Use AI for · Excel and data analysis · 26 Sep 2026
Use AI for
AI for Excel

AI for Excel:Formulas, analysis,results you can trust

AI writes formulas from your description, spots trends and outliers and builds charts from your spreadsheets. This guide covers the tools available, how to prepare your data and prompts, and how to check results before anyone makes a decision based on them.

Get in touch
On this page
Enquiry
Nikolai Schöbel und Jeremias Burger, Co-Founder Scalableloops

Let's talk about your project.

First we check whether the project fits your business model. Then you get a proposal with phases and effort.

Discuss AI for data analysis or call: +49 151 1576 5566
Use AI for · Excel and data analysis

AI for Excel and other spreadsheet tools turns plain-language questions into formulas, analyses and charts: you describe what you want to know, and the AI calculates it or suggests the right formula. You can do this with built-in assistants such as Copilot in Excel or Gemini in Google Sheets, or with AI chats such as ChatGPT, where you upload a spreadsheet. The results become reliable only with clean data, a precise question and a check by a person.

In brief
  • AI helps you write formulas, summarise spreadsheets, find trends and outliers and create charts.
  • Copilot in Excel and Gemini in Google Sheets work directly in the spreadsheet; ChatGPT analyses uploaded files.
  • Well-structured data with clear column headers is the single most important condition for useful results.
  • Every figure that feeds into a decision gets recalculated, and spreadsheets containing personal data belong only in tools with a business contract.
Published 26 Sep 2026Nikolai Schöbel and Jeremias Burger9 min read
Nikolai SchöbelJeremias Burger

Nikolai Schöbel and Jeremias Burger

Co-founders of Scalableloops GmbH. Nikolai Schöbel leads online marketing and AI strategy, Jeremias Burger the AI architecture. Both build AI systems and train teams on them in their own agency work.

On this page
  1. What can you usefully do with AI in Excel?
  2. Which AI tools work with Excel and spreadsheets?
  3. How do you prepare a spreadsheet for AI analysis?
  4. How do you phrase prompts for formulas and analysis?
  5. How do you check AI results in spreadsheets?
  6. Which spreadsheets can go into an AI tool?
  7. What does your team need to know to use AI in Excel?
  8. Frequently asked questions
  9. How to get started with AI for Excel
  10. Where the information on this page comes from
Overview

What can you usefully do with AI in Excel?

In many businesses, knowledge lives in spreadsheets: sales by month, orders by customer, survey responses, hours by project. Often only a fraction of it gets analysed, because the right formula is missing or nobody has time to build a pivot table. This is where AI comes in. You describe in plain language what you want to find out and get back a formula, a summary or a chart.

Typical tasks include having AI explain a formula a colleague built years ago; calculating a new column, such as a margin or the number of days between two dates; sorting free text such as customer feedback into categories; comparing monthly figures and flagging unusual spikes; and creating a chart for a meeting.

The limit is reached when the spreadsheet itself is wrong or the question is unclear. An AI works with whatever it finds, and it can misread a column without telling you. That is why every analysis needs a review step, described further down.

TaskWhat the AI deliversWhat you check
Write a formulaA formula based on your description, with an explanation if you askRecalculate a few rows by hand, including references
Understand a formulaStep-by-step explanation of an existing formulaWhether the explanation matches the actual formula
Summarise a tableTotals, averages, groupings, pivot tablesWhether all rows and the right period are included
Trends and outliersPointers to unusual values and developmentsWhether an outlier is real or a data entry error
Categorise textCategories or sentiment for free text such as customer feedbackRead a sample of the assignments
Create a chartA chart for the question you askedAxes, units and labels
Tools

Which AI tools work with Excel and spreadsheets?

You will come across three types in the office. The descriptions below follow what the vendors state on their own pages, as of 26 September 2026. They are not a ranking.

Copilot in Excel. According to Microsoft, you can ask Copilot questions about your data in your own words and get back summaries, trends and outliers. Copilot creates formulas that calculate new values from existing data and can explain how they work. It builds charts and PivotTables, and it can summarise free text, identify themes and sentiment, and add the results as a new column.

Gemini in Google Sheets. Google lists creating tables, formulas and charts and identifying trends and outliers among the things Gemini does in Sheets. There is also the AI function, which you type into a cell like a formula, following the pattern =AI(prompt, range). It generates text, summarises, categorises content and analyses sentiment. According to Google, it requires an eligible Workspace or Google AI plan.

AI chats with file upload. In ChatGPT, you upload a spreadsheet and ask questions about it. According to OpenAI, ChatGPT creates tables and charts from it and, for some analysis tasks, writes and runs Python code. The advantage is that a program does the arithmetic, not the language model itself. Similar chats such as Claude or Gemini also work with uploaded files; check each provider for the formats and features that apply.

Built-in assistants work where your spreadsheets already are. A chat suits one-off analyses and files you export anyway. Our article on the AI assistant for business helps you decide which type fits your office.

Preparation

How do you prepare a spreadsheet for AI analysis?

The quality of an AI analysis depends first on the spreadsheet. Things an experienced person reads correctly, such as merged cells or a subtotal in the middle of a list, can mislead an AI. These points are worth checking before every analysis:

  1. 01

    01

    One row, one record

    Each row describes exactly one item, such as an order or an invoice. Subtotals and blank rows do not belong in the data list.

  2. 02

    02

    Clear column headers

    A header row with unambiguous names such as “Invoice date” or “Net amount EUR”. Units go in the header, not in the cells.

  3. 03

    03

    Consistent formats

    Keep dates, amounts and spellings consistent. Otherwise an AI counts “Smith Ltd” and “Smith Limited” as two customers.

  4. 04

    04

    Format as a table

    In Excel, format the range as a table so that tools and people can see where the data starts and ends.

  5. 05

    05

    Share only what is needed

    Remove or replace columns with names, addresses or salaries if the question does not need them.

Prompts

How do you phrase prompts for formulas and analysis?

A good prompt names the columns, describes the result you want and says what should happen with edge cases. Here are some examples you can adapt:

Write a formula: “Create a formula for column F that applies the discount percentage in column E to the net amount in column D. If E is empty, the net amount should stay the same. Explain the formula in two sentences.”

Understand a formula: “Explain step by step what this formula calculates, and list cases where it could return an error or a wrong result: …”

Analysis: “Compare revenue by product group for the first and second half of the year. Show the result as a table with the absolute and percentage change, and name the three groups with the largest difference. Tell me how many rows you analysed.”

Categorise free text: “Assign each customer comment in column C to exactly one of these categories: delivery, price, quality, service, other. Write the category in column D.”

Asking for the number of rows analysed is a simple checkpoint: if it differs from your spreadsheet, the AI has missed data or counted it twice.

Review

How do you check AI results in spreadsheets?

Language models sound confident even when they are wrong. In a representative survey published by the German digital association Bitkom in November 2025, 42 percent of people who use AI for search said they had already received false or made-up information. With figures, a mistake often goes unnoticed because the result looks plausible. Four checks catch most errors.

Spot-check by hand. Recalculate the result for two or three rows yourself, including one with unusual values such as empty cells or zeros.

Totals check. The sum across all groups must equal the grand total of the spreadsheet. If it does not, rows are missing or were counted twice.

Read the formula. Ask for an explanation of every formula and check that the references point to the right columns and the whole range.

Question the outliers. An unusual value may be a real development or a typo in the source. A person who knows the data makes that call.

If you regularly carry analyses over into reports or bookkeeping, our guide to AI in accounting has more on records and review duties.

Data protection

Which spreadsheets can go into an AI tool?

Many spreadsheets contain personal data: customer lists, sales by contact person, working hours, salaries. As soon as you put such data into a cloud AI service, the provider processes it on your behalf. Under Article 28(3) GDPR, that requires a data processing agreement.

The German Data Protection Conference, the body of Germany’s independent data protection authorities, recommends in its guidance on artificial intelligence and data protection using work accounts instead of personal ones and checking whether inputs are used for training. With ChatGPT Business and Enterprise, OpenAI does not use your data for training by default; on personal accounts such as Free and Plus, training is on until you switch it off in the data controls. According to Microsoft, Copilot in Microsoft 365 does not use prompts, responses or your data to train the underlying language models.

Often a simple step is enough: replace names with customer numbers and remove columns that play no part in the question. If you would rather keep spreadsheets in-house entirely, our article on local AI for business covers the alternatives.

Team

What does your team need to know to use AI in Excel?

Since 2 February 2025, Article 4 of the EU AI Act has required organisations that use AI systems to take care of their staff’s AI literacy. Since 27 July 2026, it has applied in the version introduced by the so-called Digital Omnibus, Regulation (EU) 2026/1744: you have to take measures that support the development of AI literacy, but you do not have to guarantee a specific level of competence for any individual. No certificate is required. More in our article on the AI literacy obligation under Article 4.

For spreadsheet work, the key point is that your team must be able to check results. Anyone who adopts an AI-generated formula should be able to read it and verify it against a sample. So AI does not make Excel skills redundant; it turns them into the control mechanism.

Analyses often end up in presentations or project reports. See our guides to AI in project management and creating presentations with AI.

Frequently asked questions

Frequently asked questions

Can AI write Excel formulas?

Yes. Copilot in Excel, Gemini in Google Sheets and AI chats such as ChatGPT create formulas from a plain-language description and explain them on request. Check every formula against a few rows by hand before applying it to the whole spreadsheet.

What is the difference between Copilot in Excel and ChatGPT?

Copilot works directly in your Excel file and adds formulas, columns, charts and PivotTables there. In ChatGPT, you upload a file and get analyses, tables and charts in the chat, with Python code running in the background for some tasks. Which one fits depends on where your spreadsheets live and which contracts are in place.

What does the AI function in Google Sheets do?

You type it into a cell like a formula, with a prompt and an optional cell range. According to Google, it generates text, summarises content, sorts it into categories and analyses sentiment. It requires an eligible Workspace or Google AI plan.

Can I upload a customer list to ChatGPT?

Not with a personal account. For personal data you need a data processing agreement and settings that exclude training on your inputs. It is often enough to replace names with numbers and remove columns you do not need.

How can I tell whether an AI analysis is correct?

Recalculate a few rows by hand, compare subtotals with the grand total and ask the AI how many rows it analysed. If anything does not match, check the data and the prompt before you work with the result.

Do I still need Excel skills if AI writes the formulas?

Yes, but with a different focus. Instead of building every formula yourself, you need to read formulas, check references and test whether results are plausible. Without that knowledge, an AI error goes unnoticed.

What now

How to get started with AI for Excel

  1. 01

    Define the question

    Choose an analysis that comes up regularly and takes a lot of manual work today, such as a monthly comparison or evaluating customer feedback.

  2. 02

    Tidy the spreadsheet

    Make sure there is one row per record, clear column headers and consistent formats, and remove columns with personal data the question does not need.

  3. 03

    Clarify tool and contract

    Check whether your office suite includes an assistant and whether there is a business contract with training switched off.

  4. 04

    Practise checking

    Test the analysis with a spot check and a totals check, and train your team to do the same. We can help with AI training or an AI agent system tailored to your business.

AI opens up spreadsheets that hold knowledge nobody has had time to use, and it takes the hunt for the right formula off your hands. The results become reliable through clean data, a precise question and a person who recalculates before a figure feeds into a decision.

or call: +49 151 1576 5566

Further reading
Sources

Where the information on this page comes from

Projekt-Detail

Problem

Our solution

Our role

Highlights

    Got a project in mind?

    We reply personally. First a use-case check, then an architecture proposal.

    Start your inquiry