Why End-to-End Visibility Is Still a Myth Without OCPM

How Object-Centric Process Mining sees the interactions between processes, the very place where most operational failures hide, and where true end-to-end clarity finally becomes possible

End-to-end visibility is the promise process intelligence was built on, and for most organizations it is still a myth. Not for lack of data, and not because the technology fails, but because nearly every approach was designed to follow one thread at a time, in a business where work never moves one thread at a time. A failure that lives inside a single order or invoice is easy to find. The expensive ones live between them, in the handoffs, and a tool that follows one object at a time looks straight past the handoff because it isn’t tracking the object on the other side of it.

That gap is easy to miss, because on paper everything looks fine. Dashboards are green, KPIs sit within target, audit findings are minimal, and yet orders are delayed, customer issues keep resurfacing, and approvals take longer than they should. When that happens, the instinct is to reach for a better metric or a new dashboard. It rarely helps, because the problem isn’t what’s being measured. It’s what isn’t being connected.

Take something as ordinary as a customer order. Its outcome depends on inventory, procurement, supplier performance, logistics, invoicing, and payment, and when that order goes wrong, every one of those teams checks its own data and finds nothing wrong. The order team sees a completed order. Procurement sees a placed PO. Finance sees a paid invoice. Each of them is genuinely right. The failure never belonged to any single process, it lived in the handoffs between them, in the one space none of them was measuring.

This is why end-to-end visibility remains more aspiration than reality for many businesses. Traditional process intelligence analyses individual flows extremely well, but the costliest delays and risks live in the relationships between those flows. Seeing those relationships is what turns process visibility into operational visibility, and it is exactly the problem Object-Centric Process Mining was built to solve.

Process Mining Changed the Way We See Operations — But It Was Never Designed to See Everything

Process mining has already earned its credibility, and that matters to acknowledge before going any further.

At its core, process mining works by anchoring every event to a single business object which includes a purchase order, an invoice, an insurance claim, a shipment and then tracing that object’s journey from beginning to end. Researchers call this the “case notion,” but in practical terms, it gave businesses something they had never really had before: a structured way to understand how work actually moves through an organisation.

Instead of relying on process maps built in workshops or on how employees remembered things working, organisations could finally reconstruct processes directly from the digital footprints left behind in their systems. For many businesses, that changed the conversation completely.

The impact speaks for itself. The global process mining market crossed $1.1 billion in revenue in 2024 and continues to grow at more than 30% annually, a reflection of the very real value the discipline has delivered across industries (Gartner, 2024)

For processes that revolve around a single object, it works remarkably well. A finance team following invoices can quickly identify operational bottlenecks such as approval delays, compliance gaps, and cycle times. A logistics team tracking shipments gains a clear picture of delivery performance. I have seen organisations uncover inefficiencies in weeks that had remained hidden for years.

That is precisely why process mining earned its place in the modern enterprise technology stack. But it is also where an important limitation begins to emerge because while businesses may be organised around departments, real work rarely moves through the organisation one object at a time. Orders interact with shipments, shipments influence invoices, invoices trigger payments, and payments affect supplier behaviour. Outcomes are increasingly shaped not by the lifecycle of a single object, but by the interactions between many of them, and that is where the traditional single-object view run out of road.

Real Processes Are Not Linear; They Are Knots

When you examine a real enterprise process closely, it almost never flows neatly through the life of a single object. A procurement process is not just a purchase order. It is orders, line items, goods receipts, invoices, and payments, each moving on its own timeline and each constantly tangled up with the others. Wil van der Aalst, who effectively founded this discipline, has a precise name for this reality. He calls it the “Rainbow Spaghetti.” This means to picture each object’s journey as a differently coloured strand: orders in one colour, invoices in another, shipments in a third, all woven together into a single, living knot. That tangle, messy and interconnected, is what your operation actually looks like beneath the surface.

When traditional process mining is turned on, that knot must pick one colour and flatten everything else out of view. The resulting map looks tidy and navigable. But it is tidy precisely because it has discarded the interactions between strands. And the cleaner the map looks, the more confident and the more misled we become about what is actually happening. This isn’t a story about organisations being complex in some abstract sense; it shows up directly in how the people running these tools experience them. A peer-reviewed study of 41 process mining practitioners, later validated through a survey of 24 experienced analysts, found that nearly two-thirds (64.71%) had run headfirst into this exact wall where real processes, built from several interacting objects, turned out to be far messier than the single-object process their tool was designed to show them (Zimmermann, Zerbato & Weber, “What Makes Life for Process Mining Analysts Difficult?”, Software and Systems Modeling, 2023.)

The implication is significant. Most process mining platforms were designed to follow the lifecycle of a single object, while real operational outcomes are shaped by the interactions between multiple objects at the same time. The result is a structural visibility gap: organisations can see individual process flows clearly, yet still miss the dependencies that drive performance, delays, and risk. Until those interactions become visible, what many organisations call end-to-end visibility remains only a partial view of how work actually happens.

The Failures Almost Always Live at the Seams

A recurring observation across operational transformation initiatives is that the failures that survive a clean audit almost never live inside one well-measured process. They live at the handoffs, at the seam between procurement and finance, at the collision point between a workflow approval and an inventory event, and at the intersection of order timing, item availability, and logistics scheduling.

Consider a supply chain scenario most operations leaders will recognise immediately: An On-Time-In-Full (OTIF) failure, where an order arrives late, incomplete, or both, generating escalations, damaging supplier relationships, and landing in the board report. No single object caused that failure. It emerged from how orders, items, warehouse picks, delivery packages, and dispatch schedules interacted across their separate timelines, and each one of those, measured individually, might show a perfectly healthy dashboard. The breakdown is relational, living between the objects rather than inside any one of them. At the Process Excellence Network’s 2025 event, Wil van der Aalst used exactly this scenario to illustrate that finding where an OTIF failure originates requires analysing all involved objects together, not in isolation.

The same principle shows up just as clearly inside finance operations. One AP team I worked with had spent months trying to speed up their invoice processing, only to discover the real bottleneck was sitting upstream in how approvals were structured. They had been optimising the wrong step entirely because their tools could only ever show them one step at a time. Ardent Partners’ AP Metrics That Matter research found that best-in-class AP organisations process a single invoice in 3.1 days, while others take up to 17.4 days to complete the same task (Ardent Partners, 2025). Gaps that wide rarely have a single-object explanation, and often, the root cause sits in the interactions between processes rather than within any individual process itself.

OCPM Does Not Add a Feature to Process Mining — It Changes What You Can Actually See

Object-Centric Process Mining (OCPM) is often introduced in language that makes it sound more complicated than it is, so here is the plain version.

When comparing object-centric process mining vs traditional process mining, the core difference is this that traditional mining forces every event to belong to one case which includes one order, one invoice, one claim. Analysts must choose which object to follow and discard the rest. OCPM lets a single event connect to many objects at once. An invoice approval can simultaneously link to the purchase order, the vendor, the line items, and the payment terms, all together.

This is made possible by a standard called OCEL 2.0, developed by RWTH Aachen and TU Eindhoven, which does something deceptively important: it records not just which events touched which objects, but how those objects relate to and influence each other throughout a process (OCEL 2.0 Specification, 2024). The connections themselves become something you can finally analyse, rather than something that quietly disappears before the analysis begins.

For any organisation serious about AI, this matters well beyond process diagnostics. As van der Aalst argues in his research, machine learning models achieve meaningfully higher accuracy when trained on object-centric event data rather than flattened single-case logs, because the training data more faithfully reflects how work actually causes outcomes (van der Aalst, “No AI Without PI,” 2025). OCPM is not a separate investment from your AI strategy; it is the honest data foundation your AI strategy needs to stand on.

The Question Worth Taking Back to Your Organisation

This is the most productive question any operations or finance leader can sit with right now.

Where do your most persistent, most stubborn operational problems actually live? If you are candid about it, most of them are not inside a single, well-measured process. They are sitting at the seams: the points where one process hands off to another, where data converges across object types, where no single owner and no single dashboard hold the complete picture. And those seams have been invisible to most of the tool’s businesses have been buying for the past decade.

The more organizations pursue end-to-end visibility, the more they are likely to discover that their most costly delays, risks, and inefficiencies do not originate within individual processes. They emerge in the relationships between processes. Understanding those relationships is not simply the next evolution of process intelligence. It is the difference between seeing how work flows and understanding how the business actually operates.