The Boundary Problem in Supply Chain Data

Utilities and EPCs are investing heavily in internal systems, and for good reason. The systems and tools available today can organize, connect, and analyze an organization’s information in ways that were not practical a few years ago. It is worth being precise about what they can and cannot do with supply chain information, because the limitation is not a matter of features. It is a matter of where the data comes from.

Every enterprise system has a boundary. Inside that boundary, the organization controls the data model, the business rules, the user permissions, and the timing of every update. An ERP, a project management platform, a data warehouse, or an intelligence layer built across all of them works well inside the boundary for exactly that reason. The system is the system of record for information the organization creates, and anything that changes is changed by someone with access to it.

The supply chain sits outside that boundary. Lead times, pricing, ship dates, change notices, substitutions, and allocation decisions all originate at a manufacturer, and the organization has no read access to the manufacturer’s systems. That information has to cross the boundary before it can be used, and the channels it crosses on are the same ones the industry has used for decades: email, PDF attachments, spreadsheets, and phone calls that get written down afterward.

Technically, these are unstructured channels. A quote in a PDF is a document, not a record. The fields that matter, the unit price, the quantity, the quoted lead time, are embedded in a layout that was designed for a person to read, and every supplier uses a different layout. Before any of that can be loaded into an internal system, somebody or something has to identify the fields, interpret them, and map them to the internal schema.

Once that happens, what the internal system holds is a point-in-time capture. It reflects what was true on the day the document arrived. The manufacturer’s production schedule continues to move, and the internal record does not move with it. The delay between a change at the source and the same change appearing in the organization’s system is data latency, and in supply chain operations it is measured in days or weeks, not seconds. During that interval the internal system is accurate about the past and silent about the present.

Closing the gap requires one of three things. A supplier sends an update and somebody keys it in. Somebody on the project team notices an inconsistency and tracks down the answer. Or the organization builds extraction tooling to read inbound documents automatically. Each approach is repeated for every supplier, in every format, on every project, and each is only as current as the last document received.

What this means in practice is that the quality of an internal system sets an upper limit on what the organization can do with supply chain information, and the quality of the inbound channels sets a lower one. Most of the investment in this industry has gone toward raising the ceiling. The floor has stayed where it was.

‍