Is In-House Production or Outsourcing Better for On-Time Delivery?
Key takeaways
- The make-or-buy question is decided here by the goal of on-time delivery.
- If a specialist at the bottleneck is absent, such as the electrician in one work step, the whole order stops.
- Classic planning tools calculate with ideal capacities and barely capture these time delays.
- Driver trees with sequence and time effects show when an extended workbench secures on-time delivery.
Short episode, clear question: should I produce an order in-house or outsource it, with the goal of meeting on-time delivery? Classic planning tools calculate with capacities in an ideal state. What they often don't capture: the time delays caused by a small bottleneck, for example when the specialist at the fourth work step is absent.
Goal: on-time delivery
We're starting a small series: individual questions, answered briefly, best practice from industry. The first question: in-house production or outsourcing? A well-known topic that we look at under one question: should I produce an order in-house or outsource it, with the goal of meeting on-time delivery?
To break down the problem, we first ask: what is the goal, what is currently the most important KPI? Here it is on-time delivery. Contribution margin is also important, but we think from the value, and here the value lies in on-time delivery.
The bottleneck: one specialist
What planning tools often don't capture well are the time delays caused by a small bottleneck. An example: a metal part needs four work steps, and at one of these steps we need a specific specialist, let's say an electrician. He is the bottleneck in the entire system.
If he is sick, it's not simply one person missing for a few hours. The order stops at this point, because without him the electronic component can't be connected to the metal part and the assembly can't be completed. A specialist who is supposedly only missing for two hours or two days thus has a big effect on on-time delivery. Then we have to decide: do we give this step to an extended workbench on that day, or generally in planning, because we know there is a bottleneck here? Or do we keep it in-house?
Time dependencies instead of ideal capacity
Unlike the classic planning tool, we don't just look at capacities in an ideal state. Put simply: the electrician needs ten minutes per order, manages x units per shift and still has an hour of buffer, so it will probably be fine.
What is usually not captured in legacy planning tools: what happens to the rest of the company, the other orders, on-time delivery and contribution margin when this one bottleneck shifts? That requires a time-dependent component. We solve this by simulating driver trees that capture not only capacities but also sequence and time effects.
The key message: instead of only asking how often an order fits into the electrician's shift in an ideal planning world, we include the dependencies between work steps in planning. That way you can counteract immediately in daily operations when something happens. Because by Monday at 9:17 am, ten things have already changed since planning on Friday.