementation partners are using AI code generation. Openly in some cases, quietly in others. That is not automatically a problem, and I would be skeptical of a PM who reacted to it with pure alarm, because the tools are useful and the better firms are getting real value out of them. It becomes a problem when nobody has written down the controls. Usually nobody has.

What you are hiring for is not an opinion about the technology. It is somebody who will ask a partner for the written policy during selection, while you still have room to negotiate, instead of during an audit, when you have none. The specific thing to listen for is whether they ask who is named as accountable for a generated deliverable, and whether that named person can explain it without opening the tool that produced it.

Then the follow-up almost nobody asks. Does that policy flow down to subcontractors?

Firms subcontract specialist work constantly, and validation work more than most. Obligations a partner accepts in your contract are worth very little if they stop at that partner’s own payroll, so every commitment should be flowed down in writing before a subcontractor starts, and you should know the subcontractor exists at all. Silent subcontracting inside a program you are personally accountable for is a bad surprise to get late. Ask for the list.

Ask anyway. You will learn quickly whether your candidate has thought about any of this or is hearing it for the first time, and in 2026 that is a legitimate screen.

Cheaper Things That Sometimes Work Better

Three situations where I would tell you to keep your money, and I have made this call on live searches.

You have not picked a platform yet. Do not hire yet. A PM brought in during evaluation spends eleven weeks in meetings they cannot influence, gets bored, and takes another offer around the time you actually need them. Run the evaluation with an advisor or a short consulting engagement, then hire the PM against a decision. Our free ERP readiness assessment covers most of what a selection-phase PM would tell you in the first month.

Second, you have a strong operations or finance manager who already owns the process being automated. Promote them and buy them a fractional PM for two days a week. That works. Domain knowledge is harder to hire than project discipline. It is the half that is genuinely expensive to replace, and a manager who has run the month-end close for six years already knows where the bodies are buried in a way no external hire will learn before go-live.

Third, your program is under $250,000 total and lives inside one department. You need a competent implementation lead, not a program manager. Hiring up here does not buy you insurance. It buys you somebody who will be underemployed by week six.

Those recommendations cost our desk revenue every time somebody takes one. They stay because we would rather place one right hire than two wrong ones, and clients who get told no on the first call tend to come back.

From Req to Signed Offer

Assuming the hire is real, here is the sequence that keeps a search inside six weeks.

1. Name the phase before you write the req. Selection, build, cutover, or steady state. Pick one. Reqs that say “full lifecycle” attract generalists and repel the specialists you actually want.
2. Decide platform depth honestly. Does this person need to have run your exact ERP, the same tier, or just enterprise software? Each answer changes the pool by an order of magnitude, and only one of them is usually true.
3. Set the band against the failure cost. If a slipped cutover costs you a quarter of clean revenue recognition, arguing over $15,000 of base is not thrift. It is math.
4. Build the quote exercise before the first interview. Redact a real one. Generic case studies do not work because candidates have seen them.
5. Interview in two rounds, not four. Good ERP PMs are usually working, frequently on contract, and they are gone in ten days. Our average time-to-hire across technical roles is 17 days, and most searches that blow past it die in scheduling rather than in sourcing.
6. Ask the reference the specific question. Not “were they good.” Ask whether the program’s scope at the end matched the scope in the last change order. Listen to the pause.

Whether you run this as contract staffing or a direct hire depends mostly on the phase from step one. Build and cutover work is naturally contract shaped. Steady-state ownership is not.

Before You Open the Req

Does the PM need to have run our exact ERP before?

Usually the platform family matters and the specific product does not. A PM who has run two NetSuite implementations will be effective on Acumatica within a few weeks, because what transfers is the governance and the sequencing rather than the screens. The exception is deep vertical configuration, where an SAP S/4HANA brownfield conversion genuinely does demand somebody who has done that specific thing, because the failure modes are peculiar to it. Put it to them directly. Ask which parts of your platform they have never touched. A candidate who says none of it is either exceptional or not listening.

The partner’s SOW already includes a project manager. What am I buying twice?

You are not buying the same thing twice. Different jobs. The partner’s PM manages the partner’s delivery obligations, and your PM manages your interests against that contract, including every change order the partner raises. The two roles conflict by design. On smaller programs a fractional client-side seat covers it, and on anything with multiple vendors or an integration portfolio it stops being optional. We have seen mid-market companies save more in disputed change orders in one quarter than the PM cost for the year.

PMP, PRINCE2, or does the certification not matter?

Certifications are a weak positive signal. Not a filter. A PMP tells you somebody learned a vocabulary and sat an exam, which is not nothing, and it is also not evidence they have ever held a vendor to a statement of work. I would take a candidate with no certification and a real change log over a certified candidate without one, every time. Where certification genuinely helps is regulated environments and government contracts, where the client or the auditor sometimes requires it outright. Check whether yours does before you make it a screen.

How fast can somebody take over a program that is already in flight?

Two to three weeks to be useful, six to eight weeks to be fully in control, assuming the documentation exists. That last clause carries all the risk. Assume it does not. Inherit a program with a real specification, a traceability matrix, and a maintained change log and you have a handover. Inherit one where none of that got written down and your new PM is doing forensics before they can start the actual job, which realistically means four to six weeks before anybody can tell you something reliable about your own schedule. That is not their fault. Ask what they would want in week one.

We are in a regulated industry. How much does that shrink the pool?

Considerably, and it is the single biggest constraint on an ERP PM search in life sciences, medical devices, aerospace, or food. The pool is small. Validation under a formal lifecycle changes the shape of the whole program. Requirements have to be individually identifiable and testable. Criticality gets assessed per requirement, so testing depth follows risk instead of defaulting to uniform rigor across the board, which is where a lot of budget quietly goes. And a performance qualification needs a defined process to qualify against, so if the process itself is still moving you end up writing a protocol nobody can execute. On one regulated build we have benchmarks for, validation ran roughly 30% of total hours. A PM who has not lived through that will under-plan it and will not know they are under-planning it until a protocol comes back unexecutable.

What does it cost us to leave the seat empty another quarter?

More than the salary, and the mechanism is not the one people expect. Panorama Consulting’s 2026 ERP Report, released in March, found more than a quarter of organizations exceeded their project budgets, with additional technology needs the leading cause. Read that mechanism carefully. Teams discover a gap late, buy something to close it, and the purchase was never in the plan. That is a governance failure showing up as a line item. You will also see a statistic quoted all over the internet claiming 70% of ERP implementations fail. It almost never comes with a traceable source. Ignore it. The budget number is real and it is the one that should worry you.

The Hire That Pays for Itself in One Change Order

ERP programs rarely fail in a single dramatic event. There is no explosion. They fail in small approved increments, each of which made sense to somebody on the day, none of which were ever assessed together.

The person you are hiring is the one who insists on assessing them together. That is the whole hire. It is an unglamorous job, and it is why the interview should be about a marked-up quote rather than a leadership philosophy.

KORE1 has been placing technical and program talent since 2005, and our ERP desk screens against change-control judgment rather than platform checklists. If you want to pressure-test a req before it goes live, or you are not sure yet whether the hire is real, talk to a recruiter and we will tell you honestly which of the three situations above you are in. If you are earlier than that and still scoping the program itself, our ERP recruiters and NetSuite implementation consultant staffing teams cover the surrounding roles, and our breakdown of the seven most expensive implementation mistakes is worth twenty minutes before you sign anything.