Most ERP rollouts are declared successful on go-live day and quietly undermined over the following quarter. Nobody revolts. There's no meeting where the team announces it is abandoning the system. Instead, a scheduler starts keeping a side spreadsheet "just for this week." A salesperson keeps quoting from an old template because the new one takes four clicks too many. A warehouse lead writes counts on paper and enters them Friday afternoon. Six months later, leadership is looking at reports that don't match reality and wondering why the investment didn't land.
The instinct is to blame adoption and schedule more training. But people who sat through training and still built a workaround weren't confused. They were solving a problem your implementation handed them.
Workarounds Are Diagnostic, Not Disobedient
A workaround is a signal that the system costs more than the task is worth. Someone measured the official path — the clicks, the required fields, the wait for an approval, the report that doesn't exist — and found a cheaper route. That decision is usually rational at the individual level and expensive at the company level, because the data that lives in a side spreadsheet is invisible to everything downstream: forecasting, inventory, cash planning, commissions.
So the useful question isn't "how do we get people to comply?" It's "what made the official path expensive?" The answers are almost always mundane and fixable:
- A field the system requires that the person genuinely doesn't know at that moment in the process
- A screen that takes thirty seconds to load on warehouse hardware or a weak connection
- An approval step that routes to someone who approves everything anyway
- A report that exists in the old system and was never rebuilt in the new one
- A permission setting that blocks someone from finishing their own work
None of those are training problems. All of them are design problems, and most are a few hours of configuration away from being solved.
Follow the Exports
If you want to find the workarounds in your business, don't run a survey. Look at where data leaves the system and where it never arrives.
Start with exports. Every CSV download is a vote of no confidence in a report. Find out who downloads what, how often, and what they do with it after. In Odoo and most modern ERPs, you can see this directly, and the pattern is usually concentrated: a handful of people pulling the same data weekly because the dashboard they need was never built.
Then look for timestamp clustering. If an entire week of transactions gets entered between 3pm and 5pm on Friday, the work isn't happening in the system — it's being transcribed into the system from somewhere else. That gap is where your data quality goes to die, because transcription introduces errors and delay at the same time.
Finally, walk the floor. Ask people to show you how they actually do the task, not how they were taught. The sticky notes, shared drives, and personal spreadsheets will surface in about ten minutes if you make it clear you're there to remove friction rather than enforce rules.
Design for the Ten-Second Task
The fix is rarely a bigger system. It's a shorter path for the handful of actions people perform dozens of times a day.
That means stripping required fields down to what's genuinely knowable at that step and capturing the rest later in the process. It means building the one report people keep exporting, so they stop. It means turning approvals that always get a yes into rules the system enforces without a human in the loop. It means putting the three most common actions for each role on a screen those people open first, instead of expecting them to navigate a menu designed by the implementation team.
And it means accepting that some workarounds are correct. If a team's process genuinely doesn't match how the module was configured, the honest answer is to reconfigure — or to integrate the tool they're actually using so the data flows back automatically. Forcing a bad fit creates a permanent tax on everyone who touches it.
Measure Adoption Where Data Is Created
Login counts tell you very little. A better measure is what percentage of transactions originate in the system versus arriving through an import or a late batch entry. If orders, time, counts, and quotes are created in the ERP at the moment they happen, the system is working. If they're created elsewhere and loaded later, you have a reporting tool, not an operating system — and you're paying ERP prices for a database.
Track that by department and revisit it quarterly. It will point you directly at the next thing to fix, and it keeps the conversation focused on friction rather than compliance.
Where to Start
If your rollout technically succeeded but your reports still don't match what the team knows to be true, the gap is almost certainly a set of small, specific frictions nobody escalated. Finding them takes a careful look at exports, timing, and the actual work — and fixing them takes someone who knows both the platform and how operations really run.
That's the work we do. If you're staring at an ERP that your team tolerates rather than uses, reach out to Infraxio and let's find the workarounds together.