elalaouifaris
We used it for budget estimation at multi million / up to 100 m. I have seen other uses in Corp world in many departments.
IMHO I think it’s good for specific use cases. But it is routinely abused beyond that.
The key shortcomings for excel uses by non-programmers for key business applications:
- How do you test that your calculation is correct after update. The same argument holds for business applications without tests?
- how do you maintain the knowledge of the inner workings without having to follow the arrows around cells. Probably there are good practices but how likely are you to ensure them in your product / team.
- the data / computation is spread in user space. In case your app is useful many people would like to use it. How do you manage the updates and bug fixes beyond it’s the user’s Responsibility
- how do you avoid the black box effect: mission critical software that no one knows how to touch inside without a full rewrite?
- how do you convince stakeholders that they need to migrate to an adapted solution while they have a working one now and often are oblivious to the hidden costs of users copy pasting data / reformatting for hours sometimes to fit an existing tool that is no longer adapted.
The data frame structure in R or python solves a large portion of these issues. Yes you need training. It’s the same for excel if you want to avoid in inferno machine case you need the same concepts.
Why not train for programming with data frames and have the option to gradually extend to web app without a full it project that needs 3 levels of validation.