Where is the order, and why is it late?

Lead time tracking rarely needs new systems or sensors. When you know the order number, the step and the timestamp for every step of an order, you can calculate how long each step took and where the order sat waiting. In most companies this data already exists in the ERP, the manufacturing execution system (MES) and the warehouse management system (WMS). It is simply spread across systems, and nobody connects it.

What does lead time tracking require?

Lead time tracking requires three pieces of information for every step of an order: the order number, the name of the step and the time it happened. These rows form a timeline for the order, showing how long each step took and how long the order waited in between. The same three-field structure is the foundation of process mining, and Microsoft's process mining tool, for example, requires exactly these fields.

An example timeline for one order:

The order spent eight days in the company but less than five hours in production. It waited more than two and a half days before production and almost four days ready to ship. Neither wait shows up in any report if each system only looks at its own step.

Is the data in your current systems enough?

Usually, yes. A timestamp is created automatically whenever someone confirms a step in a system: the order is confirmed in the ERP, a job is started in the MES or picking is confirmed on a handheld scanner. You do not need to collect timestamps separately; you read them from the systems that already hold them.

Two obstacles usually come up in practice:

1. Identifiers. The ERP uses an order number, the warehouse a delivery number and production a job number, so they have to be linked.
2. How steps are recorded. If production steps are confirmed at the end of the shift, the timestamp tells you when someone confirmed the step, not when the work happened. In that case the step can be read directly from the machine, and production machine data can be read from existing systems without opening the plant network.

Why are deliveries late when production is fast?

An order often waits between steps rather than within them. That is why lead time is not shortened by speeding up the work, but by finding the points where the order sits ready for the next step. Those points only become visible when the time between steps is calculated order by order.

Once the timelines are in place, tracking reveals three things:

• Variation that the average hides. If nine orders out of ten ship in two days and the tenth in two weeks, the average looks fine. The customer still remembers the tenth.
• Orders that are stuck right now. When you know the normal duration of each step, an unusual wait is noticed before the customer calls.
• The cause of the wait. When the customer, product group, shift or supplier is attached to the timeline, you can see whether waiting piles up in one place.

In logistics the same pattern shows up in the transport chain. In Metsä Board's outsourced transport, the carriers' data was limited to handover points such as arrival at a terminal. With sensors travelling with the shipments, the data also showed where and for how long shipments sat in interim storage. In Metsä Board's transport tracking the same data also pinpointed where the packaging was damaged. In most companies, however, tracking starts from data that is already in their systems.

Do you need a process mining platform, or is your own tracking enough?

When an order always goes through the same steps in the same order, your own tracking is enough: lead times are calculated from existing data and shown in a view the team already uses. A process mining platform is needed when orders follow dozens of different paths and you want to find out which deviations slow them down and why.

A sensible hybrid is often this: event data is collected in the company's own database, so the data stays under the company's control, and the analysis tool is chosen to fit each question. A lightweight view is enough for day-to-day tracking, and a process mining tool comes out when a specific problem needs a thorough investigation. The same principle works for integrations, where critical workflows belong in a managed integration and the rest can stay in a no-code tool.

Where should lead time tracking start?

Start with one product line or one delivery route, not with every process in the company. Choose 4-6 steps whose timestamps are already recorded, pull a few months of history and agree on one alert threshold. A limited start keeps the work small and quickly shows whether tracking is worth expanding.

1. Choose a question. For example: why did on-time delivery fall short of target last quarter?
2. List the steps and the systems where their timestamps are stored.
3. Check the identifiers. Does the same order number run through every system, or do the numbers need to be linked?
4. Calculate from history the normal duration and variation of each step.
5. Agree on an alert. For example, a Teams message to the person responsible when an order has sat in the same step for more than 48 hours.

Tracking is also a good starting point for automation. When you can see where an order waits, you also see which manual step is holding it up: an order confirmation waiting for someone's email, or a shipment waiting for a hand-typed waybill. The Automation Assessment goes through these steps and calculates the value of automating them in euros.

← Back to homepage