Who should own your ERP implementation? The owner, IT, or an outside consultant

When an SME picks someone to own its ERP project, it almost always lands on one of three choices: hand it to IT, hand it to an outside consultant, or the owner carries it personally. Each one fails in its own way, and the cause is never skill. It is the power structure. This post lays out which month each setup breaks down and where it gets stuck, gives a role setup that actually works, and explains the one thing the owner can never delegate.

Johnny Yang13 min read
Who should own your ERP implementation? The owner, IT, or an outside consultant

"I'm giving this project to Xiao Zhang in IT. He knows computers better than anyone."

I have heard this line in first meetings more times than I can count. The owner usually says it with total confidence, because in his mind ERP is software, and software goes to the person who understands software. Six months later I come back, and Xiao Zhang always has the same look on his face. He is holding a to-do list that never gets shorter, and not one item on it is blocked by him: sales still hasn't delivered the product categories it promised, the warehouse says stocktaking has to wait for the slow season, and accounting insists her Excel reconciliation sheet cannot be retired. Xiao Zhang did nothing wrong. The project just won't move.

Let me say one thing in the owner's defense first: treating ERP as an IT matter was the right instinct twenty years ago. Back then, implementing a system really was closer to "installing something." But today you are not buying software. You are rewriting how several departments work: how part numbers are coded, when an order counts as confirmed, whose inventory number is the real one, which people are no longer allowed to use their own Excel files. Not one of these is a technical question. Every one of them is a question of who has to give ground. If you put someone in charge who has no power to make anyone give ground, of course the project stalls halfway.

An ERP project owner is not one position. It is three roles that all have to be filled at the same time: an internal lead who pushes things forward every day (with real authority to get other departments moving, usually the head of operations or finance, not IT); a decision-maker who can settle it on the spot when departments start fighting (almost always the owner personally); and an outside consultant who supplies method and benchmarks (he tells you how others have done it and what traps you will hit if you do it your way, but he should not be the one asked to "drive the project"). The three common failures, handing it to IT, handing it to an outside firm, and the owner carrying it alone, all squeeze these three roles into one person. IT has the skills but no authority. The consultant has the method but no political capital. The owner has the authority but no time to watch the details every day. And the one thing the owner cannot delegate is not project management. It is the final call on trade-offs: when two departments are both right, when an old way of working has to be killed, when someone's existing advantage is about to be laid bare by the system, only you can make that cut.

I'm Johnny Yang of Brightstream Technology in Taiwan. I've led implementation teams for seven years, across 252+ projects in 32+ industries. I have seen every one of the three failure modes below, in person, more than once.

1. Hand it to IT: he has the skills, but not the authority

To be clear, I'm not saying IT doesn't matter. Data migration, permission settings, working through technical specs with the vendor: without IT, almost none of that happens. The problem is the title you gave him, "project lead," and in a small or mid-sized company that title is empty.

Here is what it actually looks like. In month two of the implementation you need to set the part-number rules, and that requires sales, purchasing, and the warehouse to sit down together and fight it out. IT sends the meeting invite. The head of sales says they're pushing for month-end this week, let's do next week. Next week comes, and the person who shows up is a sales support staffer. She has no authority to decide anything and can only go back and ask. After three or more meetings like that, IT learns. He stops asking for things he can't get, finishes the parts he can do on his own, and writes "pending confirmation from sales" on the schedule. From that moment the project is effectively dead in the water. It only looks like it's still running.

It gets worse at acceptance. After go-live, the warehouse numbers don't reconcile, and the first person called in to get yelled at is IT, because "the system is your responsibility." But the numbers usually don't reconcile because the floor isn't following the rules for goods in and out. That is a management problem. IT has no standing to make another department change how it behaves, so he can only change the system again and again to accommodate the floor, until the ERP becomes a strange thing full of exceptions everywhere. In "Why ERP implementations fail" I wrote that many projects don't die from picking the wrong system. They die because the decision-making authority was put in the wrong place. This is the most typical case.

The test is simple: can the person you assigned tell the head of sales directly, "I need this from you by Friday," and will the head of sales actually deliver? If not, he isn't the lead. He's the point of contact.

2. Hand it to an outside consultant: he has the method, but no internal political capital

The second setup looks a lot smarter: nobody inside can carry it, so pay someone outside to carry it. The consultant has seen more projects than you, he knows how the processes should be designed, and he has ready-made project templates. All of that is true. This is my own line of work, so of course I believe an outside view has value.

But there is one thing an outsider can never get: internal political capital. The consultant has no history in your company. He doesn't know that accounting and sales fought over a returns form three years ago and still haven't made up. He doesn't know that the veteran in the warehouse is the owner's wife's cousin. He doesn't know which sentence will leave which person unable to save face. What he can do is make the right recommendation, but in an organization where nobody is willing to give ground, the right recommendation has no force.

The scene I see most often: the consultant finishes a meeting, sends everyone a nice process diagram, and three days later nobody has replied. He follows up and gets "we've been pretty busy lately." What can he do? He can't evaluate anyone. He can't affect anyone's year-end bonus. So all he can do is escalate to the owner, but escalate too many times and the owner starts thinking, "Why does this consultant need to ask me about everything?" At that point the consultant switches to protecting himself: say everything that needs saying, keep the meeting minutes on file, close the project per the contract. The system goes live. Nobody's behavior changes.

So the right place for an outside consultant is not "lead." It is "method and brakes." His job is to tell you how others do it, what problems your decision will grow three months from now, and whether what the vendor just said holds up. That is a different role from the system vendor. If you haven't separated the two yet, read "How to choose an ERP vendor" first. Separating "the person selling you something" from "the person helping you judge" is step one of this whole thing.

3. The owner carries it alone: it won't last three months

This third one is the one I talk owners out of most often, because it performs best at the start. When the owner personally takes over as lead, almost every project flies for the first six weeks. Nobody dares to be late to meetings, any data you ask for shows up the same day, and when two departments fight it gets settled on the spot. At this point the owner is usually very confident and thinks he should have done it himself from the beginning.

The problem comes in weeks seven through ten. A rush order from a big customer, a long-time employee quitting, a shipment stuck in customs: any one of these is enough to pull the owner's attention away completely. And the nature of an implementation is that it doesn't sound an alarm when you're distracted. It just sits there quietly. Two weeks later the owner remembers to ask, "Where's that system at?" and the answer is usually, "Waiting on that document you said you'd look at last time."

There's a side effect here that is harder to deal with. When the owner is the only lead, the other managers automatically step back into the role of spectators. This is the owner's project, not ours. They won't openly oppose it, but they won't commit either, and when something goes wrong they say, "This is what the boss wanted." When it's time after go-live for them to change how they work every day, in their heads they're still just "going along with it." Whether an implementation succeeds is not decided on go-live day. It is decided by whether people slide back to the old way in the three months after, and that depends almost entirely on whether the managers treat it as their own.

So when the owner carries it alone, it doesn't fail. It fades out. The project is never officially stopped. It just drags on and gets thinner until everyone quietly agrees not to mention it again. This one leaves the worst aftermath, because the company draws the wrong conclusion: "We tried. ERP isn't right for us."

4. A setup that actually works

That's the three ways it dies. Now, how to arrange it. The core idea is one sentence: separate the person who pushes things from the person who makes the call, put the outsider in the method seat, and never let any one person play two roles at once.

RoleWho should fill itWhat the job really isWhat happens with the wrong person
Internal project leadHead of operations or finance, with real cross-department authorityPushes progress every day, chases data, puts blockers on the tablePick someone without real authority and the project looks like it's running but is actually stalled
Decision-makerThe owner personallyOnce every two weeks, makes trade-offs and final calls onlyWithout this role, every dispute has no resolution and gets postponed forever
System contactIT or whoever knows the system bestTechnical specs, data migration, permissions, working with the vendorMake him the lead too and he is forced to dodge management problems by changing the system
Outside consultantExternalProvides benchmarks, points out traps, acts as the brakesMake him the lead and all he can do is protect himself and close out the project

The row in this table people most often overlook is the second one. The owner's role is actually not much work. The hard part is keeping the cadence fixed. What I do on site: once every two weeks, meeting kept under an hour, and only two items on the agenda. Where are last time's decisions now, and which cross-department disputes from the past two weeks need you to decide on the spot. Don't sit through technical reports at this meeting. That wastes your time. And don't skip it. Skip it once and four issues will be waiting for you next time.

There's one more step many owners don't want to take, but it nearly decides the outcome: a formal announcement. Pick an occasion when everyone is present, and the owner personally makes three things clear: who is in charge of this project, that this person asks for things on my behalf, and that how well you cooperate will count in your performance review. It takes less than three minutes, but it is the only way to actually put authority in that lead's hands. If you don't say it, he has to rely on personal favors to get every piece of data, and by the third request the favors run out.

As for who should be the lead, the test is not whether they understand systems. It is three things: do they know how the company actually works (not the process on paper), will other departments listen to them, and can they genuinely free up time every week. The third is the most underestimated. If you don't take some of their existing work off their plate, you're asking them to do this project on overtime, and they will rationally let the project slow down.

5. The one thing the owner cannot delegate

You can hand off project management. You can skip the meetings. You can skip the schedule. But there is one thing you cannot hand off no matter how much you trust someone: the final call on trade-offs.

The disputes that truly deadlock an implementation all look alike. Sales wants the system to follow how customers are used to doing things, accounting wants it to follow accounting rules, and both have a point. The warehouse has done it the same way for over ten years, and when you want to switch to barcode scanning, the veteran flatly says no. A senior employee knows that once the data in his hands goes into the system, he is no longer irreplaceable, so he objects with every technical reason he can find. What these disputes have in common: there is no technically correct answer, only an answer to "which way has the company decided to go." And the person who makes that decision has to have the authority to own the consequences. The head of operations doesn't. The consultant even less.

So this is what the owner should do, and the only thing: when both sides are right, decide who gives ground. When an old way of working has to go, say "Starting next month, we don't do it this way anymore." When someone uses seniority to block the road, you go talk to them. Don't leave your project lead to face it alone. Making that cut will be uncomfortable and someone will be unhappy, but if you don't make it, the problem will live on as "add one custom-built feature to the system," and that customization will follow you for the next five years.

Put plainly: where implementations get stuck is usually not something the system can't do. It's that nobody inside has the authority to decide whose convenience gets sacrificed. That authority is only in your hands. You can ignore the schedule, the interface, the report formats. But you cannot skip this decision. Leaving it undecided is itself a decision, just one with a worse outcome.

6. So what should you do now

Do three things this week. None of them cost money.

First, take a sheet of paper, write down the name of the person you have in mind, and answer one question: if they told the head of sales tomorrow, "I need this data by Friday," would they get it? If you can't answer "yes," pick someone else, or first give them the authority they're missing. Second, talk to that person one-on-one. Not about how important the project is, but about which of their current duties you'll take off their plate, and how they should come to you when someone blocks them. Without those two things, they're just there to take the blame. Third, put the one-hour decision meeting every two weeks straight into your calendar, filled out for the next three months, and treat it like a client meeting: no cancellations.

Once you've done these three things, your implementation still hasn't started, but you've already avoided the three most common ways it dies. Vendor selection, quotations, and acceptance are all still open for discussion. But get the lead wrong, and everything you do after that gets diluted.

If you're not sure whether your company even has the right person, book a free 30-minute diagnostic. I'll work through who should fill this role and what authority they need, together with you.

Found this useful? Pass it on
Prices published · free range estimate

So what does a build to your process actually cost?

From NT$300,000, priced against your spec, with same-scope changes free during the maintenance period. The prices and the estimator are in the first screen — get a range first, you do not have to decide anything yet.

Written scope, then a fixed price · staged acceptance · the code and the data are yours · or just book a free 30-minute diagnostic

Johnny Yang (熊董)
Johnny Yang (熊董)
Founder, Brightstream Technology

Serial founder leading a small team that builds systems into an owner's everyday work.

As a founder
7 yrs
Projects delivered
252+
Industries served
32+
About Johnny