Project management
An ERP programme has two plans: the one your systems integrator sells you and the one your business can actually live with. Client side ERP project management exists to own the second, hold the first to account and tell the board the truth about both.
Delivered for and alongside
































The systems integrator will arrive with a project manager, a solution architect and a delivery team. None of them work for you. Good ERP project management starts by staffing the client side properly, and that means four distinct jobs rather than one overworked finance manager.
Write the responsibilities down early in a simple matrix with the integrator's roles beside yours. The gaps that matrix exposes, usually data cleansing, test execution and training, are the gaps the vendor's proposal assumed you would fill.
Governance is not a monthly slide pack. It is a small number of forums with the authority to decide, and a set of gates at which the programme must prove it is fit to continue. The UK government's gate review model is a useful pattern even for a private company: independent reviews at key decision points, anchored to the Five Case business case, from business justification through investment decision to a post go live review of operations and benefits. Scale it down, keep the principle.
For an ERP programme we set six gates: mobilisation complete, design signed off, build and unit test complete, user acceptance exit, go live readiness, and hypercare exit with benefits tracking in place. Each gate has written criteria agreed before the phase starts, and each is judged by someone other than the people who did the work. A design gate, for example, asks whether every workstream lead has signed their process designs, whether the data mapping covers every object in scope, whether every interface has a named owner on both sides and whether the open design decisions are below an agreed number. If the answer is no, the phase does not close, however inconvenient the date.
Above the gates sit three forums. A programme board, chaired by the sponsor, meets monthly to take decisions, not to be briefed. A design authority, chaired by your programme director or an independent assurer, settles configuration and process disputes within days so the build does not stall. A weekly delivery meeting between your project manager and the integrator's runs the plan. Every decision goes in a log with the date, the owner and the reasoning, because the question in month fourteen is always why something was decided in month three. Our ERP implementation plan article sets out how the phases and gates map onto a real calendar, and the ERP implementation timeline guide shows how long each phase genuinely takes.
There should be one integrated plan, owned by your project manager, into which the integrator's plan fits as a set of dependencies. Client activities sit on it with the same rigour as the vendor's: data extraction, cleansing and reconciliation cycles, test script writing and execution, training design and delivery, interface build on your side, cutover rehearsals and the business calendar of year end, audit, payroll runs and peak trading that nobody may plan a go live across. The critical path on most ERP programmes runs through data and testing, both client activities, which is why a plan drawn only from the integrator's side looks healthy until it is not.
Budget control means tracking four pots separately: software subscriptions, integrator fees, internal costs including backfill, and contingency held by the programme director and invisible to the integrator. Change requests are where budgets die, so every one is traced back to the statement of work before it is priced. If a gap was in the agreed scope, it is not a change. In 19 years of running programmes, the client side habit that saves most money is reading the statement of work properly before signing and then holding the integrator to the deliverables it lists, including the format and acceptance criteria of each.
Managing the integrator is a relationship with teeth. Agree named people and a substitution clause. Tie payments to accepted deliverables and gates, not calendar dates. Hold a weekly performance review of the integrator's plan, resourcing and quality, and record the position in writing. Escalate early and formally, because the account executive will only hear a problem that is on paper. Where the contract allows remedies for late or defective delivery, decide in advance when you would use them, so the threat is credible without being reached for in anger. None of this is hostility; it is the behaviour a good integrator expects from a competent client, and the ones worth hiring welcome it.
A risk register owned by the integrator records the risks to the integrator. Yours records the risks to the business: data that will not reconcile, a finance team that cannot test and close at the same time, a payroll interface nobody has specified, one person holding all knowledge of the old system, a parallel programme competing for the same people, a go live on a quarter end. Each risk has an owner, a trigger and a costed response, and the board reviews the top ten monthly until they are closed or accepted.
Dependencies are mapped as carefully as tasks. Which decisions block which configuration? Which data objects must be clean before which test cycle? Which third parties, banks, HMRC gateways, customers and suppliers, must change something on their side, and how long do they take? When a dependency slips, the plan shows what moves with it the same day.
When the vendor runs the plan instead, the failure pattern is consistent. The plan is optimised for the integrator's resource calendar rather than your business calendar. Milestones are documents delivered rather than business readiness achieved. Test exit criteria are set by the people who built the system. Scope that was in the proposal reappears as change requests. Data problems surface in the final rehearsal because nobody owned them earlier. The go live date hardens for commercial reasons while readiness is still amber, and benefits have no owner because the integrator leaves at hypercare. The National Audit Office found the same faults in March 2026 across the UK government's shared services programme: no clear owner with the levers to deliver, boards with no clear remit and integration problems that forced a reset. Our five signs an ERP programme is in trouble article will tell you whether you are there; ERP programme recovery is what follows. We offer client side project management as assurance beside your integrator or as delivery by our own practitioners, with fees shown plainly on our pricing page.
A client side ERP project manager owns the integrated plan, runs the client workstreams for data, testing, training, integration and cutover, manages actions, risks and dependencies, and holds the systems integrator to its statement of work. The role reports to a programme director or sponsor and must be independent of the integrator, with authority to challenge dates, scope and quality on the client's behalf rather than simply relaying the vendor's plan.
For anything beyond a single site, single ledger implementation, yes. The project manager runs the plan; the programme director owns the business case, budget, vendor relationship and go live decision, and carries the board's authority into disputes with the integrator. Combining the roles leaves nobody above the daily firefight to make the hard calls, and giving the job to the finance director alongside their day job usually means neither is done well.
Stage gates are decision points at which the programme must prove, against criteria agreed in advance, that it is ready to move into the next phase. Typical gates are mobilisation complete, design signed off, build complete, user acceptance exit, go live readiness and hypercare exit. Each is judged by someone independent of the delivery team and recorded, so that a phase cannot be closed by the pressure of a date alone.
The client should own the integrated plan, with the integrator's delivery plan sitting inside it as a set of dependencies. The critical path on most ERP programmes runs through client activities, data cleansing, testing and training, which an integrator's plan treats as assumptions. When the vendor owns the only plan, it is optimised for the vendor's resourcing and commercial milestones, and business readiness becomes somebody else's problem.
Size it by roles rather than headcount: a programme director, a project manager, a PMO function, business owners for each process area and leads for data, integration, testing, training and cutover, each with real time released and their day job backfilled. On a mid sized finance implementation several of those roles can be part time or shared; on a multi entity rollout they are full time. The common mistake is under resourcing data and testing.
Where next
Tell us where the programme stands and who is on your side of the table, and we will show you what control would look like.