Do We Still Need Excel Skills in the Age of AI?

AI can build formulas, charts and summaries in seconds. The real business skill is knowing whether the workbook can be trusted.

By Sara Hassouneh

Artificial intelligence has changed the way people use spreadsheets. A manager can describe a task in plain language and ask an AI assistant to create a formula, summarise a table, identify a pattern or draft a chart. Microsoft says Copilot in Excel can work with formulas, tables, charts and PivotTables, and can make changes directly in a workbook. These capabilities can accelerate everyday work, but they do not make spreadsheet knowledge obsolete.

AI increases the value of a different kind of Excel literacy. The priority is understanding how data is structured, how calculations behave, where errors enter a workflow and what evidence is needed before a result can support a business decision. For a UAE business using spreadsheets to connect marketing exports, customer records and management reports, these skills remain essential.

AI changes the task, not the responsibility

An AI tool can generate a plausible formula without knowing whether the business definition behind it is correct. It may calculate revenue when the team needs gross margin, use a simple average when a weighted average is required, include cancelled orders, misread dates or summarise a partial range. A polished chart can therefore present an incorrect answer convincingly.

Microsoft’s own guidance tells users to review, edit and verify AI-generated content. AI can accelerate spreadsheet work, but the person approving the result still needs to understand what it means and how it was produced.

What Excel knowledge still means

Useful Excel knowledge in an AI-assisted workplace is practical. Employees do not need to remember hundreds of functions, but they should be able to:

• Recognise sound data structure. One row should represent one record; columns should have consistent meanings; dates, currencies and percentages should use consistent formats; and record identifiers should remain unique.
• Read the logic. A user should be able to trace where a number came from, understand the inputs to a formula and identify whether a cell reference is absolute, relative or broken.
• Test the output. Important totals should be reconciled against a source system or checked using a second method. Reviewing individual records can reveal duplicated rows, missing information or unexpected exclusions.
• Control changes. Teams should preserve source data, document assumptions and be able to restore an earlier workbook when an automated edit produces an unwanted result.
• Recognise the limits of a workbook. A spreadsheet deserves closer review when it depends on one person’s undocumented knowledge or supports a critical process without adequate access controls and validation. A CRM, database, ERP or analytics platform may be more appropriate for some tasks.

Why this matters for UAE businesses

The UAE Digital Government describes its role in developing policies and supporting digital transformation. For businesses pursuing their own digital improvements, a practical question follows: how can familiar tools support better decisions without introducing avoidable errors?

Excel can be useful for budgets, sales pipelines, inventory checks, campaign reporting and scenario planning. A small business can use a workbook to compare exports from different systems or investigate a question before investing in a more specialised solution.

However, a spreadsheet can become an unofficial business system without the controls normally associated with specialist software. AI may make it easier to create a sophisticated workbook, while also making incorrect logic harder for an inexperienced user to notice. A sensible approach is to assign an owner, separate source data from calculations, document business rules and require review before using the results for significant decisions.

A practical example from lead generation

The following scenario is hypothetical, illustrating the type of marketing work handled by Snap Story. Imagine a campaign produces 1,200 leads for a client. The export contains a lead ID, company name, contact name, email address, phone number, source and submission date. An AI assistant is asked to clean the data, identify possible duplicates and produce a summary before the file reaches the sales team.

Suppose a review of the resulting workbook finds three problems: some company names have been paired with the wrong contact details; repeated email addresses were not consistently recognised because of spaces and capital letters; and some phone numbers lost their original formatting during conversion.

The reviewer should keep the original export unchanged and work on a separate copy. A stable lead ID allows records to be compared with the original export. Email addresses can be standardised in a helper column, while phone numbers should be imported and stored as text. Changing a damaged phone field to text afterwards will not restore characters that have already been lost; those values must be recovered from the source.

For the formulas below, assume row 1 contains headers and the 1,200 records occupy rows 2 to 1201. Column D contains the original email addresses. Column H is a new helper column for normalised email addresses. Enter the cleanup formula in H2 and copy it down to H1201. Enter the duplicate-check formula in I2 and copy it down to I1201.

Control	Excel check	Business purpose
Stable key	Compare records using the unique lead ID	Checks that each company and its contact details still match the original record
Email cleanup	=LOWER(TRIM(D2))	Converts email text to lowercase and removes ordinary leading and trailing spaces for comparison
Duplicate flag	=AND(H2<>"",COUNTIF($H$2:$H$1201,H2)>1)	Returns TRUE for non-empty normalised addresses that appear more than once
Phone format	Import and store the field as text	Preserves country codes, leading zeros and plus signs present in the source
Reconciliation	Compare source and final record counts	Requires an explanation for every deletion, merge or exclusion

These formulas support review; they do not validate an email address or prove that two records describe the same lead. Shared company inboxes may legitimately appear on several records. Preserve the original addresses and investigate flagged records before merging or deleting them. Adjust the row range for a different dataset; some Excel regional settings also require semicolons instead of commas in formulas.

The person approving the file needs enough spreadsheet knowledge to challenge the output. A workbook that looks clean may still contain mismatched records, and an AI-generated explanation does not establish that the checks were performed correctly.

A practical checklist for AI-generated spreadsheets

• Purpose: Is the workbook answering the business question that was actually asked?
• Source: Can every important figure be traced to an identifiable data source?
• Scope: Are the correct dates, entities, products, currencies and status categories included?
• Structure: Are headers, data types, identifiers and table ranges consistent?
• Logic: Do formulas use the intended inputs, filters and weighting method?
• Exceptions: How are blank cells, errors, returns, cancellations and possible duplicates handled?
• Verification: Have critical totals been recalculated independently or reconciled against the source?
• Security: Is the AI tool approved for the personal, confidential or commercially sensitive information being processed?
• Change control: Can the team inspect what changed and recover the previous version?
• Decision: Can the person approving the result explain how it was produced?

Where AI is most useful

AI is useful when the task is clearly defined and the output can be checked. It can help draft a formula, explain an unfamiliar function, suggest a PivotTable, identify possible trends, standardise labels or create an initial chart. It can also help less experienced users formulate questions about their data.

For work that affects important decisions, the source data, assumptions and resulting changes should remain reviewable. Microsoft’s Copilot in Excel guidance discusses reviewing workbook changes and recovering previous versions. Businesses should combine the available features with clear ownership and approval procedures.

The skill is judgement

We still need Excel skills in the age of AI, but the emphasis is changing. Understanding how a spreadsheet represents a business process helps users describe a task clearly, organise the inputs and recognise when an automated result is misleading.

For UAE businesses, the practical goal is to use AI to save time while maintaining reliable checks. Spreadsheet literacy gives teams the ability to explain the numbers, correct mistakes and decide when a result is ready to use.

About Sara Hassouneh

Sara Hassouneh is an entrepreneur and the founder of Snap Story. She helps businesses across the MENA region strengthen their brands through marketing, storytelling and brand positioning. She holds an MBA from Al-Quds University and a bachelor’s degree in Mass Media Communication from Abu Dhabi University. Her work combines creativity, AI and strategy, and she also works as a marketing mentor.

LinkedIn: linkedin.com/in/sarahassouneh
Website: snapstory.biz

AI-generated illustration created with Google Gemini, courtesy of Sara Hassouneh.