Amazon outage blindsides sellers as ad spend keeps burning

Orders vanished. Ad budgets didn’t.

An Amazon marketplace outage on March 5 (US Eastern Time) reportedly lasted three to four hours, wiping out sales while advertising campaigns continued to spend unabated. Chinese media TMTPost (钛媒体) collected accounts from cross‑border sellers who awoke to Amazon’s signature dog‑error pages, frozen dashboards, and evaporating revenue. One veteran seller said the disruption cost 40% of expected orders; another in Zhengzhou estimated losses of more than $20,000 when “Lightning Deals” collided with the blackout. For many Chinese merchants who time promotions to the US afternoon peak, those hours are everything. When the storefront goes dark, who foots the bill?

‘Money burned, no orders’ fuels refund push

Sellers say Amazon’s ad system kept serving impressions and charging for clicks even as checkout failed, turning ACOS metrics upside down with zero conversions. The lack of a circuit‑breaker between commerce availability and ad billing has drawn sharp criticism: if orders can’t be placed, why does ad spend keep running? Some merchants reportedly secured case‑by‑case refunds after submitting campaign IDs and outage timestamps to support. But there is no platform‑wide guidance, and Amazon’s advertising terms give it wide latitude on performance guarantees—leaving smaller sellers at a disadvantage. The episode underscores a dependency risk: as more budget shifts to automated and AI‑driven placements, manual fail‑safes are harder to pull in a crisis.

What caused it? Rumors, risk—and a familiar culprit

Speculation abounded. One theory—unverified—linked the disruption to Middle East tensions, with reports that AWS data centers in the UAE and Bahrain were affected by drone activity. Another, also unverified, revived debate over automation risk after past incidents allegedly tied to internal AI coding tools and misconfigurations. It has been reported that an Amazon spokesperson later attributed the latest outage to a software code deployment and said the issue was resolved. Either way, the stakes are clear: Amazon’s retail front end is tightly coupled to AWS, and when the cloud stumbles, marketplace commerce and ad economics stumble with it.

The China cross‑border angle

Chinese merchants power a significant share of Amazon’s third‑party marketplace and rely heavily on scheduled deals, tight inventory turns, and paid traffic to hit KPIs. That model amplifies the cost of even brief outages, particularly during US peak buying windows when Beijing is asleep. Industry voices, including Cifnews (亿恩网), advise diversifying channels and retaining manual controls—however limited—over bids and budgets. The lesson? Reliability is a two‑way contract. Sellers want more than ad credits after the fact; they want platform‑level safeguards that pause spend when sales are impossible.