On paper, cross-training is irresistible: if everyone can do everything, no absence is scary anymore. In practice, the store where "everyone does everything" is often the store where everyone does everything averagely, and where people who are excellent at one thing spend their days plugging holes elsewhere, wondering why.
The difference between the two versions is not goodwill. It is the method used to build the versatility.
Start from the risk, not the ideal
The right question is not "how do I teach everyone everything?" but "which gap hurts us most?". If the only person who can handle complex returns calls in sick, the store limps for days: that is the first skill to duplicate. The map takes ten minutes: for each critical task, how many people can genuinely do it? The rows where the answer is "one" are your priority list, and there are usually three or four of them, not twenty.
That also flips the numbers: useful cross-training is not "everyone on everything". It is "at least two people on every critical thing". A far more reachable goal.
Learn in the quiet slots, make mistakes without an audience
The second rule is obvious and ignored everywhere: someone's first time on a new task cannot be Saturday afternoon. Whoever is learning the till should learn it on a Tuesday morning, with an experienced colleague alongside and three customers an hour, not with a huffing queue. Which means shadowing sessions belong in the shift plan like any other commitment, in the low slots, with names attached: who shadows whom.
It also helps to cut theory to the bone: for each area, one card with the three things you must know and the two situations where you stop and call the point person. The rest is learned by doing.
Rotation is not a way to hide a staffing hole
Honesty required here: if rotation only exists because a position has been vacant for six months, the team knows it perfectly well. Calling it "professional growth" makes things worse, because it adds insult to fatigue. Versatility holds when it is a declared investment with a visible return for the people doing it: more varied shifts, better pay or more say in preferences, and the reasonable certainty of not being stuck forever in the least loved department.
And one clear limit: rotating is fine, rotating daily is not. Someone who never spends two consecutive weeks on the same task does not become versatile. They become temporary everywhere.
Keep score, for fairness and for safety
Last piece: rotation needs tracking. Who has done the till shadowing? Who is autonomous on returns? Who has not touched the stockroom in eight months and needs a refresher before the stocktake? Without a minimal record, versatility lives in the manager's memory, and the manager's memory goes on holiday too.
When skills per role live inside the shift plan, coverage gets built with them in mind from the start: that is how Sked Solve works, assigning shifts based on each person's roles. The site has the full overview.