The worst phone call in retail is the one where you have to ring somebody back and say the thing they bought is not actually there.
It happens like this. Saturday evening, two orders arrive within a minute of each other on the shop page. Both are for the last pair of size 40 shoes. Neither customer did anything wrong. Neither did you. The page said one was available because when the first person loaded it, one was.
At the counter this problem barely exists. Two people cannot both pick up the same box, because a box is a physical object and one of them is holding it. Online you are selling a promise, and promises can be made twice.
Why it costs more than the sale
Losing the sale is the cheap part. You refund, you apologise, it is over.
What you actually lose is the thing that took you a year to build. A customer who was told yes and then told no does not think "unlucky timing". They think this shop does not know what it has. And they are right, which is the uncomfortable bit. That impression is very hard to undo, and it travels: in a Kathmandu neighbourhood, or in a Facebook group for a particular hobby, one bad story about an order does more damage than five good ones repair.
There is a second cost that is easy to miss. Overselling turns your phone into a queue of apologies. An hour spent on three awkward calls is an hour not spent packing, buying, or photographing. It is the most expensive kind of admin because it also makes you feel bad.
Count the things worth counting
The instinct after being burned once is to count everything. Resist that.
A shop that tries to track stock on four hundred items will have four hundred wrong numbers within a month, because every sale that does not get recorded, every breakage, every item a cousin took, every miscount on delivery pushes one of them out. And a stock number you do not trust is worse than no number, because you start ignoring the warnings and then the warnings stop meaning anything.
Count the things where running out is expensive and where the quantity is small enough to stay honest. The shoes in size 40. The last three of a colour. The imported thing that takes six weeks to replace. Anything you sell by weight out of a sack, where "how many" is a fiction anyway, is usually not worth a number.
In Pasal each product can carry a count, and it is a per-product decision rather than a mode you switch on for the whole catalogue. That is deliberate. Most shops have twenty items worth counting and two hundred that are not.
Where the refusal has to live
Here is the part that matters technically, and it is worth understanding even if you never think about software.
It is not enough for the shop page to hide the Add button when something runs out. The page is a copy of the truth from a few seconds ago, sitting on somebody's phone, and two phones can hold the same copy at the same time. A page that stops you is a courtesy. It is not a rule.
The rule has to live where the order is actually written down. In Pasal an order that would take a counted product below zero is refused at the point of writing, not at the point of asking, and the customer is told to change the basket rather than being thanked for an order that cannot be filled. Both shopfront doors cap the quantity as well, and they add up the sizes and colours of the same product rather than capping each line on its own, because eight in stock should not go out as small eight, medium eight and large eight.
The page being helpful and the server being strict are different jobs. You want both, and if you only get one, you want the second.
The counter is deliberately allowed to oversell
There is one place Pasal does not stop you, and it is on purpose: the till.
If you ring up a sale at the counter for something the count says you do not have, it goes through. This looks like a bug and it is not. The person selling is standing there holding the goods. If the shelf has one and the number says zero, the shelf is right and the number is stale, and the last thing a shop needs is software arguing with a customer who is holding money out.
So the counter trusts you, and the internet does not. That is the correct way round.
Fixing the number without drama
Counts drift. That is not a failure, it is the nature of physical objects and busy Saturdays.
The habit worth building is not perfect accuracy, it is quick correction. When you notice the shelf and the number disagree, change the number then, while you are looking at the shelf. Not tonight, not at month end. Anybody who has tried to reconcile stock from memory a week later knows that the memory loses.
For the twenty items you actually count, a two minute walk along the shelf once a week is enough to keep them honest. That is a completely different job from a full inventory, and it is one that a shop can genuinely sustain.
What good looks like
You will still run out of things. Running out is not the failure. Selling something you have run out of is.
The shop that has this right does not have perfect stock records. It has accurate numbers on the small set of items where being wrong is expensive, a server that refuses rather than hopes, and an owner who corrects a count the moment they notice rather than adding it to a list.
That is enough to never make the worst phone call again, which is a low bar and a surprisingly rare achievement.
Turning a count on for the products worth counting is counted stock.