We assume ERP projects fall apart when a consultant leaves. What I keep seeing in the field tells a different story: projects can collapse just as quietly with the same consultant in the chair. In a 312-employee automotive parts manufacturer in Eskişehir, a 16-month ERP engagement ran with full team continuity throughout. No one quit. No one was replaced. By month 14, the finance module had to be rebuilt from scratch — because a design decision made in week one, how the chart of accounts would map to the existing structure, had never been written down. It lived in memory, not in documentation. The problem was never the consultant. It was the architecture around the consultant.This leads to a thesis that most ERP sponsors resist: consulting continuity in long-term projects is a structural problem, not a staffing one. Keeping the right person on the engagement for the full duration is necessary but not sufficient. Without a working protocol that makes knowledge independent of any single person, even the most stable team is fragile. The owner of the business says ‘don’t let the consultant leave.’ The consultant stays. The project still breaks down at a knowledge seam. Because knowledge transfer isn’t an event you schedule at project close — it’s a discipline that has to run every single week.The first structural condition is a decision log with teeth. Before the first configuration session, establish a simple standard: every material design decision must be recorded with three data points — who made it, which business requirement drove it, and which alternatives were considered and rejected. This sounds obvious. In practice, it almost never happens in Turkish SME ERP projects. Progress reports get written. Steering committee slides get produced. But the reasoning behind decisions evaporates. Four months later, a new team member or a scope change forces you to revisit a decision no one can fully reconstruct. In the Eskişehir case, four field sessions and two progress reports had been delivered before anyone asked ‘why was the chart of accounts mapped this way?’ The answer no longer existed in retrievable form.The second condition is a weekly knowledge transfer ritual — and it is not the same as a status meeting. A status meeting answers ‘what did we do?’ A knowledge transfer session answers ‘what did we learn and where did we record it?’ A construction firm in Ankara ran a structured Friday session throughout an 11-month ERP engagement, working through a fixed template: decisions taken, rationale documented, open questions flagged. The consultant on that project changed in month seven. The incoming consultant was functional within two weeks, because the preceding six months of decisions were accessible and coherent. That outcome wasn’t accidental — it was the product of a project lead who treated the Friday ritual as non-negotiable. The system wasn’t sophisticated. It was consistent.The third condition is developing a technically capable internal counterpart on the client side. This is the rule most commonly ignored in Turkey. The owner runs the ERP project directly with the consulting team. The finance manager defers all technical questions to ‘the consultants.’ The IT contact — if one exists — manages hardware and connectivity, not process design. When the consultant eventually steps back or changes, the firm discovers that no one inside the organisation truly understands how the system was built or why. The solution is simple to describe and hard to execute: within the first 60 days of the project, designate an internal ‘ownership coordinator.’ This person doesn’t do consulting work. But they attend every critical session, they ask questions, they understand every decision. By month eight or ten, they are the most knowledgeable person in the firm regarding the ERP. That investment costs little and pays repeatedly.A clear boundary needs to be drawn here, and I want to draw it honestly. These three conditions work in projects where scope is reasonably stable and management decisions don’t reverse on a weekly basis. Where scope creep is severe — where requirements shift faster than documentation can follow — no knowledge transfer protocol will save you. The knowledge you documented last Tuesday is already partially obsolete. In that environment, the prior work is to establish management commitment and route scope changes through a formal approval process. Knowledge transfer architecture only becomes meaningful once that foundation is in place. Attempting to document a moving target without fixing the movement first produces clean-looking records of decisions that no longer reflect reality.That automotive parts manufacturer in Eskişehir eventually delivered a working system. The finance module rebuild added 14 weeks and a meaningful budget overrun. The internal review attributed it to ‘technical issues during integration.’ From where I was sitting, the cost was entirely structural — the bill for a project that ran without a decision log, without a weekly transfer ritual, and without a client-side counterpart who truly owned the process. If you are managing a long-term ERP project and want a practical test, ask yourself this: if your consultant were unavailable tomorrow, what specific knowledge would you lose, and where would you go to recover it? If that question doesn’t have a clean answer, your project is still running on personal memory. The consultant may be sitting right there — and the fragility is still waiting.
This article was originally published in Turkish by Gökhan MERCANOĞLU on January 14, 2010. The English edition has been reviewed and edited by the author.