A promised delivery time used to stem from staffing: how many people were available to schedule orders, handle exceptions and inform customers when something went wrong. Whoever had more capacity could switch faster and thus promise a shorter delivery time. That was the playing field, and on that playing field a commercial manager knew roughly where he stood relative to the competition: staffing is visible, or estimable.
That assumption no longer holds everywhere. Where planning, order processing and customer updates are largely done by AI, delivery time is no longer a function of how many people are working, but of how well the system behind it is set up. Two companies with the same headcount can then promise completely different delivery times. That is not a future scenario: it is happening today, already at one company and not yet at another.
Behind a promised delivery time sits a series of tasks that falls into three categories.
Part of it AI can fully take over: combining inventory positions, production capacity and transport times into a feasible delivery date is calculation work, and calculation work is exactly where AI is systematically faster and more consistent than a planner doing this alongside other tasks.
Part happens with human oversight: in case of a disruption — a supplier that fails, an order that deviates from the pattern — AI can propose an adjusted delivery time, but someone approves or rejects that proposal, with reason. The speed of that approval determines how quickly the new commitment reaches the customer.
Part remains human work: the customer who calls because a delivery has been postponed twice wants an explanation and a judgment call that goes beyond pressing a button. How much of that conversation is still needed depends on how well the first two categories already work — the fewer disruptions go unnoticed, the fewer conversations of this kind are needed.
The difference between companies does not lie in whether they use AI, but in what is already connected. A planning system that is disconnected from inventory administration delivers little gain, however advanced the model behind it. A company where order data, inventory and transport planning already interlock can lay a planning model straight on top and adjust the delivery time downward. A company where that data sits in separate systems and spreadsheets must first straighten that out before there is anything to automate.
That explains why the shift is not happening at the same pace everywhere, and why two competitors in the same market with the same headcount can nonetheless promise a different delivery time. It says nothing about how many people work there or should work there; that is a separate consideration with its own legal requirements, which is not at issue here.
If delivery time shifts from a staffing question to a system question, what you compare competitors on shifts too. Asking how many people sit on planning says little anymore. More relevant is whether a competitor consistently meets delivery times, how quickly it adjusts in case of disruption, and whether that is already visible upfront in how confidently it dares to name a delivery date.
That is immediately one of the places where how you obtain usable information about competitors who don't publish figures becomes relevant: delivery time is one of the few dimensions that can be tested in practice, through your own orders or customer signals, even without a competitor sharing figures on it.
Delivery time is also not separate from other dimensions. A shorter delivery time accompanied by more errors is not a win; how those two relate to each other is addressed in how AI is changing the competition on quality. And a delivery time that is only short for standard orders says something different than a delivery time that also holds up across a broad product range, as discussed in how AI is changing the competition on product range. Anyone who mainly encounters recognizable promises about speed without distinction, in their own marketing and that of competitors, recognizes the pattern from why competitors all sound the same in their promises.
The question of whether delivery time at your company is a staffing question or already a system question is difficult to answer with a broad view; which work behind it can genuinely be taken over by AI is answered task by task with the work scan from FTE TO AI.
For the comparison with competitors, a first step is smaller: name where you believe you win on delivery time, and test with the free dimension check which of those claims can be defended with evidence. The full benchmark, with scores and an evidence matrix per dimension, is under construction.