Evan Schuman's StorefrontBacktalk

For somewhere between one and five hours on Monday (Aug. 3), e-tail executives get a harsh reminder from EBay how they are all potentially one coding error away from millions in lost revenue.

EBay’s PayPal group went dark worldwide for all users for an hour Monday, starting at about 1:30 PM (New York time). Many users were unable to make purchases for a much longer period, until the final users were restored by about 6:30 PM. The Wall Street Journal quoted EBay spokesperson Anuj Nayar as saying that the cause of the outage was an “internal network hardware issue” and that EBay was “looking into how to address our affected merchants.”

In all probability, there won’t be much compensation needed, for several reasons. Many customers would have likely paid some other way or simply tried making the purchase a few hours later. But maybe not and that’s where this gets interesting.

Unlike anything that much-larger payment processors Visa and MasterCard have ever seen, there is a growing (and decidedly younger) group of consumers that are making purchases based on the payment method and that method is often PayPal. That’s fine if you’re a retailer that accepts a wide range of payment options, including PayPal. Unless, of course, PayPal pushes all transactions through a single point.

Even worse, when a consumer runs into a payment hiccup in-store, it is recognized and can be dealt with instantly. But in an automated online system, are systems set up to ask the customer to return in an hour? To complete the purchase with an alternative payment method? Or does a failed transaction close out the purchase entirely? (We’ve seen plenty of examples of both.)

If EBay wants to compensate—or, more precisely, console—merchants who were scared POSless on Monday, forget about compensation checks. Instead, do two things. First, provide the community a detailed explanation of what went wrong and what EBay plans to do about it, something that is specific enough so that retail partners can judge for themselves the probability of this glitch happening again. Secondly, invest in more offsite rollover plans and make sure that they don’t need five hours to fully kick in.