Excel is still worth learning properly — even with AI

A reasonable question to ask right now is whether it’s still worth investing time in learning Excel properly. AI tools can generate formulas on demand, explain functions in plain English, and write macros you’d have spent an hour researching. Why develop expertise in a tool when you can just ask an AI to do it?

The answer, frustratingly, is that it matters more than ever. Here’s why.

AI generates formulas. You still have to judge them.

When you ask an AI assistant for a formula, it will usually give you one that looks right. Sometimes it is right. Sometimes it’s subtly wrong — a lookup that doesn’t handle duplicates correctly, an aggregation that includes rows it shouldn’t, a date calculation that breaks at month boundaries. The formula works in the obvious case and fails in the edge case you haven’t tested yet.

If you don’t understand how the formula works, you can’t catch these errors. You’ll accept the output, build downstream work on top of it, and only discover the problem when the numbers don’t add up at month-end — or worse, when someone else finds it.

Excel knowledge is what turns AI assistance from a potential liability into a genuine productivity gain. The people who use AI tools most effectively are the ones who can read the generated formula, spot the issue, and correct it. The people who struggle are the ones who accept the output without understanding it.

The functions that are worth learning cold

Not every Excel function deserves deep investment. These ones do:

XLOOKUP and INDEX/MATCH — the foundation of almost every data-matching task. XLOOKUP is more intuitive; INDEX/MATCH is more flexible and still more common in practice. Understanding both, and specifically understanding why they behave differently with approximate matches and duplicates, saves significant time across your career.

SUMIFS, COUNTIFS, AVERAGEIFS — conditional aggregation is the engine of most financial analysis in Excel. Knowing exactly how the criteria ranges work, how to reference them dynamically, and how to nest conditions correctly makes this family of functions genuinely powerful.

Dynamic arrays (FILTER, SORT, UNIQUE, SEQUENCE) — introduced in Excel 365, these changed how Excel works. A FILTER formula that spills its results is fundamentally different from a VLOOKUP — it updates automatically as the data changes, and it can return multiple columns. Accountants who haven’t made the switch are still doing work manually that could be dynamic.

IFERROR and error handling — knowing when and how to handle errors gracefully, rather than suppressing them blindly, is the difference between spreadsheets that communicate clearly and ones that silently hide problems.

Named ranges and structured tables — not functions exactly, but features. Accountants who use Excel tables (Ctrl+T) get automatic formula extension, structured references, and cleaner pivot table behaviour. Skipping this is leaving practical value on the table.

Understanding the model matters as much as the functions

The accountants who get into trouble with Excel aren’t usually the ones who don’t know enough functions. They’re the ones who don’t have a clear mental model of how Excel calculates — the order of operations, what volatile functions are and why they slow down workbooks, how circular references occur, why a formula returns the wrong number even when it looks right.

This kind of structural understanding isn’t taught in most Excel training. It develops through practice and exposure. But it’s the difference between someone who can use Excel and someone who can diagnose Excel — who can look at a complex workbook built by someone else, understand what it’s doing, and identify where it might be wrong.

In audit especially, this matters. Being able to open a client workbook, understand its structure, and spot calculation errors is a core skill — and it’s one that requires genuine Excel depth, not just formula knowledge.

AI doesn’t change the need for data discipline

A significant part of Excel competence isn’t about formulas at all — it’s about structuring data so that it can be analysed without constant manual intervention.

The accountant who receives a client’s data export and knows immediately how to reshape it for analysis — using Power Query, or structured tables, or well-designed pivot tables — is faster and less error-prone than one who works around the structure, adding helper columns and manual steps that need to be redone every month.

AI can advise on data structure in theory. But applying that advice requires understanding what good data looks like in Excel and how to get there in practice.

The AI floor vs the Excel ceiling

Here’s a useful way to think about it. AI tools raise the floor — they make it easier for less experienced users to produce reasonable outputs. But they don’t raise the ceiling. The upper limit of what you can do with Excel is still determined by what you know.

In a team where everyone has access to the same AI tools, the differentiator is expertise. The person who understands Excel well builds better models, catches errors earlier, works faster, and is less dependent on external help. That gap doesn’t close when everyone has an AI assistant — it becomes more visible.

What investment in Excel skill actually looks like

For most accountants, the path to genuine Excel proficiency isn’t a formal course — it’s focused practice on the functions that appear in your actual work, combined with deliberate attention to the areas where you know you’re working around gaps rather than through them.

A few specific things worth doing:

• Replace every VLOOKUP you write with XLOOKUP until it becomes automatic
• Learn one dynamic array function properly — FILTER is the most immediately useful
• Understand why your workbook is slow before reaching for a workaround
• When AI gives you a formula, read it carefully before using it

The goal isn’t to know every function in Excel. It’s to be fluent enough that you can work confidently, build reliable models, and use AI assistance without being misled by it.

How tech+bash can help

Coaching from tech+bash is available for individuals and teams who want to build genuine Excel proficiency, not just function familiarity. Sessions are grounded in accounting and audit context — the examples are relevant, the problems are ones you’ll actually encounter, and the focus is on the understanding that doesn’t appear in formula documentation.

Get in touch to discuss what you’re working on.