Queuing Theory
A long hold time does not necessarily mean too much demand. I think a lot about queuing theory. Today, it’s mostly about contact centers, but in the old days it was basic voice engineering. Most telecom engineers today missed all the Erlang theory.
Queuing theory can be applied to almost everything. I note the different delays in lanes on the road, the different lines at the grocery checkout, and my favorite the boarding and unboarding of planes (they don’t even pretend to be optimal).
Consider a healthcare system where the wait averages 8~10 weeks to see a specialist. Month after month, year after year there is a 8~10 week wait to see a specialist. The fix here is pipeline optimization, and it’s easy.
If there is a relatively constant 8~10 week wait to make an appointment with a specialist, this means that customer requests are arriving at about the same speed as they are being fulfilled. The flow of clients into the “pipeline” equals the flow out. It actually means that you have enough supply (doctors in this case) to meet demand.
The 8~10 week delay is caused by a one-time backlog. If that one-time backlog didn’t exist then you would be able to serve new clients quickly, perhaps in just a week or so. And, from that point forward you would have a 1~2 week scheduling lead time. Or less.
Pipeline optimization says that if the delays are not getting longer, you have a match between new requests and fulfillment capacity, so the problem is your own bureaucracy which is easier to fix than hiring more doctors. What is flowing into the pipeline matches what is flowing out.
The optimal point for a business to operate at is so that the resources are 80% to 100% utilized and the queue length is not increasing. If the queue is constant, but long, it is not an issue with resources or customer load. It is a management problem.
In my healthcare system it is always 8~10 weeks to wait to see a specialist. So the problem is one of middle-management accepting (or wanting) a long queue for whatever political purpose.
This is universal: if the queue/wait time is constant, the problem is not being understaffed.