ERP projects rarely fail on go-live day. They fail in the weeks after, when the first month-end close, the first stock count, and the first busy sales week hit the new system at the same time. By then the warning signs have usually been visible for months. Below are the seven patterns we see most often, what each looks like in practice, and what to do while there is still time to fix it.
The first signal is unclear ownership between business and IT. If you ask who decides how sales orders are approved in the new system and get three different names, nobody owns it. IT assumes finance will decide, finance assumes the vendor will propose something, and the vendor waits. Every process area needs one named owner from the business side who can sign off a design and answer questions within a day, not after the next steering meeting.
The second signal is scope that keeps quietly expanding. It rarely arrives as a big change request. It arrives as small additions: one more report, an approval step for a single customer, a field that only one branch uses. Each looks harmless, and together they push the timeline back by months. Keep a written scope list, log every addition with its cost in days, and make the project owner approve or reject it openly.
The third signal is data migration treated as an afterthought. Teams spend months on configuration and leave customer, item, and opening balance data for the last two weeks. Then they discover duplicate customers, item codes that mean different things in different warehouses, and receivables nobody can reconcile. Start cleaning master data in the first month, assign an owner per data set, and run at least two trial imports before the real one.
The fourth signal is training compressed into the final week. Users sit through a long demo, nod, and on day one go back to the spreadsheets they trust. Training belongs in the project plan from the start, with key users involved in testing and end users practicing their own daily transactions on a test environment. If the training plan is one session per department, it is not a plan.
The fifth signal is no defined cutover plan. Cutover is the weekend or week when you stop the old system, move the final balances, and start transacting in the new one. Without a written plan, nobody knows when the last invoice goes into the old system, who counts stock, or what happens if the import fails. A good cutover plan lists every task, an owner, a time, and a rollback decision point.
The sixth signal is custom development without governance. Developers build whatever is requested because nobody asked whether the standard function would do. Each customization adds testing work now and upgrade work later. Before approving any custom feature, ask what the standard system does, why that is not enough, and what it will cost to maintain after the next version update. Many requests disappear once that question is asked.
The seventh signal is leadership that disappears once the contract is signed. The director attends the kickoff, then delegates everything. When a department refuses to change how it works, there is nobody with the authority to settle it, so the system gets bent around old habits. Leadership does not need to attend every meeting, but it needs a short monthly review and must make the hard calls when teams disagree.
Each signal is fixable on its own. The danger is when three or four appear together, because then the project stops being an implementation and becomes a recovery effort. Buying more software or adding modules does not help. What helps is naming owners, resetting the timeline to something realistic, freezing scope to the original business case, and getting data and training back on the critical path before go-live, not after it.
If you recognize several of these signals in your own project, an outside review can help before the problem reaches month-end. Kerja Itu Mudah works online with companies across Indonesia on ERP implementation and project recovery. Tell us where your project stands and which signals you are seeing, and we will help you decide what to fix first.