UT, SIT and UAT are three different gates, and training should start in Analysis, not the week before go-live. Here is where each one sits in the six Sure Step phases, and why a manufacturing or warehousing project takes longer than the standard timeline.
Two questions come up on almost every Business Central project: "when do we test?" and "when do we train?". The honest answer is that both start much earlier than most people expect, and that getting them in the wrong place is the most common reason a go-live slips.
We run three distinct kinds of testing, each with its own sign-off, and we do not start the next one until the previous one is signed off.
Done by the developer, during the Development phase, on every AL extension and every integration. Each piece is tested on its own against the design before it is handed over. If a posting routine or an API call does not behave in isolation, it has no business being in an end-to-end test.
Still in the Development phase, but now the whole thing together: Business Central with your migrated data, the extensions, and the integrations to your other systems (web shop, EDI, payroll, warehouse scanners, Power BI). The test scripts follow real business processes end to end: order to cash, purchase to pay, month-end close. SIT sign-off is the gate into Deployment.
In the Deployment phase, and run by your people, not ours. Your super users work through the scripts using their own data and their own judgement about whether the system does what the business needs. UAT sign-off is the gate to go-live. If UAT is the first time your users see the system, something has gone wrong earlier in the project.
This is the part that is usually done backwards. Training is treated as a thing that happens the week before go-live, in a room, from a slide deck. By then it is too late for the users to influence anything and too late to find out that a process was misunderstood.
We do it in two parts:
The side effect is that your super users become the first line of support after go-live, which is what makes hypercare short and the support contract cheap.
| Phase | Testing | Training |
|---|---|---|
| 1 Diagnostic | – | – |
| 2 Analysis | – | Super users named, training plan agreed |
| 3 Design | Test strategy: UT, SIT and UAT scripts drafted against the design | Super users in the design workshops |
| 4 Development | UT on every extension, then SIT end to end with migrated data | Super users trained on the sandbox as it is built |
| 5 Deployment | UAT run by the super users; sign-off is the gate to go-live | End-user training alongside UAT |
| 6 Operation | Regression testing on each Microsoft release | Refreshers as the system grows |
The typical durations on our lifecycle graphic (one to two weeks for Diagnostic, two to four for Analysis and Design, four to twelve for Development, two to four for Deployment) are for a standard implementation: finance, sales, purchasing and stock on largely standard Business Central.
Projects that include manufacturing (routings, BOMs, capacity, subcontracting), advanced warehousing (directed put-away and pick, bins, scanners) or complex integrations with other systems take considerably longer, mostly in Design, Development and Deployment, because there is more to design, more to build, more to test end to end and more people to train. We scope those timelines separately once we understand what is involved, rather than quoting a standard number and watching it slip.
If you would like to talk through where your project sits, book a 30 minute call.
