Programming skills are essential for data scientists and engineers, and becoming more commonplace for data analysts.

This is a great thing – automating activities, solving problems more efficiently and more easily accessing datasets turns any good analyst into a great one overnight.

However, as these newer, shinier tools are adopted, I’ve seen Excel scoffed at and discarded. Yes, it has substantial limitations compared to Python and R, but it wins out in at least one big way: Direct auditability by the business.

You can’t take a piece of code outside of the data team and expect anyone to understand what you are doing.

Even if you lay out your assumptions in modelling and analysis clearly at the top of your code, people don’t know how to read or understand it.

This is known as communication bias – where your audience doesn’t have the capacity or competence to question your work, so they accept it as is.

This matters because important business decisions might be made from your analysis – but they only see an end result, and not the workings within. As business management can’t ask informed questions, this can lead analysts to think they are right, when instead they just aren’t being questioned – which isn’t good for anyone.

Many more people can use Excel than can code.

Using Excel where you know the business will leverage the end result allows the business to use the model, alter assumptions, and add to and modify the workbook. This frees up you to do other analytical work, rather than maintaining code for eternity, and empowers others in the business to contribute to your work.

The best lawyers know that the law is complex and difficult to interpret, so they make a major effort to communicate in the language of their clients.

The best analysts do the same. This means not using the shiniest technology in the room but, where the business will be affected, using tools like Excel that business stakeholders can understand.

This is obviously not always possible, but where available it should be considered. A common example is financial models. This is possible to do through programming but understanding budgets is important for every business department, not just the ones who code.