Pick the role whose absence would hurt most and make sure a second person can do the job.
Why does cross training never happen?
The cost is immediate and specific. The benefit is deferred and hypothetical.
Training hours come out of this week, from a person who has work that is already late. Somebody has to cover, or the work slips, and both are visible immediately.
The payoff arrives on a day nobody can name. Somebody is sick, somebody quits, somebody is on a plane when a system fails. Nobody can point at that day in advance, so nobody can price it.
Every argument for cross training loses to every argument against it, right up until the day it does not.
What did cross training look like on a small team?
I put it in so that no role was a single point of failure.
My team did not like it at first. They did not know why it was happening, they did not understand the reasoning, and they did not think they had time for it.
That reaction was reasonable. From their side it looked like being pulled off their own work to learn somebody else’s job for no stated reason.
Then they could all do the job, and they were happy with it. Cross training was very important for our survival, and I would not run a small team without it again.
What changes when everyone can cover?
People can be away.
On a team where each person owns their systems completely, one person being unavailable means an entire area is unattended. Not degraded, unattended. Vacations become negotiations and illness becomes a problem for the company rather than for the person.
Cross training converts that. Somebody can be away and the area still has an owner, which means the rotation works, holidays work, and a resignation is a hiring problem rather than a crisis.
The second effect is quieter. When two people know a system, they discuss it. The person who has owned something alone for years has assumptions nobody has ever questioned, and a second set of eyes finds those.
How do you find hours for training that nobody has?
Take work away first, because willingness does not create time.
The failure I have watched most often is a manager announcing cross training, scheduling it, and then leaving everyone’s existing workload untouched. The training slips every week, and after two months it is quietly abandoned and everyone learns that the initiative was not serious.
Something has to give. Defer a project, accept slower response on something for a month, or bring in outside help to cover the gap while the training happens.
If nothing can give, the honest conclusion is that cross training is not going to happen, and the company should plan around the single points of failure it has rather than pretending they are being addressed.
Which role should be cross trained first?
The one that touches everything, and that is where I got it wrong.
My database administrator overlapped every system and every person. Every application depended on that layer, and every problem eventually became a question for that role.
That role was not cross trained.
The reasons were practical. Database administration is deep, specialized work that takes years rather than weeks, and nobody else on the team had the background or the hours to acquire it. It was the hardest role to cover and therefore the one that did not fit the plan.
So the single point of failure I had designed the entire scheme to eliminate survived in the one place where it mattered most.
Can you buy your way out of a single point of failure?
Partly, and less than I expected.
I hired a database consulting company as a backstop. The theory was that somebody could always step in.
It did not always work out that way.
A purchased backstop is a different thing from a trained colleague and the difference shows up exactly when you need it. The consultants knew databases. They did not know that database, that schema, those decisions, or why any of it had been built the way it was.
Cross training transfers context. A contract transfers capacity. Those get treated as interchangeable on an organization chart and they are not.
What is the honest version of covering your gaps?
The plan covers most of the risk, the hardest piece does not fit, and it gets solved with money that buys something weaker.
That was my situation and I suspect it is most people’s. The scheme worked. Cross training was one of the better decisions I made. And the role I most needed to cover was the one I could not, so I paid for a substitute and the substitute was not equivalent.
Nobody writes it that way. Every account of resilience planning describes a complete solution, because a partial one sounds like failure.
A partial one is what most companies have. The useful step is knowing which gap you did not close, so that when it matters, nobody is surprised by what the backstop cannot do.
