The short answer
A preorder converts uncertain future availability into a committed position, at the cost of tying up capital before the product exists. The discipline is separating non-binding interest from a binding order, and sizing the binding stage against cash you can afford to have parked.
The products that sell hardest are frequently the ones with the least availability after release. A store that waits to see how a release performs before ordering is competing for whatever is left, usually at a worse cost and often at no availability at all. Preordering is how a store converts uncertain future access into a committed position.
The cost is capital and time. Money committed to a release that arrives in four months is money not available for the three releases in between, and for a store operating on working capital rather than a credit line, that trade-off is the entire decision.
It follows that preordering is not a strategy in itself. It is a tool that works when applied to the specific products your store genuinely sells, and destroys a quarter when applied broadly out of a fear of missing out.
A well-structured preorder process has three distinct stages, and conflating them is where disputes come from.
The first is indicated interest: non-binding, before the distributor has committed capital. You say what quantity you would take at an indicated price. Nothing is owed by anyone. This stage exists so the distributor can aggregate demand and take a real number upstream.
The second is a binding preorder: the distributor has opened the product for order and you have requested a quantity. You are now committed subject to the terms, and so is the distributor to the extent it can be.
The third is allocation confirmation: the distributor tells you, in writing, what you are actually getting. Between stages two and three the quantity can change, because the distributor’s own supply can change. A process that does not distinguish these stages will eventually tell a store it is getting something and then not deliver it.
The sizing question is not "how much do I want" but "how much can this capital be unavailable for, and what am I giving up". Three tests are worth applying before committing.
The first is the cash test. If this money is unavailable until the release arrives, what does the store miss in the interval? A store that preorders itself out of the ability to restock its consistent sellers has made a bad trade regardless of how the release performs.
The second is the evidence test. What in your own sales history says this product sells here? Not the category — the specific product line, in your store, in previous years. A store’s own data is the only reliable input, and a release that performed nationally and not locally will do the same thing again.
The third is the downside test. If this release underperforms, how long does this inventory sit? Sealed product does not spoil, but capital parked in a slow release is capital not compounding through the calendar. A quantity you would be comfortable holding for a year is a quantity you can preorder.
Cancelling a confirmed allocation is more expensive than it appears on the invoice. From the distributor’s side, that product was committed against your request and withheld from an account that would have taken it, and by the time you cancel, redistribution is harder and often later than release.
Distributors respond by weighting cancellation history in future allocation. That is not punitive — it is the same logic that produced the allocation in the first place. Product goes where it is most likely to reach a retail floor, and an account with a cancellation record is, on the evidence, less likely.
The practical implication is to request the quantity you are confident about rather than the quantity you hope for. Requesting less and being disappointed is a cheaper error than requesting more and cancelling.
You should be able to tell, at any moment, which stage each of your outstanding requests is at, what quantity is attached, at what price, and what the current expected arrival is. If any of those four is unavailable, you cannot plan cash against it.
You should also expect to be told when something changes. Upstream dates move constantly and no distributor controls that. What a distributor does control is whether you hear about it promptly or discover it when the product does not arrive. An expected date a distributor already knows is wrong is worse than no date at all, because you have planned against it.
Indicated interest is non-binding — you state a quantity you would take at an indicated price, before the distributor commits capital. A preorder is a binding request against product the distributor has opened for order. Allocation confirmation is a third, separate stage that comes after both.
Only as much as can be unavailable until the product arrives without preventing you from restocking your consistent sellers. There is no universal percentage, and any figure quoted as one is ignoring your cash position, which is the variable that actually decides it.
Terms vary by distributor and should be published. Beyond the direct terms, cancellation history is weighted in future allocation at most distributors, because a cancelled allocation is product that could have gone to an account that would have sold it.