Ah, that master data that’s out of date!

In the age of AI, where data reigns supreme, the feedback from companies that shared their experiences was unanimous: the quality of master data is a critical issue to address when implementing a demand-driven approach, or more broadly, a planning solution.

This was a central topic of discussion at our user conference.

No company is immune to this issue. The quality of technical data is cited as a prerequisite for success.

ERP is rarely a reliable "out-of-the-box" source

Surprising? Yes and no.

Surprising, yes, because it’s not new: this data has been at the heart of ERP systems, deployed since the 1990s. By now, we should be well-versed in it!

Not surprising, when you consider that ERP systems are rarely actually used for planning, scheduling, and procurement—tasks that are often shifted to Excel spreadsheets and the individual expertise of our planners.

If the data isn’t used, it’s no secret—it isn’t maintained.

Feedback shows that the data in the ERP is either missing, obsolete, or not structured in a way that actually enables planning.

Procuring components without actually knowing supplier lead times? Far from being the exception, this is very common!

Items without planner codes (who handles that?), minimum and order multiples, production time, resources used, bills of materials, scrap rates… in fact, the ERP system we hoped would serve as a reference database is rarely a source of ready-to-use data.

Should we wait until we have perfect data?

The companies that spoke at our user conference initially took different approaches in this regard.

Some advocate a top-to-bottom audit before getting started: for example, systematically reviewing and correcting lead times, MOQs, etc. - by reaching out to the supplier base with the help of the purchasing department. This is a sound approach, but it’s tedious and can delay implementation by several long weeks.

A consensus is emerging among our clients: it’s better just to go ahead and load the data into Intuiflow as-is, because any anomalies will stand out like a sore thumb anyway. Master data issues “surface very quickly in DDMRP”—the method acts as a brutal revealer of existing gaps.

This is notably the message shared by Groupe Atlantic, Greif, Hutchinson, and the Sicame Group:

Load the available data, let the model run, let the problems surface, then fix what really matters. This “load and learn” approach has two advantages:

• It removes the psychological barrier of “our data isn’t ready yet”
• It prioritizes corrections based on actual impact on planning, rather than theoretical completeness

This is a well-known phenomenon when implementing an improvement initiative: the pursuit of perfection is the enemy of the good—get started and address issues by priority according to the Pareto principle.

This approach carries a risk: it may cause users to doubt the relevance of the planning solution’s recommendations. To avoid this, they must be involved in the design process and fully understand the rationale behind the calculations and recommendations. In this way, they become active participants in ensuring the system’s reliability and its future maintenance under operational conditions.

Migration to Intuiflow forces a complete overhaul that the ERP had never required.

It will never be perfect!

Remember that the data used in the supply chain is mostly estimates. Lead times are estimates. A setup time is an estimate. Capacity is an estimate. Average daily demand is an estimate. A scrap rate is an estimate. Don’t strive for perfection; strive for the relevance of the orders of magnitude!

Drift is the real enemy in the long run

Greif, with seven years of experience and 70 sites deployed, offers the most valuable lesson over time: data degrades silently. Without proactive monitoring, data drift can go unnoticed for weeks. Their concrete recommendations:

• Implement exception reporting from Day 1
• Establish a schedule for reviewing data at predefined intervals
• Conduct regular post-implementation audits

Human Governance

Data discipline does not maintain itself. It requires defined owners, a documented update schedule, and formalized rules - not tacit practices. In a global environment involving dozens of sites and possibly heterogeneous ERP systems, it is crucial to establish shared rules to avoid uncontrolled local decisions.

The underlying message common to all cases: master data is not a one-time project requirement to check off a box; it is an ongoing operational process - to be launched pragmatically and maintained rigorously!