Define ERP Success in Numbers, or You'll Never Agree You Got There

By Steve Leadbeater 

There's a line that shows up in roughly every third article written about ERP: you cannot manage what you cannot measure. It's almost always attributed to Peter Drucker.

He didn't say it. There's no source for it anywhere in his work, and the sentiment sits awkwardly against what he actually wrote — Drucker was fairly explicit that some of the most important things in an organization resist measurement entirely, and that managing only what's easy to count is a good way to manage the wrong things.

Worth mentioning not to be pedantic, but because the misattributed version does real damage. Taken literally, it encourages programs to measure whatever's convenient and call it strategy. A weaker claim is both truer and far more useful: if you don't decide in advance what success looks like, you will never agree afterward on whether you achieved it.

The Problem with a Vague Target

"Improved efficiency." "Better visibility." "A platform for growth." These are perfectly good reasons to start a program and completely useless as a definition of done — because they can be argued either way, indefinitely.

Without a hard definition, the assessment becomes political. The project team can point to a working system, delivered on schedule. A skeptical department head can point to a process that still annoys them. Both are telling the truth. Neither can be contradicted. The verdict ends up depending on who has the floor — which is how the same implementation gets described as a clear success and a costly disappointment by two people in the same company, in the same month.

What a Real Target Looks Like

Specific, numeric, dated, and evidenced by the system — not by opinion. For a manufacturer, that usually means something from this family:

• Days to close the books — from wherever you are now to a specific number, by a specific month-end.
• Inventory record accuracy, measured by cycle count, to a stated percentage.
• Percentage of orders shipped complete against the first promised date.
• Average age of open work in process, and the count of jobs open beyond a threshold.
• Elapsed time from shipment posting to invoice.
• Manual reconciliations performed per month — with a target of zero, which is one of the few targets on this list that's genuinely unambiguous.

Four or five of these is plenty. A dozen is a dashboard nobody reads.

The Step Almost Everybody Skips: Baseline Before Go-Live

This is the single most common omission we see, and it's unrecoverable. Once the legacy system is switched off, you can't go back and find out how long it used to take to close the books, how accurate stock actually was, or how often an order shipped complete the first time. That comparison is gone for good.

What you're left with instead is a year of people saying "it feels slower than before," with no way to test the claim. That's corrosive precisely because it can't be checked — and quite often it's wrong, and being able to demonstrate that it's wrong would be worth a great deal to the program.

So measure the old world first, even roughly, even if it takes a couple of weeks of somebody's time. A rough baseline beats no baseline by a distance.

Who Agrees the Numbers

The people who will later be asked whether it worked. Typically the finance lead, the operations lead, and whoever holds the budget — and they need to agree with each other, in writing, before selection, not after implementation.

The reason to do it that early is that this conversation surfaces real disagreement about priorities. If your CFO is optimizing for close speed and your operations director is optimizing for shop floor responsiveness, that's fine — but it will shape design decisions, and it's far better to know about it in month zero than to discover it in a heated conversation in month nine.

What This Buys You

Two things. First, an honest answer at the end — one you'll want whether it's flattering or not. Second, and less obviously, a much easier program in the middle, because when a scope decision comes up mid-project, you have something concrete to test it against.

"Does this move one of our six numbers?" is a far better question than "do we want this?" — and it's the only version of the question that a room full of people with different priorities can actually answer together.

Setting these numbers is easier with someone who's watched a hundred manufacturers get it right — and wrong. If you're heading into a selection process and want a second pair of eyes on what "success" should actually look like for your operation, talk to us before you're locked into a scope.