What Happens After Odoo ERP Go-Live?
Go-live is the day everyone plans for and almost nobody plans past. The project team disbands, the consultants leave, and the business discovers that an ERP is not a thing you install — it is a thing you operate.
The first ninety days after go-live decide whether the system becomes the way the company works or the thing people work around. This is what actually happens in that window, what goes wrong, and what to have agreed before anyone switches over.
Week one: the system is slower than the spreadsheet
Expect productivity to drop. A store keeper who could write a line in a register in four seconds now scans, selects a location and confirms. A salesperson who quoted from memory now picks a pricelist. Everything takes longer, and everyone notices.
This is normal and temporary, but it is also the moment adoption is won or lost. Two things help. First, somebody internal who knows the system should be physically near the people using it, not reachable by email. Second, the consultant should still be engaged — a question that takes two minutes to answer on day three turns into a workaround that survives for years if nobody answers it.
What does not help is telling people it will get better. It will, but on day three that sounds like an excuse.
Weeks two to four: the exceptions arrive
Testing covers the normal path. Live operation finds the rest: the customer who pays in two currencies, the return that came back without the original invoice, the item that ships in a unit nobody configured, the approval needed when the approver is on leave.
These are not implementation failures. They are the cases a business handles by judgement and never wrote down, and they surface only under real volume. Expect a steady stream of them for about a month, and expect most to be resolved by configuration rather than development.
Keep a list. The pattern in that list tells you more about your business than any process document, and it is the basis for the first round of post-go-live changes.
The first month end
This is the real test, and it is where problems that have been quietly accumulating become visible. Stock that was wrong on day one is still wrong. Transactions keyed into the wrong period have to be found. The trial balance does not match what the old system said, and somebody has to explain the difference.
Three things make the first close survivable:
- Opening balances reconciled before go-live, not after. If physical stock and system stock disagreed on day one, every report built on them is suspect and the close takes twice as long.
- A cut-off decision everyone followed. Which transactions belonged to the old system and which to the new, agreed in writing and actually observed.
- The old system still readable. Not writable — readable. You will need to look something up, and losing that access makes the close an archaeology exercise.
Months two and three: the system stops being a project
By the second month the exception list has slowed and people have stopped narrating what they are doing. This is when two things should happen that often do not.
Reports get built. During implementation everybody focuses on transactions. The reports managers actually need become clear only once real data is in, and they are usually different from the ones specified up front.
Process gets revisited. Some of the workarounds adopted in week one were sensible under pressure and are now permanent habits that cost time. Reviewing them at month three, while the alternatives are still remembered, is cheap. Reviewing them at month twelve means retraining.
What to agree before go-live, not after
Who answers questions, and how fast
Support scope and response times in writing, agreed while you still have leverage. Negotiating a support contract during your first month-end is the worst possible timing. See Odoo support and maintenance.
Who owns the system internally
One named person who can say "this process changes" and who holds the backlog of post-go-live requests. Without them, every small change becomes a committee.
What the hypercare period covers
The weeks immediately after go-live, when issues are fixed as part of the project rather than billed. The length and the boundary should be explicit — a configuration fix is hypercare, a new report is a change request.
How changes get requested
A single place requests land, so the fifteen things people want in month two are visible as a list rather than arriving as fifteen separate phone calls to whoever they know.
When the first review happens
Book it before go-live, for about six weeks after. Something scheduled gets done; something intended does not.
What usually goes wrong
The internal champion goes back to their day job. They were seconded to the project, the project ended, and the knowledge left with them. Keep some of their time ring-fenced for the first quarter.
Training was given to managers. Adoption is decided by the people who key transactions daily — store keepers, accountants, operators, counter staff. If they were not in the room, expect week one to be harder than it should be. See Odoo training.
Opening stock was never counted properly. The most expensive shortcut in any implementation. Everything downstream inherits the error and nobody trusts the system afterwards.
The old system was switched off immediately. Keep it readable for at least one full close.
Nobody owns the error queue. Integrations fail. If no named person checks the failures, data diverges quietly and the divergence is found at audit.
Frequently asked questions
Roughly ninety days. Week one is slower than the old way and that is normal; weeks two to four surface the exceptions nobody wrote down; the first month end is the real test; and by month two or three the system stops being a project. The main variable is whether support and internal ownership were agreed before go-live.
Because a system enforces steps that people previously held in their heads. A register line takes four seconds; a scanned, located, confirmed stock move takes longer. Productivity recovers as the steps become automatic, but in week one it genuinely costs time, and pretending otherwise damages trust.
The period immediately after go-live when issues are resolved as part of the project rather than billed as support. Its length and boundary should be written down before cutover: a configuration fix is hypercare, a report that never existed is a change request.
Keep it readable, not writable, for at least one full financial close. You will need to look something up during the first month end, and losing that access turns the close into an archaeology exercise. Writable access, on the other hand, guarantees someone keeps using it.
Opening stock that was never properly counted. Every report built on it is suspect, the first close takes twice as long, and people conclude the system is wrong rather than the data. A physical count before cutover is part of the project, not an optional extra.


