Returns are inventory: folding returned units into your reorder math
Restocking returned items quietly breaks your reorder math — sellable units re-enter stock weeks later. Net returns into your reorder points.
InventoryX is in development and not available today. ClearStock is the live Shopify app.
Every restocked return re-enters your Shopify inventory on its own clock — usually two to four weeks after the sale that created it. That lag means the two numbers your reorder math runs on, your sales velocity and your available stock, are each computed on figures that don't match what's physically happening on your shelf. Get the netting wrong in one direction and you over-order into a cash squeeze; get it wrong in the other and your bestseller runs empty in the middle of a lead time. This article shows you the three numbers you actually have, where the standard reorder point formula breaks, and how to fold returns in without a forecasting suite.
The three numbers your store actually has
Merchants tend to track one number: sales. With returns in play you have three, and the differences between them are the whole game. The illustration below uses a pet-gear store — say you sell no-pull dog harnesses, and your medium size is the workhorse of the catalog.
| Number | What it counts | Harness example (illustrative) |
|---|---|---|
| Gross demand | Units sold, before any refunds | 9.4/day |
| Restock inflow | Returned units that go back to sellable stock | 1.4/day, lagged 2–4 weeks |
| Net consumption | Gross demand minus restocked returns | 8.0/day |
The arithmetic behind that table: an 18% return rate (sizing is the usual suspect with harnesses) on 9.4/day gross demand means about 1.7 units/day come back. If 85% of returns are sellable again, restock inflow is 1.7 × 0.85 ≈ 1.4/day, and net consumption is 9.4 − 1.4 = 8.0/day. That 8.0 is the number your next purchase order has to cover. The other two are context, not buying signals.
Where the reorder math breaks
The standard formula is reorder point = velocity × lead time + safety stock (the reorder point formula walkthrough covers the mechanics). The question is which velocity goes in, and the two failure modes run in opposite directions.
Failure mode one: planning on gross demand. If you reorder against 9.4/day when your stock is only actually depleting at 8.0/day, you over-buy by 9.4 × 0.153 ≈ 1.4 units of pipeline cover per day of lead time. With a 21-day domestic lead time and 60 units of safety stock:
- Reorder point on gross: 9.4 × 21 + 60 = 258 units
- Reorder point on net: 8.0 × 21 + 60 = 228 units
A 30-unit gap on one SKU. Across a 40-SKU harness line with the same return profile, that's roughly 1,200 extra units sitting on shelves — cash that arrived because you let returned units count as demand twice: once when sold, again as the reason to replace them.
Failure mode two: netting returns against the wrong week. The tempting fix is subtracting this week's refunds from this week's sales. That breaks because of the lag. Restocked units from a late-October promo land in available inventory in mid-November, when the sales they belong to are three weeks old. Your monthly net velocity then swings: an artificially low October, an artificially high November. If you reorder on those swings, you under-buy right after every strong promo — precisely when the pipeline is emptiest — and over-buy the month after.
The fix for both is the same discipline: measure demand on a rolling 90-day window (long enough that return lags average out), count restocked returns as supply arriving, and feed net consumption — not gross, not weekly net — into the reorder point.
The lag problem, worked through
Here's what the lag actually does to a reorder cycle. Your reorder point on the medium harness is 228 (the corrected figure). It fires on November 3. Stock lasts until November 26; the PO lands December 12; you cover the gap. But during that window, restocked returns keep dribbling in — roughly 1.4/day × 39 days ≈ 55 units of unexpected supply that stretch your cover by about seven days beyond what the reorder point promised.
That sounds like a bonus. It isn't free money: it's inventory you already paid for and already sold once, and the danger is that it trains you to under-buy. "The reorder point fired late and nothing bad happened" is the most expensive lesson a returns-heavy store can learn, because the inflow depends on last quarter's sales, not this quarter's. Cut your ad spend, and the returns inflow shrinks with a delay while your gross velocity assumptions from the fat months linger.
Not every return belongs in available stock
When you refund an order in Shopify, the restock option returns the items to available inventory — it's the default path, as of November 2026, and it only applies when you track inventory (Shopify Help Center, "Refunding orders," accessed November 2026). That's the behavior to plan around, and it creates one specific trap: units that come back damaged, worn, or washed get restocked into available quantity anyway if nobody deselects the option at the counter. Those units are phantom stock — numbers Shopify shows and customers can't buy.
So the restock decision needs a gate before it reaches Shopify. Sellable returns go back into available stock; damaged ones go to a written-off or B-grade state and leave the sellable count. How returns physically move through your back room — quarantine, inspection, then restock or disposal — is its own workflow, covered in the returned-inventory processing guide; the point for this article is the sequence: inspect first, restock second, because whatever touches available quantity becomes an input to your reorder math.
A practical check: once a quarter, compare Shopify's available quantity for your top returners against a physical count. A widening gap is the signature of damaged returns being restocked by default.
How the return rate changes what you do
| Return rate | What it means for reorder planning |
|---|---|
| Under ~5% | Noise. Ignore it in the reorder point; net it in monthly reporting only. |
| ~5–15% | Net it in. Use net consumption in the reorder point on a rolling 90-day window. |
| Above ~15% | Treat restocked returns as a supply line. Track inflow per SKU, gate restocks behind inspection, and recheck your safety stock — it was sized against demand that partially doesn't stick. |
Those bands are operator rules of thumb, not benchmarks — the right trigger is when the gap between gross and net is big enough to change what you'd order. If you're not sure where you stand, pull units sold and units refunded by SKU for your last 90 days from Shopify's reports and compute the restock-adjusted rate per SKU; five minutes per top seller.
The Q1 returns wave
November and December sales generate a returns wave that lands in January — and January is exactly when you're deciding spring buys. A store that sold 2,800 harnesses in Q4 at an 18% return rate should expect roughly 500 returns, of which maybe 425 come back sellable. Those 425 units are the first line of Q1 supply, before any new PO.
The January discipline, then: before you place spring orders, count what's physically back on the shelf (not what Shopify said in early January, while the wave is still landing), net it against Q1 demand, and order the remainder — timing spring buys against the season, not the calendar, which is the retail buying calendar's job. The returns piece of the January reset is simply "don't buy what's already in the building, still arriving."
Common questions
Does restocking returned items increase available inventory in Shopify automatically? Yes, when you or your staff keep the restock option selected during the refund — it's the default and it puts units back into available quantity at that location. It only works on inventory-tracked items, and you can deselect it to handle restocking manually. As of November 2026 there's no setting that inspects condition first; that gate is manual.
Should I subtract returns from my sales velocity? Not week by week — the lag between the sale and the restocked return distorts both months. Instead, compute velocity on a rolling 90-day window and use net consumption (gross sales minus restocked returns) as the demand input in your reorder point. Weekly refund counts are reporting data, not buying data.
How do returned items affect safety stock? Indirectly. Safety stock buffers demand variability, and a high return rate adds a second source of variability — lumpy restock inflow on a lag. If your return rate is above roughly 15%, recheck your buffer after netting returns; some of what you were buffering against is actually supply that arrives late, not risk.
Do restocked returns distort my inventory value? They can, if damaged units get restocked by default: you're carrying available quantity that isn't sellable, which overstates both stock and value until it's written off. Gate restocking behind a quick inspection, and send unsellable units through a write-off or B-grade path so they leave the sellable count.
What return rate should change my reorder math? There's no universal threshold, but a useful test: if gross minus restocked demand changes what you'd order by more than a week of sales, start netting. For most stores that lands in the 5–15% range. Compute it per SKU — a harness line at 18% and an enamel-pins line at 2% shouldn't share a policy.
Next step
Pull units sold and units refunded per SKU for the last 90 days, compute restock-adjusted net consumption for your ten fastest movers, and compare each against the reorder point you're currently using. Where the gap exceeds about a week of net demand, recalculate — the reorder point formula walkthrough has the worked template. Then set a calendar note for mid-January, when the Q4 returns wave finishes landing and your spring buys should net against it.