When a Supplier Discontinues a Bestseller

Price increases announce themselves. Discontinuations mostly don't. The line that earned you the most money last quarter is simply not in this month's file, and unless something in your process is watching for absence, you will find out six weeks later when a customer orders it and the purchase order bounces.

How it actually shows up

There are five ways a supplier tells you a part is finished, roughly in descending order of how much warning they give you:

• An email or bulletin. The polite version, usually with a last-order date and a replacement part number. Enjoy it when it happens.
• A status column in the price file. D, DISC, OBS, NLA, EOL — whatever this supplier's convention is. Present in the file, easy to miss if your import only reads the price column.
• A supersession. The old number is still listed, but a replacement column now points at a new one. This is not a discontinuation in the sense that matters — the demand continues, under a different number. In some categories it is routine; in auto parts it is constant.
• The line disappears. No status, no note. Last month's file had 4,902 lines; this month has 4,871. Nobody mentions the 31.
• The price does something absurd. A 60% jump with no announced increase, or a minimum order quantity that leaps from 1 to 250. That is frequently a last-time-buy price on remaining factory stock, priced to discourage anyone who isn't serious.

The first 48 hours

The order matters here, because step two frequently cancels the rest.

1. Confirm what kind of event this is

Three different things get reported to you as "discontinued", and they demand three different responses:

• Superseded — a replacement part exists and covers the same demand. Your job is a catalog and cost update, not a stocking decision.
• Discontinued by the manufacturer — nobody will make it again. Whatever exists in the channel is all there will ever be.
• Dropped by this distributor — the manufacturer still makes it; your supplier stopped carrying it. This is a sourcing problem, not a product problem, and it is the best of the three outcomes because someone else has the part.

One message to your rep resolves all three: is this superseded, discontinued at the factory, or dropped from your line card? Ask for the last-order date and the replacement number in the same message.

2. Get the replacement's real cost

If a replacement exists, its cost is the only number that decides everything downstream. It is almost never the same as the old one, and the direction is predictable.

3. Count your exposure before you decide anything

Units on hand, units on open purchase orders, weekly velocity, and the weeks of cover those imply. Then the same for anything that depends on this part — kits, bundles and assemblies where a discontinued component quietly kills a parent SKU that is still selling fine.

4. Decide the last-time buy

The one genuinely hard decision, worked through below.

5. Reprice what you already own

Your existing stock is now a different asset than it was last week. It stops being replenishable inventory and becomes a finite one.

6. Leave the catalog alone until 1–5 are done

Deleting the product page is the one irreversible move available to you, and it is almost always premature.

The last-time-buy math

Take a part that is discontinued at the factory with a replacement available at a higher cost. Real numbers, one SKU:

Margin is (price − cost) / price throughout: (32.95 − 18.40) / 32.95 = 44.2% today, (32.95 − 23.10) / 32.95 = 29.9% once you are buying the replacement. Holding 44% on the replacement would need a retail of 23.10 / (1 − 0.44) = $41.25, which is a 25% price rise on a bestseller. That is the pressure that makes people over-buy.

The naive version of the calculation says buy enormous quantities. Carrying a unit costs 0.22 × $18.40 = $4.05 a year, or about $0.078 a week. Against a $4.70 saving, a unit pays for its own storage for roughly 60 weeks. At 14 a week and an average holding time of half the batch, that maths out to about 1,690 units — more than two years of sales — before carrying cost alone kills the deal.

So model the decay instead. A defensible shape for a superseded part: full velocity for a quarter, then roughly halving each quarter after, with a long thin tail. That gives about 342 units in year one (182 + 91 + 46 + 23) and perhaps 80 more over the following year before it is not worth the shelf space — call it 420 units of remaining lifetime demand.

Now the same decision at three sizes. Carrying cost is unit-weeks held × $0.078; dead stock is assumed to clear at half of cost.

Three things fall out of that table. The optimum sits at roughly your honest estimate of remaining lifetime demand, not above it. The curve is asymmetric — buying half of what you should costs you $395 of foregone saving, buying 67% more than you should costs $4,280. And the whole exercise is worth $1,150 on a bestseller, which is real money but not enough to justify the cash risk of guessing high.

Repricing the stock you already own

The instinct is to leave the price alone. That is a mistake, and it is easiest to see by asking what the units on your shelf are worth rather than what they cost.

Once the replacement lands at $23.10, the market price of that function goes up. A store holding the old part at $32.95 is selling a finite asset at a price set by a cost that no longer exists — and every unit sold at the old price is one that can't be sold later at the new one. If the replacement needs $38.50 to hold a 40% margin (23.10 / 0.60), then moving the remaining old stock to $38.50 as well earns (38.50 − 18.40) / 38.50 = 52.2% on it, and stops your last-time buy from quietly subsidising customers who would have paid the new price anyway.

Two constraints on that. If the category carries advertised-price rules, check them before touching anything — MAP is set by someone else and does not care about your supply situation. And if you sell to trade accounts on a fixed price list, you are contractually somewhere else entirely; that case is covered in B2B trade pricing and cost increases.

The opposite case — no replacement at all, demand that will simply evaporate — argues for the reverse. Clear it while there is still a reason for anyone to buy it, and take the margin hit early rather than the write-off later.

Don't delete the product page

A discontinued bestseller has accumulated something you can't rebuild: search rankings, inbound links, reviews, and a URL that other people's pages point at. Deleting it throws all of that away for the sake of a tidy catalog.

• If a replacement exists, keep the old page live while stock lasts, then redirect it to the replacement. The demand that arrives at the old URL is the demand for the new part.
• If nothing replaces it, redirect to the closest category page rather than to the homepage. A visitor who wanted a specific part is better served by a shelf of similar parts than by your value proposition.
• Either way, mark the SKU as discontinued in your own data with a date, and keep the cost history. A year from now, "why did this category's margin fall in Q3" has an answer, and the answer is in that record.

Making it visible earlier next time

Everything above is a response. The cheaper work is detection, and it is four rules on top of a price-file diff you probably already run:

1. Report missing keys. Lines present last time and absent now, listed explicitly, every file. Sort them by trailing revenue, not alphabetically — 31 dropped lines are noise until you see that one of them did $18,000 last year.
2. Read the status column. If the file has one, map its codes once and treat a status change as an event of the same weight as a price change.
3. Flag structural jumps. A minimum order quantity that multiplies, a pack size that changes, or an increase far outside the file's normal distribution. Any of these can be a discontinuation wearing a price change's clothes; the pack-size version is worked through in MOQ and pack size economics.
4. Rank by your exposure, not by the supplier's. A discontinued line you sell twice a year is a footnote. The same event on a top-50 SKU is a week's work and a five-figure cash decision, and the only way your process knows the difference is if it joins the diff to your own sales data.