August 01, 2017
Do you think about the “nines” as a trade-off, or a standard?
As a payment processor, we’ve assumed that any cash equivalent payment – debit cards, prepaid cards and electronic benefits transfer (EBT) cards – should be as reliable and immediate as cash. To that end, we established the “five nines standard.” That stands for 99.999 percent uptime, or less than one second per day or five minutes per year of downtime. Practically speaking, not every platform in the payments industry can or will perform at this level, but the commitment among key players is to strive to achieve the highest level of uptime possible.
Often, the focus on uptime comes at a major cost, and discounts the other values cards and digital payments offer over cash, such as greater convenience, security, brand and loyalty affinity, control, data availability, etc. In recent engagements with procurement groups, I am certain we have all touched on the topic of uptime without truly examining what it takes to deliver such performance.
Is it time to rethink the five nines?
The Cost of the Nines
The five nines cost millions of dollars every year. And they do so for uptime that is only slightly greater than standards that would cost far less. (It’s simple math, but dropping one of the “nines” would result in eight-and-a-half seconds of downtime daily. Three nines – 99.9 percent – would result in roughly one-and-a-half minutes of downtime per day, and so on…). Would you accept eight and a half seconds of inconvenience to your customers? Is it reasonable to ask for millions of dollars in annual investment for infrastructure to shrink that inconvenience by a few seconds? In some cases, that is absolutely a reasonable ask, but in others, it may not make sense, and that is an important discussion to have.
Is it Time to Rethink the Five Nines?
As a payment processor, we must maintain a standard applicable to all our clients. Although some retailers may be willing to bear a little more inconvenience on their and their customers’ parts to keep costs down, the processor maintains an infrastructure that meets the most demanding customer’s standard. Let’s keep in mind also that many platforms are architected such that further optimization isn’t feasible without complete overhaul – which takes both considerable time in addition to cost.
Processors routinely deal with the usual balls in the air – potential machine failures, network failures and people failures. But now, security and fraud issues are adding a layer of complexity that works against the notion of delivering the fastest possible processing. How much is stopping fraudulent transactions worth? That extra security and fraud protection increases either time or money. Who pays for it, when holding to the five nines standard? As prices are always expected to go down (a topic for another day), is it reasonable that the standards are expected to rise?
It seems to me that it’s time to rethink what the marketplace really expects in terms of speed instead of defining the five nines as the gold standard. What is the marketplace willing to pay for five nines when costs of meeting that standard escalate? Do the benefits still jibe with the new realities of fraud, security, and risk?