An ERP touches every order, invoice and stock count in your business, so when an ERP implementation goes wrong, everyone feels it. The software is rarely the reason. Most problems trace back to a handful of decisions made before or early in the project. This article walks through eight common mistakes, how to spot each one while there is still time, and what to do instead.
Why ERP implementation goes wrong so often
The track record is sobering. Based on more than 6,000 client conversations, Gartner estimated in 2024 that more than 70% of recently implemented ERP initiatives will fail to fully meet their original business case goals by 2027, and that as many as 25% of those will fail catastrophically.
Projects are also long and expensive. In Panorama Consulting Group's 2024 ERP Report, the median project cost was $450,000 and the median timeline was 15.5 months.
The encouraging part is that the same research points to causes you control. Here they are.
Mistake 1: Treating it as an IT project
What it looks like: the IT manager owns the project, department heads attend the kickoff and then disappear, and success is defined as "the system went live".
Why it hurts: an ERP changes how sales, production, stock and finance work together. Gartner's 2023 survey found that almost 75% of respondents with an ERP strategy said it was not strongly aligned with their overall business strategy, and it names that alignment as one of the top predictors of success.
What to do instead: write down three to five business results the ERP must deliver, such as "quote a delivery date from live stock" or "close the month in five working days". Make a senior operations person accountable for them, with IT in support.
Mistake 2: Switching everything on at once
What it looks like: one go-live date for every module and every site, after a year or more of building.
Why it hurts: nothing is tested by real daily use until the end, so every problem arrives on the same weekend. Rolling back is hard because the old systems have already been switched off.
What to do instead: start with the module that removes the most manual work, run it with one department, and add the next module once the first has earned its place. A phased plan also lets you stop and adjust if priorities change.
Mistake 3: Bending your business to a template, or copying every old habit
What it looks like: either the team accepts the vendor's standard process for everything, including the steps that make you different, or it asks for every quirk of the old system to be rebuilt exactly.
Why it hurts: the first route forces workarounds into spreadsheets beside the new system. The second rebuilds old problems at new prices.
What to do instead: sort your processes into two groups. Keep the ones that are part of why customers choose you, and simplify the rest. A custom ERP makes sense when the first group is large; a packaged system fits when it is small.
Mistake 4: Leaving data migration to the end
What it looks like: the plan has a single line near go-live that says "migrate data".
Why it hurts: old systems are full of duplicate customers, retired product codes and stock figures nobody trusts. Discovering this a few weeks before launch causes delays or, worse, a new system loaded with bad records.
What to do instead: pull a sample of real data in the first month. Decide who owns each data set, what gets cleaned, what gets archived, and run at least one full trial load before go-live. If the data sits in a system too old to export cleanly, plan that work as its own track, which is often part of legacy system modernization.
Mistake 5: Underestimating your own people's time
What it looks like: key users are expected to design processes, test screens and train colleagues on top of a full workload.
Why it hurts: in Panorama's 2024 study, the most common reason projects ran over schedule was resource constraints, meaning staffing, budget or both.
What to do instead: name the people who will work on the project and free up part of their week in writing. Backfill their routine work for the busiest phases. Budget for this time as a real project cost.
Mistake 6: Treating training as a one-day event
What it looks like: a slide deck and a demo the week before launch, then everyone is on their own.
Why it hurts: people go back to spreadsheets and phone calls when the new system feels slower, and the data in the ERP quietly goes stale. Prosci's research found that projects with excellent change management were about seven times more likely to meet their objectives than those with poor change management. Panorama's 2024 report also found that smaller organizations were more likely than larger ones to give change management little or no focus.
What to do instead: show staff what the system fixes for them, involve them in testing screens built with their own products and customers, and keep support close for the first weeks of daily use.
Mistake 7: Forgetting the systems that stay
What it looks like: the ERP plan covers the ERP, but not the accounting software, e-commerce site, warehouse scanners or payroll tool it must exchange data with.
Why it hurts: integrations discovered late are expensive. In the Panorama report, the most common reason for going over budget was an unexpected need for additional technology.
What to do instead: list every system that will send or receive data, and decide for each one whether it is replaced, connected or retired. Price the connections up front.
Mistake 8: Nobody can make the final call
What it looks like: decisions wait for a steering committee that meets once a month, and disputes between departments go round in circles.
Why it hurts: every week of waiting is a week of paid project time and a week of staff losing interest.
What to do instead: give one named person the authority to settle scope and process questions within days. Set a simple rule: anything that changes the budget or the go-live date goes to the owner or MD, everything else is decided by that person.
A pre-kickoff checklist
Before you sign a contract or start a build, check that you can tick each of these:
Three to five written business results the ERP must deliver, with an owner for each.
A map of how an order moves through your departments today, from quote to invoice.
A ranked list of modules, starting with the one that removes the most manual work.
A list of every system the ERP must connect to, with a decision for each.
A sample of your real data pulled, with known problems written down.
Named key users with protected time on the project.
One decision-maker with clear authority and a fast turnaround.
A plan for training and support after go-live, with a budget line.
A test copy of the system where every change is tried before staff see it.
If several of these are blank, that is useful information. Filling them in first costs far less than discovering the gaps halfway through.
Frequently asked questions
How long does an ERP implementation take?
It depends on the number of modules, sites and integrations. Panorama's 2024 median was 15.5 months for the projects it studied. A phased approach gets the first module into daily use much sooner, while the rest follows in order of value.
Should we choose a packaged ERP or a custom one?
Packaged systems suit businesses whose processes match the standard well. Custom makes sense when the way you work is part of your advantage, or when a packaged system needs a workaround for every order. You can also start with one custom module next to your existing accounting software.
Can we keep our accounting software?
Often, yes. Many businesses connect the ERP to the accounting package they already trust and move other departments first. Treat that connection as a planned piece of work, priced at the start.
What is the first step if a project is already in trouble?
Pause new features and get an independent review of scope, data and integrations. A short technical advisory engagement can tell you what to keep, what to fix and what to stop.
If you are planning an ERP or rescuing one, we follow your orders through each department before suggesting anything, and you can read how we work before talking to us. When you are ready, book a 30-minute call with one of our founders.
