“I've been working on digital transformation lately. I had two system vendors come in for a quotation, and they were very specific: these modules, this many person-days, this much in total. A friend also introduced me to a consulting firm. Their slides were beautiful and everything they said was right, but when they finished I still couldn't tell what they would actually do for me. Who should I hire?”
This is one of the questions I get asked most, and the person asking has usually been stuck at this fork for a while. It isn't that he won't spend the money. He doesn't know what the money will buy.
Let me start with something that may make you a little uncomfortable: the question itself is wrong. “System vendor or consultant” asks who the other side is, but identity doesn't decide the outcome. Something else does — whose income is tied to whether you actually save labor hours after go-live. Swap in that question, sit three vendors in front of you, and you'll know who's who within ten minutes.
I'm Johnny Yang, Brightstream Technology, Taiwan. I've led teams doing implementation for seven years, across 252+ projects in 32+ industries. In those seven years I've seen owners spend serious money on a cabinet full of reports nobody opens. I've also seen owners buy a fully featured system and end up with one extra person on the floor whose job is retyping data. These two kinds of waste look nothing alike, but they have the same root: when the owner signed, he never figured out how the person across the table makes money.
Conclusion first.
There are really only three kinds of outside teams: pure software vendors (sell licenses, do standard implementation), management consultants (do interviews, write reports, don't touch the system), and system integrators or hands-on consultants (change the process and actually go in and change the system). A software vendor's interest is selling more modules and more custom-built work. A management consultant's interest is delivering the report on time and closing the project. Neither role, by the way its contract is written, is accountable for whether the floor is actually easier six months after go-live. That isn't because they're bad. It's what their business model looks like. So don't look at who sounds more convincing. Look at whose pay structure forces him to come back and care about the result. If your needs are standard and you already know exactly what to buy, you don't need to pay consulting fees at all — going straight to a software vendor is the best deal. But if your process has never been mapped, spans departments, or has already failed once, you need someone in the consultant role, no matter whose payroll they're on.
First, match each role's name to what it actually does
In the Taiwan market, all three roles print nearly identical business cards. Plenty of them just say “Digital Transformation Consultant.” So ignore the card. Look at where the company's revenue comes from.
Type one: pure software vendors sell licenses, not results
This one is the easiest to spot. The product already exists. The first half of the pitch is always modules, features, and screens. The biggest line on the quotation is usually the license or annual fee, followed by a block of person-days for implementation services.
Their core skill is installing the product, importing your data, and teaching people to use it. That is genuinely valuable, especially when your needs fall inside what the product was designed for. Then every dollar you pay goes where it should, with nobody in the middle adding a markup.
But understand their blind spot: they won't tell you “this process shouldn't exist.” It isn't their job, it isn't in their skill set, and more importantly — saying it would cost them a module sale.
Here's a very common scene. You tell them: “After sales takes an order, they have to ask purchasing to check if we have the material, and then come back to give the customer a delivery date.” A pure software vendor hears that and says: “No problem, our approval workflow module handles that.” They quote it, build it, launch it, and it does exactly what you described.
The problem is that this “sales asks purchasing, purchasing answers sales” step only grew in the first place as a patch for inventory data that used to be wrong. Once you have real-time inventory, the step can simply disappear. Nobody will tell you that, because saying it means one less approval module sold. And they genuinely have no obligation to question your process — you're the customer, they build what you ask for.
The result: you paid to digitize and preserve a step that should have been deleted. This is the waste I see most often on the floor, and the most expensive. Which part of the process to touch first, and which steps are really patches left over from old problems, I cover in more detail in the first step of digital transformation.
Type two: management consultants deliver reports, not systems
These usually come from larger consulting firms or are senior independent consultants. Their method: interviews, inventory, current-state process maps, gap analysis, a recommended blueprint, a phased roadmap.
Their value is real, and in some situations nobody else can provide it — especially when the company has multiple business units, when the owner himself can't say where he wants to go, or when there's internal politics and you need an outsider to be the one who says the true thing. Interviews sound vague, but a good consultant can take the things “everyone knows but nobody dares say in a meeting” and put them in writing. That alone is worth money.
But their blind spot is just as clear: they don't do the work.
The report will say “Recommend implementing a system with lot traceability to strengthen quality management.” But who picks it, how to negotiate, what to do when the auntie in the warehouse refuses to scan barcodes at go-live — none of that is in the contract. Their contract ends the day the report is delivered and the closing presentation is done.
So you get something correct, but it's a map, not a pair of legs. You take the map to a system vendor, who reads it and says: “This is well written, but getting our product to do this requires custom work. Here, let me give you a quotation.” You pay again.
In fairness to management consultants: they aren't lying. The interviews were real. The analysis was real. The problem is that the contract's finish line is the report, not the result. When the finish line is the report, a complete report delivered on time with nice slides is a successful project — even if nothing in your company has changed a year later.
Type three: system integrators / hands-on consultants change the process and touch the system
The third type sits in the middle. They look at your process first, question how you do things now, recommend cutting some of it, and then go in themselves and change the system to match what they recommended.
There's variation within this group. Some are large system integrators with strong project-winning ability, but the actual execution is often subcontracted, so the people you talk to and the people who build aren't the same. Some are smaller hands-on teams where the person you talk to is the person who builds. The upside is nothing gets lost in handoffs. The downside is they can only take on so many projects at once.
The defining trait of this group: they have to answer for their own recommendations. They made the recommendation and they changed the system, so if the floor gets harder after go-live, there's nobody to blame — they can't say “the consultant's plan was flawed” or “that's what the client asked for.” That structure forces them to think through whether something can actually be built at the moment they recommend it.
I have to be honest about this group's blind spots too, or this post turns into a sales pitch for myself. First, they may shrink their recommendations to fit what they can build — if the best answer is to replace a system they don't know well, they may not say so. Second, their capacity is limited. They can't go deep on many projects at once, so you may not get a slot, or you get one but the lead consultant only shows up in the early phase. Third, they will look more expensive than a pure software vendor, because the diagnostic and process work is real labor, and there's nowhere to hide it.
Pricing models tell you more than the pitch deck
Every recommendation you hear in that meeting room has an incentive structure pushing it. That's not a conspiracy theory. It's what happens in any business. Your job isn't to defend against it. It's to figure out which direction the push is coming from.
Where a software vendor's money comes from
License or annual fees, number of modules, number of users, person-days of custom-built work, and annual maintenance fees after that.
So their best outcome is: the more modules you buy, the more users, the more custom work, the better. That directly shapes their advice. Ask “should we roll this out at the same time?” and the answer leans toward “doing it together is more complete.” Ask “can we skip customizing this part and just handle it manually?” and the answer leans toward “customize it once and you'll never have to do it by hand again.”
Those recommendations aren't necessarily wrong. Often they're right. But know this: when a recommendation happens to benefit them, you need to verify it one more layer yourself. As for which lines on a quotation are most often padded, and which costs only surface in year two, how to calculate ERP implementation cost breaks it all down.
Where a management consultant's money comes from
Project fees, usually paid in stages: signing, midpoint, close. Some also bill by person-day.
So their best outcome is: deliver what the contract describes on time, then close. Report quality matters, of course, because it affects reputation and the next project. But note — quality is measured by “is the report good,” not “did the client actually make it work.”
There's a subtle consequence here. Management consultants tend to write recommendations that are correct but cautious. Write something too specific or too aggressive, and it bites back when execution fails. Write something neutral, comprehensive, and principled, and you can never be wrong. So you get sentences like “Recommend strengthening cross-departmental information transparency” — impossible to argue with, and no clue what to do tomorrow morning.
Whose pay is tied to your result
This is the core of the whole post.
Ask yourself one question: six months after go-live, if the floor isn't any easier, who feels the pain?
- Pure software vendor: no pain. License sold, acceptance signed, maintenance fees still coming in. If you're unhappy you can skip renewal next year, but by then your data lives in their system, and switching costs you more than losing you costs them.
- Management consultant: no pain. Report delivered, project closed, invoice issued. Six months later they're on another project.
- Hands-on consultant / integrator: feels it. Every complaint call you make six months later is labor hours their people have to spend. Every change you need is debt from something they didn't design right. That eats directly into their margin.
See the difference? It isn't about who has more of a conscience. It's whose cost structure gets worse when you fail. With the first two, you carry the cost of failure. With the third, part of that cost comes back to them. That's the one you want.
The three roles side by side
| Comparison | Pure software vendor | Management consultant | System integrator / hands-on consultant |
|---|---|---|---|
| Main deliverable | A working system, training | Current-state analysis, recommended blueprint, roadmap | A reworked process plus a system that actually runs |
| Revenue source | License fees, modules, user count, custom-work person-days, annual maintenance | Project fees, by stage or person-day | Diagnostic and implementation service fees, plus ongoing maintenance |
| Their best outcome | You buy more modules and more custom work | Deliver a passable report on time and close | You change less, customize less, call less after go-live |
| What they do best | Installing a standard product, data migration, user training | Clarifying direction, mapping across business units, being the outsider who says the true thing | Cutting the process down, then mapping it to the system and executing |
| Biggest blind spot | Won't question whether your process should exist | Recommendations are correct but don't land; execution is out of scope | Recommendations tend to shrink to what they can build |
| Accountability if go-live saves no labor hours | Usually none in the contract | Usually none in the contract | Comes back as their own labor cost |
| When they're the best deal | Standard needs, you already know exactly what to buy | Multiple business units, direction undecided, internal politics | Process never mapped, cross-department, or already failed once |
| What to confirm | What costs sit outside the quotation, what you pay from year two | Who executes after the report, and what that costs separately | Is the person you talk to the person who builds, how long the lead consultant stays involved |
Print this table and keep it next to you at your next meeting. The point isn't picking a column. It's confirming which column the other side actually belongs in — a lot of owners' confusion comes from thinking they're talking to type three when the person across the table is type one.
When you don't need to pay for a consultant at all
I'll start with the part that works against me, because it's what you actually need to know.
In three situations, just go to a software vendor. Hiring a consultant is wasted money.
Situation one: your needs are standard and your industry's practices are mature
If you run a straightforward trading business — purchasing, shipping, inventory, invoicing, no triangular trade, no consignment stock, no complex assembly or disassembly — then almost any standard product on the market can handle your needs. What you need is to get it installed, not to rethink the process, because the whole industry has already thought through that process for decades.
In this case, put your effort into comparing vendors and reading contracts, not paying for a process diagnostic. The checklist in how to choose an inventory system is enough.
Situation two: you already know exactly what to buy and which specific problem to solve
If your problem is “I need a system to track warranty periods and repair records” — clear scope, one department, doesn't change how anyone else works — then you need a tool, not a transformation. Find a software vendor that specializes in exactly that, spell out the specs, write the acceptance criteria clearly, done.
A consultant adds very little on a project like this, because there's no process to sort out — you've already defined what you want.
Situation three: the company is still small enough that one person can hold every process in their head
If you can walk through the company's entire operation from start to finish by yourself, without having to ask anyone “what happened with that part,” your process isn't complex enough to need an outsider mapping it. What you need is a tool that's good enough, to connect the most painful part first.
And honestly, if cash flow is still tight right now, keep that money. I said this very bluntly in survive first, then talk transformation, so I won't repeat it here.
When you must have someone in the consultant role
On the other side, there are four situations where going to a software vendor alone has a very high failure rate. Note I said “consultant role,” not “consulting firm” — the role can be an outside team, or it can be someone inside your company who truly understands the whole process and has the authority to act.
One: your process has never been fully mapped
The test is simple: can you draw on a whiteboard every person who touches an order, from the moment it comes in to the moment the money arrives? If you have to call someone halfway through, your process has never been mapped.
What happens if you go straight to a system without mapping it? The vendor asks you to “provide requirements.” You describe things the way you think they work. The floor does things the way they actually work. There's a gap. It only blows up on go-live day, and by then the contract is signed, the specs are locked, and any change is a change order.
Two: it spans departments, and there are scores to settle between them
As soon as two or more departments are involved, and there are questions of “whose responsibility, whose performance numbers, who has to do more work,” you need someone who belongs to neither side to make the call.
A software vendor can't do this. Not because they don't know how, but because they have no standing — they're a vendor, and offending any department head can affect acceptance. So their standard move when departments fight is: build both. The result is two parallel logics in the system, and in the end nobody finds it usable.
Three: you've already switched once, and it failed
This is the one I'm most firm about. If you've already done one implementation that didn't work, the second time absolutely cannot be done the same way.
The reasons a first attempt fails are almost never in the software itself. Maybe the specs weren't clear. Maybe the acceptance criteria were too vague. Maybe the process was never changed and the system was forced on top of it. Maybe key users were never involved from start to finish. If those causes aren't found, you'll switch vendors, switch products, and fail exactly the same way again — just with more money spent and people in the company believing in this even less.
The first thing to do the second time isn't selection. It's a post-mortem. A software vendor can't help you with that, because what needs reviewing includes “how we chose, how we negotiated, how we did acceptance.” I've laid out the common failure patterns in why ERP implementations fail. Before you look for a new vendor, check your last attempt against it.
Four: the owner can't say what he wants either
This is more common than you'd think. The owner knows “things are a mess,” “I'm always waiting for numbers,” “I don't know if we're making money.” But ask him what specifically he wants and he can't say — because what he wants is a result, not a feature.
This is when going to a software vendor is most dangerous, because the vendor will very helpfully “translate” your needs into features their product already has. You'll look at the translation and feel it's roughly right, but it's no longer what you originally wanted.
How we charge, because that is the difference
At this point I have to take a position, or this post becomes a comparison with no point of view. But I don't want to do it like an advertorial, so I'll just tell you our pricing structure and you can judge for yourself.
Brightstream is the third type described above: we change the process, and we touch the system. Every project we take starts with a diagnostic. We look at how the process works now, cut the steps that shouldn't exist, and only then decide what the system should look like. Our AI ERP starts at NT$300,000. The actual number depends on your process complexity and scope. I'm not going to put a fake market rate in a blog post.
The real difference is after go-live.
Here's how we do it: after implementation is complete, as long as it's within the original spec, refinements to usability and workflow fall under ongoing service. That means if a field order is awkward and needs adjusting, if a common action takes too many steps and needs shortening, if a table needs one more filter — adjustments within the originally agreed spec are part of ongoing service, not a new change order every time.
I want to be very clear about the qualifier, because it sets your expectations: “within the original spec” has a boundary. If six months after go-live you decide to add a production line, open a new sales channel, or add a module that was never discussed, that's new scope. We renegotiate and requote. That's reasonable, and it's the part we have to be honest about — nobody can do unlimited work for free.
Why design it this way? Not because we're more generous. Because this structure forces us to get it right during implementation. When post-go-live refinements within the original spec aren't billed separately, every “this is hard to use, can you adjust it” call eats our own labor hours. So in the design phase we work hard to shorten workflows, cut unnecessary fields, and minimize approval layers — because the cost of doing it badly comes back to us.
That's what “pay tied to results,” which I mentioned earlier, looks like in practice. You don't need to trust my attitude. Just read how the contract is written. And I'll say the thing that works against me one more time: if you're in one of the three “don't need a consultant” situations above, don't spend this money. Going straight to a software vendor is the better deal, and we won't force the project.
Tell which type you're facing in the first meeting: ask these three questions
This is the most practical section in the post. You don't need any technical knowledge. Any owner can ask these three questions, and the other side's reaction will sort them immediately.
Question one: “If this isn't faster six months after go-live, what happens on your side?”
This asks who carries the cost of failure.
- Typical pure software vendor reaction: “Our system is very mature, and lots of customers use this feature.” — He's answering whether the product can do it, not where the accountability lies.
- Typical management consultant reaction: “We'll set KPIs and tracking mechanisms in the report.” — He's answering how to measure it, still not where the accountability lies.
- A typical hands-on reaction includes a specific mechanism: how the acceptance criteria are written, which outcomes count as not delivered, whether later adjustments fall within the original scope.
What matters isn't how good it sounds, but whether he mentions a mechanism that would cost him something. If he doesn't mention one, there isn't one.
Question two: “How do you plan to find out what my process looks like now?”
This asks whether he intends to understand you at all.
If the answer is “we'll ask you to provide a requirements spec” or “we have a requirements questionnaire for you to fill out,” he's assuming you know what you want: you define it, he implements it. That's standard for a pure software vendor, nothing wrong with it, but it means if the spec is wrong, that's on you.
If the answer is “we'll run several interviews and draw current-state process maps,” that's the management consultant approach. Your follow-up should be: after the maps are drawn, who does the work?
If he says he wants to visit the floor, sit with the people who actually do the work for a while, and see the Excel files you're actually using right now, he plans to execute. Only the people who will actually do the work need to know the details on the floor.
Question three: “How is the end of this project defined? After it ends, who do I call when there's a problem, and what does it cost?”
This is the toughest one, because it tests two things at once: where the finish line is, and how billing works after the finish line.
You're listening for three specific answers: what the acceptance criteria are (“the system runs normally” isn't a criterion, it's filler); how long after go-live adjustments count as within the original scope; and how work beyond the original scope is priced, and whether there are tiers.
A vendor who glosses over this question, no matter how good the rest of the pitch was, I'd set aside for now. People willing to spell out the finish line and the accountability after it are usually the ones who actually intend to finish. For which other contract clauses to watch, use this together with how to choose an ERP vendor.
One more: “Is it possible I don't need to do this at all?”
This is my personal favorite, because it tests directly whether the other side dares to say something that hurts their own interest.
If a requirement you raise actually has a simpler solution, or doesn't need to exist at all, someone who's really on your side will tell you. Someone who makes money selling modules will naturally say “we can do that.”
Neither answer is illegal, but you should know the difference.
FAQ
What's the actual difference between a digital transformation consultant and a system integrator?
The deliverable, and where accountability ends. A management consultant delivers an analysis report, a recommended blueprint, and a roadmap. The contract ends the day the report is delivered and the closing presentation is done. Who executes, and whether it works, is usually out of scope. A system integrator delivers a system that actually runs, handling data migration, integrations, training, and go-live. In between is the hands-on consultant, who maps the process first and then touches the system, connecting the two. In practice these three terms are used interchangeably in the Taiwan market, and any of them may print “Digital Transformation Consultant” on their card. So ignore the label and check two things: does their revenue come mainly from licenses or from services, and on what day does their contract end. Ask those two and the role is clear.
How do I choose an outside team? If I only get one quotation, will I get taken advantage of?
Get at least three, but the point isn't comparing prices. It's comparing how different their solutions to the same problem are. Put three quotations side by side. If one total is clearly much lower, don't look at how cheap it is — look at what they left out. Usually it's data migration, training, or post-go-live support, and all of it comes back to bill you later. A more effective approach: describe your most painful specific scenario to all three and see who asks you follow-up questions and who just answers “we can do that.” The one asking questions is usually actually thinking about how to build it. The one saying yes right away is mostly thinking about how to quote it. Compare price last. Compare understanding first.
I already have an MIS or IT person. Do I still need an outside consultant?
It depends on what your MIS person does day to day. If the job is maintaining the network, fixing computers, and managing account permissions, that's IT operations, not process consulting. They're completely different skills. A process consultant does cross-department coordination, questions existing practices, and redesigns workflows, which takes an understanding of the business and a certain amount of authority. On the other hand, if your IT person already sits in business meetings, knows where each department hurts, and gets listened to, they're actually better suited to this role than an outside consultant. What you need to do is give them time and authority, and bring in outside help only to fill the system experience they lack. The test is simple: can they tell the sales manager, in a cross-department meeting, “you don't need to do this step anymore”?
Consulting fees look more expensive than software licenses. Is that reasonable?
It depends on what you're buying. A license fee buys software that's already written, with development costs spread across all customers, so the price can be low. Consulting or implementation fees buy people's time, which can't be spread, so the price looks high. The real comparison isn't which number is bigger, but which side your problem is on. If your problem is a missing tool, putting the money into licenses is right, and keeping consulting fees to a minimum is fine. If your problem is a messy process and departments that don't line up, even the most expensive system won't fix it, because the system will faithfully copy the mess. The most expensive waste I've seen isn't consulting fees. It's paying to digitize and preserve a process that should have been deleted.
Once I hire a consultant, do I still need to spend my own time?
The opposite. A consultant brings method, experience, and an outside view, but two things an outsider can never do. One is decisions — which department gives ground, which habit gets dropped, who gets resources first. Only you can make those calls. The other is signaling — people in the company judge whether something matters not by what the consultant says, but by whether the owner shows up for it. In practice, the most common failure pattern in implementation projects is the owner “handing off” the project and then disappearing, only to discover a month before go-live that the direction is wrong. The time you need to put in isn't much, but you must be at the key meetings, and you must make decisions in front of everyone. Nobody can do that for you.
If you're stuck right here
If you have two or three quotations in hand, every vendor makes sense, and you just can't decide, what you're missing usually isn't more information. It's an outside opinion on whether your situation actually needs someone in the consultant role.
You can do two things on your own first. One, take the comparison table above and go through it column by column against each vendor you're talking to, and confirm which type each one is — for a lot of people the confusion disappears at this step. Two, take those three questions exactly as written and ask them, and see which cell each answer falls into.
If you find you're in one of the three “don't need a consultant” situations, good. You've saved money. Go compare software vendors directly. That's my honest advice.
If you find you're in one of the four “must have a consultant role” situations — especially the one where it's already failed once — we offer a free diagnostic. We'll map your current state with you and pin down the problem: is it the process, the selection, or is it simply not time to take this on yet. The diagnostic doesn't commit you to anything further, and if we judge you don't need us, we'll tell you directly. If you'd rather map it yourself first, start with the SME ERP buying guide, list the questions you need to ask, and bring that list to your next meeting. It makes a big difference.




