Imagine you are preparing a short report on two training events. Your manager asks a simple question: “What percentage of registered people attended?”
You could now ask an AI tool to create the spreadsheet, calculate the rates and write the summary. On October 29, 2026, Google announced that the Gemini app can generate, preview and edit spreadsheets directly in a chat. Google says spreadsheet editing is available on the web wherever the Gemini app is available.
That makes the first draft easier. It also makes one skill more valuable: knowing whether the answer actually follows from the data.
The small calculation that changes the whole report
Here is a fictional example:
| Event | Registered | Attended | Attendance rate |
|---|---|---|---|
| Dhaka webinar | 100 | 60 | 60% |
| Chattogram workshop | 20 | 18 | 90% |
What was the overall attendance rate?
It is tempting to average the two percentages: (60% + 90%) ÷ 2 = 75%. That would treat a group of 20 people as if it were the same size as a group of 100.
The correct calculation uses the people behind the percentages: (60 + 18) ÷ (100 + 20) = 78 ÷ 120 = 65%.
The difference is ten percentage points. If you send “75% attendance” to management, the report looks convincing but gives the wrong picture. This mistake can happen in a manual spreadsheet too; AI simply makes it easier to produce a polished report before anyone notices.
The lesson: whenever you combine percentages from groups of different sizes, ask what each percentage is out of. For an overall attendance rate, divide total attendees by total registrations. Do not simply average the event rates.
Give the AI the calculation rule, not just the task
“Analyze this spreadsheet” leaves too much open to interpretation. A useful prompt tells the tool what counts, how to calculate it and what to show for review. Try this with the fictional numbers above:
Create a simple spreadsheet for these two events: Dhaka webinar, 100 registered and 60 attended; Chattogram workshop, 20 registered and 18 attended. Show the attendance rate for each event. Calculate the overall rate as total attendees divided by total registrations, not as the average of the two event rates. Show the formula, the totals and a two-sentence summary for a manager. Use only the figures I supplied. If a figure or definition is missing, flag it instead of guessing.
You can use the same approach in an AI tool that creates a file or ask for formulas to enter yourself in Excel or Google Sheets. The point is to make the calculation visible.
Before you share, check the source rows
The formula may be right while the underlying data is wrong. If you are working with a detailed registration sheet, look for three common problems:
- Duplicate IDs: Did one person register twice? Decide whether you are counting registrations or unique people.
- Missing values: Does a blank attendance cell mean “did not attend” or “not recorded yet”? Those are different answers.
- Mixed formats: A number saved as text may be missed by a total. Dates and event names written in different formats can also split one group into several.
Ask AI to flag these rows and explain its assumptions. Review the original sheet before you delete, fill or change anything. For real student, customer or company records, follow your organization’s data rules before uploading a file to an external tool. A fictional or anonymized sample is a good place to practice.
A two-minute review you can repeat
Before sending an AI-assisted spreadsheet or report, answer these questions:
- What does one row represent? A registration, a person, a sale or an event?
- What is the denominator? Which total sits beneath the percentage?
- Can I reproduce one important number? Check the source and calculate it yourself.
- Does the written conclusion match the calculation? A chart and a confident sentence cannot fix an incorrect total.
This is the practical value of AI spreadsheet analysis: spend less time building the first draft and more time asking the questions that make the result useful.
The skill worth learning now
AI can help make the sheet, suggest formulas and turn numbers into a readable summary. You still need to define the question, recognise a misleading average and decide what the data supports. Those habits work in student projects, marketing reports, sales tracking and everyday office tasks.
At Transfotech Global Bangladesh’s Prompt Engineering & Claude Cowork program, learners practice turning plain-language requests into spreadsheet formulas, cleaning a sample BDT sales sheet with free AI tools and checking outputs before using them. Start with a small dataset you understand. If you can explain where every number came from, your next AI-assisted report will be much stronger.
Frequently asked questions
Can AI create a spreadsheet from a prompt?
Yes. For example, Google says Gemini can now create and edit spreadsheets in the web version of its app, where the feature is available. You can also ask an AI assistant for formulas and enter them into your usual spreadsheet tool.
Do I need to know advanced mathematics?
No. Start with the meaning of the numbers: What am I counting, and what am I dividing by? In the example above, total attendees divided by total registrations gives the answer.
Should I trust a summary if the spreadsheet looks professional?
Check at least one key result against the source data and inspect the formula. A clean layout tells you nothing about whether a duplicate row or a wrong denominator changed the answer.
