Giant queue outside the supermarket. They're operating a one-in one-out policy.

Kind of interesting throughput analysis problem.

Assuming the number of tills remain the same, does this increase the time it takes to shop or does it stay the same?

You'd think it increases it but are you just moving time queueing to pay, to time queueing to get in the shop?

I suspect that, as long as the time taken to shop on average is greater than the time taken to checkout, that it actually stays pretty similar.

Ok, I did a process diagram to check and it turns out that if the number of people allowed in is greater than the number of tills multiplies by the ratio of time browsing to time checking out then it doesn't make a difference.

It just splits the queueing into till and outside.

But that if it is less than that number then it does lower throughput and it takes longer to shop.

So, ideally you'd max out at these numbers being balanced so you hit the maximum throughput without allowing any more people in at once than necessary.

This doesn't take into account that the number of tills was cut in half in the semi-lockdown to spread people out a bit more.

So, process definitely is lowering throughput rather than insufficient demand or you wouldn't see queues at all.