If paid media can see the last unit, you will sell the last unit to a stranger while a wholesale order or a VIP waitlist still needs it. Zero is an ops number. Paid needs a buffer.
Questions that bring people here
The buffer argument is far easier to win with a number attached: see how to estimate variant-level ad spend and waste.
Why zero is too late
Between “1 left” and “0 left” sit catalogue lag, checkout races, and creative that still looks fully available. Buffers buy time for the bridge to catch up. They also protect sizes that keep the size curve honest.
This is the sibling of days of cover: cover says how long you can sell; buffer says when paid must step back.
Starting points (steal, then measure)
| Category shape | Starter buffer | Notes |
|---|---|---|
| Apparel sizes | 2 to 5 units per size/colour on heroes | Protect the curve; dead 10s kill conversion |
| High AOV hardgoods | 1 to 2 units or 2 days cover | Oversell is expensive |
| Clearance | 0 to 1 unit | You may want to sell the last unit on purpose. Be honest on-site. |
| Flash drops | Policy-led | If the story is scarcity, creative and page must match |
Use the free days of cover tool when you think in days instead of units.
Encode it or it did not happen
A buffer in a Notion doc is not a buffer. It must hit:
- Feed / channel availability rules, or
- Tags that product sets respect, or
- Explicit set filters on inventory
Otherwise media will keep spending while ops “meant to” hold stock.
One shared number beats three clever people with three private rules.
Buffers and lag
If catalogue lag is 40 minutes, a buffer of one unit on a fast seller is pretending. Raise the buffer or fix lag. Measure lag with the lag test. Variant delivery still needs the map in DPA OOS.
Worked numbers
Hero tee, size M, sells 8 units/day. Buffer of 3 units is under half a day. With 40 minutes of catalogue lag, you will still oversell under spike traffic. Either raise buffer to ~10 units (a bit over a day) or fix lag. Pick consciously.
Clearance colourway with 4 units left and a “final sale” story can use buffer 0 if the page is honest and support is ready. Different SKUs, different rules. One global buffer for the whole catalogue is lazy.
Politics
Finance wants zero buffer (max sell-through). Brand wants infinite buffer (zero risk). Paid wants a number that keeps learning healthy. Write the tradeoff once. Revisit monthly with refund and lag data, not with whoever shouted last.
Questions people actually ask
Is a buffer the same as safety stock for wholesale?
Related idea, different audience. Paid buffer is specifically what you refuse to advertise into.
What number should I start with?
Many apparel brands start with 2 to 5 units per size/colour on heroes, or 1 to 2 days of cover. Tune with refund and lag data.
Won't I leave money on the table?
You leave some. You also leave fewer apology emails and fewer trained bad conversions.
Where do I implement the buffer?
In feed rules, channel settings, or product tags that exclude from paid sets. Write it where both ops and media can see it.
how do i avoid running out of stock on my top-performing products in the middle of a meta catalog ad campaign?
Compute days of cover per variant, not per product, and withhold the last few units of your converting sizes from ads before they hit zero, so organic and repeat orders still ship. Set the buffer from your catalogue lag and click-to-purchase delay, and cap scale on any variant under your cover threshold. The maths is on this page.
If a sold-out size is the problem: Ralph reads every size against every campaign overnight, finds the ones still paying for an empty shelf, and stages the exclusion with a revert for restock day. You say yes. See Ralph · How Ralph handles out-of-stock sizes · Docs