How to approach Customer dimensions and work hours relation problem

I am thinking about how to solve the following problem related to a supermarket and its employees’ workload prediction:

I have customer dimensions (attributes):

• CustomerID
• Loyalty status
• Uses e-store
• Monthly Spend
• Monthly Spend in e-store
• Delivery %
• Pick-up %
• In-store purchase %

Interested in FTE requirements in the following:

• Selves restocking
• Cashier
• Product compiler for e-store
• Take-away

Based on the customer’s behaviour, I would need to build a model to estimate the workload that the store must do. E.g. if the customer uses e-store, this means the store must have an employee, who compiles the products and ships them. When a person comes to the store, then a cashier is needed and shelves must be filled. et cetera. Each FTE calculation is according to a ratio.

My idea:

Cluster the customers according to their dimensions. Either K-means and using the elbow or by PCA to understand relevant dimensions.

Then build linear regression on top of it for each of the contributions to the FTE number. Sum the FTEs for the total effect.

Any other suggestions?

Much appreciated!