For analyst, FP&A, pricing, operations and office roles, the Excel test is usually not a vendor product but a workbook the team built: a few sheets of raw data, a list of questions, and 20 to 60 minutes to answer them in your own Excel, sometimes while sharing your screen. Finance roles stretch the same idea to one-to-three-hour modeling tests and take-home cases. There is no published rubric; graders look for accuracy first, then completeness, then structure, then formatting.
Across analyst guides, Big Four and bank interview reports, the same seven tasks appear. Candidate reports from EY, J.P. Morgan and Deloitte name SUMIF, VLOOKUP and COUNTIF and a one-hour live finance case; data-analyst guides add pivot summaries, date standardization and QA checks.
Bring the product category or customer segment onto the transactions with VLOOKUP, XLOOKUP or INDEX and MATCH, and check the key for duplicates first.
Revenue by region and month, orders above a threshold, conversion rates per channel: SUMIFS and COUNTIFS, sometimes AVERAGEIFS.
Late deliveries, margins below target, eligibility rules: IF with AND or OR, IFERROR around divisions.
Sales by rep by quarter, grouped rates, a top-10 list. Build it, or rebuild it with formulas when the test forbids PivotTables.
Dates stored as text in two formats, month and year columns, working days between two dates.
Remove duplicate rows, treat "N/A" and blanks consistently, unify "West" and "west ".
Share of total, growth versus last year, a column or line chart of one series with a sensible title.
Speed on messy data is the whole skill. Practise on datasets with dirt in them, keep a QA habit (totals, duplicate keys, impossible values), and rehearse saying what a number means in one sentence.
Do the lookups track: exact and approximate VLOOKUP, XLOOKUP, INDEX and MATCH, and the #N/A handling exercise. Then practise counting duplicates on the key before any join.
Do the SUMIFS and COUNTIFS exercises with two and three criteria.
Rebuild a PivotTable with formulas, the way out when the test says formulas only: a category-by-month grid with one SUMIFS, then a summary by region with GROUPBY in Excel 365.
Work the text and date exercises: text to real dates, EOMONTH due dates, NETWORKDAYS, split and trim names.
Do the UNIQUE and FILTER exercises and the conditional-formatting rules for overdue items.
Time yourself. Give every practice session a clock of 20 minutes for five tasks, and write one line of interpretation under each answer, as if the grader will read it.
Learn the keyboard moves that finance tests assume: Ctrl + Shift + Arrow to select, F4 to lock a reference, F2 to edit, Alt + = for AutoSum, Ctrl + 1 for the format dialog.
Our Level 2 test puts 20 tasks into two business workbooks of two to three sheets with one 25-minute clock, the closest free rehearsal for this family.
A lookup key that appears twice in the reference table doubles every matching row. Analyst guides call it the hidden risk; count the key with COUNTIF before you join, and compare totals before and after.
"Most spreadsheet mistakes are really filtering mistakes": a rate computed over all rows when the question asked about one region, or over filtered rows when it asked about all.
Candidates report tests that took two to three hours when the email said 30 to 60 minutes. Answer every question roughly first, then refine; a blank question scores nothing.
A total typed in, a tax rate inside a formula, a copied number instead of a link: graders check formulas, and a number nobody can trace fails the accuracy test.
A full three-statement model when a growth table was asked for, colours and borders before the numbers are right. Structure and formatting rank after accuracy and completeness.
Modeling-test guides list "did not read the prompt" next to hard-coding as the most common failure: wrong period, wrong unit, wrong definition of margin.
Most are 20–60 minutes. Live screen-share versions run about 30 minutes; emailed assignments are often an hour; investment-banking and private-equity modeling tests run one to three hours; take-home cases give you 24–72 hours for four to eight hours of work.
Against the team's own answer key, in roughly this order: accuracy, completeness, structure, formatting. Interview guides for banks say balance sheets that do not balance and hard-coded numbers are the most damaging errors, and that reasoning is weighed as heavily as the final numbers.
Often, yes, for a summary by category and period. Know how to build one and how to produce the same table with SUMIFS and COUNTIFS when the test says formulas only.
Only if the employer offers it. Finance teams expect Excel on Windows and keyboard speed; data teams are often flexible. Ask before the test, not during it.
Usually. Expect a written-answer cell or a conversation about why you chose a method and what the number means for the business. Practise saying it in one or two sentences.
TestGorilla-style, iMocha, Adaface, Mettl, Testlify
10–30 minutes 10–30 questions
One question per screen with a clock, usually no going back. The first filter after the CV.
Qué esperar de esta pruebaSHL-type, eSkill, Criteria Interactive, Testlify, Vervoe
25–35 minutes 20–30 tasks
A mock Excel window with a task under it. One attempt per task, graded on the end result, some shortcuts switched off.
Qué esperar de esta pruebaMicrosoft Office Specialist exams MO-210 and MO-211 (Certiport)
50 minutes 23–37 tasks
Several projects in live Excel, each a workbook with 4–7 tasks. Scored 1–1000, 700 to pass.
Qué esperar de esta pruebaGratis, sin cuenta y sin instalar nada.
Facts checked on November 15, 2026 against Interview Query's data-analyst Excel assessment guide (May 1, 2026), investment-banking modeling-test guides (updated November 3, 2026), Mergers & Inquisitions on timed LBO tests, the AFP's FP&A case-interview article (updated May 7, 2024), and candidate interview reports for EY, Deloitte and J.P. Morgan. Employers do not publish rubrics; the grading order is how those guides describe it.
Excel es una marca comercial del grupo de empresas Microsoft. Indeed, TestGorilla, SHL, iMocha, Adaface, Criteria y otras marcas son propiedad de sus respectivos titulares. GridBelt es independiente y no está afiliado a ninguno de ellos, ni respaldado ni patrocinado por ellos.