The freezer, the certificate and the one person who knows

Single points of failure are not always hardware. As John Sansbury argues in The Journey to Service Excellence, they can be software, a licence, a certificate, a supplier or one irreplaceable person whose knowledge exists only in their head, and they usually stay invisible until they break.

• One expiry can have global reach. Carriers in eleven countries were affected on the same day by a failure nobody knew was a single point of failure or was about to expire; the compensation Ericsson was believed to have paid O2 alone was in the order of £100 million.
• Universities have their own hidden weak links. Examples include a minus eighty freezer with an untested alarm, a research licence whose renewal date sits in one inbox, a computing cluster relying on a script only one postdoc understands, and a long-serving technician about to retire with undocumented knowledge.
• Know what downtime costs. Very few leaders know what an hour of downtime costs, yet it is the most effective business case for prevention. TSB's 2018 migration failure cost £366 million once compensation, fraud losses, fixing costs and lost income were counted.
• Look deliberately, with simple techniques. Single point of failure analysis asks, at each step of a service, what happens if this one thing fails and whether anything stands behind it. Fault tree analysis works backwards from an unaffordable failure to map every path that could cause it. Neither needs expensive tooling, only time to look before failure strikes.