From a Q4 stockout to a 2.4× peak season the following year

The brand's season runs from October to February, and in the year before we started it ran out of its two hero ASINs on 6 November. By the time stock landed in mid-December the listings had fallen from the first page to the third, the Buy Box had been handed to resellers, and the advertising that was still running was paying to send shoppers to an unavailable product.

The cost was not just the lost quarter. Amazon's ranking is a memory: the sales velocity that disappears in peak is exactly the velocity the algorithm uses to rank the listing next season. The brand entered the following spring with lower organic rank than it had eighteen months earlier, and read that as “the category got more competitive”.

We planned the next peak backwards from the week the brand ran out.

• Forecast built on last year's true demand, not on last year's sales — the stockout weeks were modeled as lost demand rather than treated as a slow period.
• Rank protection instead of blanket ad cuts — a defined floor of spend on the hero terms through peak, with bids managed to hold position rather than to hit a monthly ACoS target.
• Depth behind the heroes — secondary sizes and a bundle promoted into real listings, so demand had somewhere to go if a hero ran tight.
• Inventory and advertising in one plan — weekly cover reviewed against ad spend, with an agreed rule for throttling spend before a stockout rather than after it.

The following peak delivered $497k in its best month against $204k the year before — 2.4× — with zero days out of stock across the quarter.

Holding availability is what let the rest work: the hero term finished the quarter at #4 organically against #27 a year earlier, and peak ACoS fell from 39% to 24% because the account was no longer buying traffic to compensate for lost organic position. The brand now runs a single plan for inventory and advertising, which is the change most likely to keep this from repeating.