ERP programmes: what to look out for
A personal reflection across the thirteen stages of an ERP journey
ERP & Transformation · 5 June 2026 · 4 min read

Across the ERP programmes I've worked on, the same patterns keep appearing. Not exotic failures — ordinary, avoidable ones, repeated at each stage of the journey. This is a personal reflection on where things tend to come unstuck, and what I'd do differently.
1. Business case
Almost every business case I've come across has been built on numbers that were just a little too optimistic — because everyone knows the programme won't get approved otherwise. The result? A gap between what the business thinks it's getting and what's actually deliverable.
2. Requirements gathering
I've sat through many workshops where we spent weeks documenting how things work today — and then basically rebuilt the same thing in the new system. All we did was take the inefficiencies and bake them in permanently. Start from the vendor's reference processes or industry templates, not a blank page. In my experience this alone can cut the requirements phase by at least 30%.
3. Evaluation & selection
I've seen selections decided on from a slick demo built on perfect reference data, or criteria that didn't reflect how complex the business actually was. Two or three vendors, max. A structured scoring matrix everyone agrees on upfront. Keep the RFP tight — a 200-page RFP doesn't make better decisions; it just makes slower ones.
4. Detailed specification
Sounds radical, but detailed functional specs at this stage are almost always going to change — and they slow everything down. Save the detail for after you've picked a vendor.
5. Programme planning
Plans that looked great on paper but had no bearing on reality — built to pass a governance gate, not to reflect what the team could actually deliver or how much capacity the business had.
6. Change & training
Change teams brought in too late, too light. Training delivered once, a few weeks before go-live, on a system that didn't quite look the way it would in production. Then everyone was surprised when adoption was slow.
7. Phasing
Plan for a phased go-live from the very start, and define your Minimum Viable ERP — the smallest coherent scope that delivers real value.
8. Build & configuration
Custom development that everyone assumed would take two weeks but takes four. Configuration decisions made verbally, never documented, impossible to unpick later. Configuration over customisation — always. A process adapted to work with standard ERP functionality can be done in days. Custom code takes weeks, breaks on upgrades, and no one remembers why it was built that way.
9. Testing
The phase that got squeezed when earlier phases ran late. UAT became a box-ticking exercise, not a genuine check that the business was ready. Invest in automated regression testing early. It feels like extra effort upfront, but every time configuration changes (and it always changes), manual re-testing kills weeks.
10. Go-live readiness
Go/No-Go decisions made under huge organisational pressure, with leaders choosing to proceed despite knowing things weren't ready — then paying for it dearly in the weeks that followed. Run a mock Go/No-Go vote two weeks before the real one, and have a rollback plan even if you hope never to use it.
11. Go-live & roll out
The Big Bang. Everything switching on at once, the whole business affected simultaneously, and one bad data issue or integration failure bringing everything to a halt. Make your lowest-risk business unit Wave 1. A clean first wave does wonders for team confidence. And set up a War Room — a dedicated space, physical or virtual, where the right people are all available immediately.
12. Hypercare
The programme team demobilised on a date, not on a readiness signal. The business left to fend for itself before it was genuinely stable. Publish a live dashboard from Day 1. When the data is visible to everyone, the decision to extend or exit hypercare becomes objective. It takes the politics out of it.
13. Closure & continuous improvement
Business case reviews from Phase 1 that nobody ever went back to check. Millions spent, and no one quite sure whether it was worth it. Book the 3, 6 and 12-month reviews — once the team disperses, those meetings simply don't happen. And set up a Centre of Excellence or a Product Owner function: someone whose job it is to keep the system evolving with the business, long after the programme is closed.
None of this is rocket science. Most of it is just about being honest — with the business, with the vendor, and with yourself — earlier than feels comfortable.


