How to test Excel skills in an interview

Start from the job, not from a list of functions

Write down three to five things the person will actually do in Excel in their first few months. "Update the weekly sales summary", "reconcile two exports", "build a quote from a price list". Those tasks are your test. A list like "VLOOKUP, PivotTables, macros" copied from a job ad tests what sounds advanced, not what the role needs.

Then pick the level. These are rough guides; your own task list wins if it disagrees.

• Admin, coordinator, customer support: enter and tidy data, totals and counts, SUM, COUNTIF, a simple IF, sorting and filtering, copying a formula down without breaking it.
• Operations, sales ops, junior analyst: everything above plus lookups (XLOOKUP or VLOOKUP with FALSE), SUMIFS and COUNTIFS with criteria in cells, cleaning text, dates and working days, a PivotTable.
• Financial or data analyst, FP&A: everything above plus multi-condition lookups, SUMPRODUCT, dynamic arrays such as FILTER and UNIQUE, finding a bug in someone else's formula, and laying out a sheet other people can trust. Hiring for accounting, AP or AR? See the Excel test for accounting roles for reconciliation, aging and tie-out tasks.
• Leave out VBA, Power Query and keyboard-shortcut speed unless the job genuinely uses them. Testing them filters out people who would do the real job fine.

Why a work sample beats self-ratings and quizzes

"Advanced Excel" on a CV means different things to different people. Some people who wrote "intermediate" are better than the ones who wrote "expert", and a rating scale in an application form has the same problem. You learn nothing until you see them work.

Multiple-choice quizzes are better than nothing and quick to mark, so they're fine as a first screen when you have many applicants. But picking the right formula from four options is recognition. Writing one from a blank cell, on data with a blank or a typo in it, is the actual job. A short work sample shows that directly, and it shows how they work: whether they check, whether they hard-code, whether they get stuck and recover.

Our list of Excel assessment questions for employers has ready-made tasks at three levels, with an answer key.

Pick a format: live, take-home or online

• Live, with screen share or in the room (15-30 minutes): you see their process and can ask "why that formula?". Some good people freeze when watched, so keep it short and friendly, and allow them to look things up the way they would at work.
• Take-home file (30-60 minutes of work): calmer, and closer to real work. You can't see the process, so follow up with a short call where they walk you through it. Don't send a four-hour project; strong candidates with jobs drop out.
• Timed online test: easiest to run at volume and the same for everyone. Check that it uses real spreadsheet tasks, not just multiple choice, and that you can see the formulas they wrote, not only a score.
• Whichever you pick, tell candidates in advance: the format, the time, roughly what it covers, and which version of Excel they'll use. Functions like XLOOKUP and FILTER need Excel 2021 or Microsoft 365.

An example task, and what a good answer looks like

The sheet above is a typical small task. Orders are in A:F, a price list in H:I, and the region to report is in K2. Here are two tasks and how to read the answers.

Score with a rubric you write first

Decide what full marks look like before you see any answers, and score every candidate on the same points. Scoring 0, 1 or 2 on each line keeps it quick.

• Correct result on the data they were given.
• Points at cells: inputs like the region, a rate or a price come from cells, not numbers typed into the formula.
• Still works when the data changes: change K2, edit a number or insert a row, and see whether the answer follows. You can do this yourself in two minutes on a take-home file; some online tools do it automatically. Cellwise Hire, for example, reruns each candidate's formula on hidden versions of the sheet with different numbers and an extra row.
• Checked their work: a quick cross-total, a count, or a glance at the status bar before saying "done".
• Can explain it: in two sentences, what the formula does and why they chose it.

Green flags and red flags

• Green: looks at the data before writing formulas and spots the blank or the text in a number column; asks one clarifying question; uses the simplest method that's right; says what they'd check next.
• Green: gets stuck, says so, and tries another route. That is what working competently in Excel looks like.
• Red: typed results or hard-coded numbers presented as formulas; IFERROR wrapped around everything so problems disappear instead of being fixed; ranges that stop one row short of the data.
• Red: can't say what their own formula does. A pasted formula from a search is fine at work, but they should understand it before they rely on it.
• Not a red flag: using VLOOKUP instead of XLOOKUP, a PivotTable instead of formulas, or looking up syntax. Judge the result and the reasoning, not the method you'd have picked.

Keep the test fair

• Same test, same instructions, same time limit and same scoring for every candidate for the role.
• Keep every task tied to the job. If you can't say which part of the role a task reflects, drop it.
• Set a realistic time and say what happens if they don't finish. Most people should be able to complete the core tasks.
• Make it clear how candidates can ask for adjustments, such as extra time, and handle requests consistently.
• Use invented or anonymised data, never live company work. A test is not unpaid labour.
• Follow the hiring laws where you recruit, and record scores the same way for everyone.