Finding the single points of failure on your project
In May, the tower crane operator on a ten-storey rental building mentions that he has booked two weeks off in August, the month the curved glass corner goes in. The crane supplier has no other operator who knows this crane on this site, and only one plant in the region bends the glass. The project manager starts to wonder what else on the job has no second.
List everything the job depends on that has no ready substitute — a person, a supplier, a machine, sometimes a file or an access route — estimate how long each would take to replace, and put a backup behind the ones that would stop critical work. Single points of failure hide because they work: the operator turns up, the supplier delivers and the crane runs, until the day one does not. Recovering once one has failed is covered in when a supplier fails.
- Walk the critical and near-critical work and ask of each activity: which person, supplier and machine must be there for it to happen?
- Mark those with no second: the only certified operator, the one person who holds the survey control or the owner's approvals, the only supplier able to make an item, one crane, one access road.
- Estimate the time to replace each: the days to find, qualify, orient or deliver a substitute, and what stops meanwhile.
- Choose a backup that fits: a second person named, qualified and oriented, a second source qualified early, spares ordered with the first order, a rental agreed in advance. Cross-training your own crews is covered in planning around absences.
- Protect what cannot be backed up with a buffer, maintenance and early orders, and check it weekly.
- Keep the list with the risk register, review it monthly, and revisit it whenever a key person, supplier or machine changes.
Common mistakes
- Listing machines and forgetting people and single-source suppliers.
- Naming a backup who has never been asked, trained or oriented.
- Accepting "we'll find someone" as a backup.
- Making the list once and never updating it when people or suppliers change.
Checklist
For each single point
- What depends on it, and whether that work is critical
- Days to replace it, and what stops meanwhile
- A named, tested backup, or the reason there cannot be one
- Protection where there is no backup: buffer, maintenance, early orders
- An owner who reviews it with the risk register
Check your understanding
Your only survey technician keeps the site's control points in a file on his own laptop. Is that a single point of failure, and what is the backup?
