When to kill a product (and clear the stock)
Every product on your store spends your cash, your storage, and your attention. A slow one spends what a winner could use. Killing it isn't failure, it's freeing resources.
Judge on real criteria: contribution after returns and support, velocity, and review quality. Then run a clean sunset instead of hoping. AI can score your catalog and build the markdown plan; you make the call. The honest math usually says kill it sooner than it feels.
Fastest path: one prompt, end to end
🤖 AI prompt — paste into ChatGPT / Claude
You are a merchandising analyst. Use MY data only.
Attach my product report with, per SKU: units sold last 90 days, revenue, unit cost, return rate, and (if I have it) support tickets tied to it.
My target contribution margin %: [%] On-hand units + storage cost if any: [# / $]
Do this:
1. For each product, compute contribution AFTER returns and support load (not gross revenue). Rank from strongest to weakest.
2. Flag kill candidates using: contribution below my target, a declining velocity trend, OR a bad review pattern (recurring complaint). Say which criterion each one fails.
3. For each kill candidate, lay out the sunset ladder: stop reorder now, markdown ladder (e.g. 15% then 30% then 50% over 3-4 weeks), bundle-out with a strong seller, then donate/liquidate the tail. Give the on-hand units a target clear-out date.
4. Show the honest carrying-cost math: what this dead stock costs me per month in cash + storage if I hold vs clear.
Do not invent numbers. If return rate or support data is missing, list it and ask, do not estimate it.
Output: contribution ranking + kill candidates with the failing criterion + per-product sunset ladder + carrying-cost math.
Or do it in 4 steps
- Judge by contribution after returns and support, not revenue. A product can look like it sells and still lose money once you subtract returns and the support time it eats. Compute true contribution per SKU. A high-return, high-complaint item is often your worst product wearing a good-revenue mask.
- Read the velocity trend and the reviews. One slow month is noise; a steady decline over a quarter is a signal. Pair it with the reviews. A recurring complaint (sizing, breakage, 'not as pictured') means a real defect that discounts won't fix. Trend down plus bad reviews is a clear kill.
- Run the sunset ladder in order. Don't just delete it, recover the cash. Stop the reorder first. Then a markdown ladder: ~15%, then 30%, then 50% over a few weeks, clearing at the best price the market will pay. What won't sell, bundle with a strong seller. The tail, donate or liquidate; a bookable write-off beats a shelf you keep paying for.
- Do the honest carrying math. Work out what the dead stock actually costs per month: tied-up cash, storage, and the attention it steals from products that work. Holding 'in case it sells' almost always loses to clearing now and redeploying the cash. The number usually says kill it sooner than your gut does.
Worked example (labeled): a SKU showing $4k/quarter revenue looked fine until the report subtracted a 22% return rate and the support hours it generated. Real contribution was negative. It failed on contribution and on a recurring 'runs small' review pattern.
Sunset: reorder stopped; 15% then 30% then 50% over four weeks cleared 80%; the last 20% bundled with a bestseller; tail donated for the write-off.
That freed roughly $1,800 of cash and a chunk of weekly support time, both redeployed to the top seller. Killing it was the growth move.
A product earns its place on contribution, trend, and reviews, not on revenue; when it fails, run the sunset ladder and redeploy the cash, because carrying dead stock is a cost you're choosing to pay.