◆ AI · analyse

Analyse your data

Upload a messy spreadsheet and ask in plain English — you get the numbers that matter, one clear chart, and the 'so what', without a single formula.

4heights
8tasks
8roles do it
See the prompt↓More analyse workflows

AI prompts

do it · improve it · decide · become
AExecute — “help me do it”You are a sharp data analyst. I'm giving you a spreadsheet. 1. In one line, tell me what this…+
You are a sharp data analyst. I'm giving you a spreadsheet. 1. In one line, tell me what this data is. 2. Clean the obvious problems (blank rows, wrong types, duplicates) and list what you changed. 3. Give me the five numbers that matter most for my goal (ask me the goal if I haven't said it). 4. Make one clear chart of the most important trend. 5. End with three plain takeaways and one thing worth digging into.
when the reply comes backPush once: ask it to sharpen the weakest part, and to say what it assumed.
BImprove — “do it better”Here is my first-cut analysis of the attached data. Make it sharper: - tighten the summary to…+
Here is my first-cut analysis of the attached data. Make it sharper: - tighten the summary to what a busy manager needs - replace any vague claim with the exact number behind it - redo the chart so the single most important message is obvious at a glance - flag anything in my reading that the data does not actually support.
when the reply comes backPush once: ask it to sharpen the weakest part, and to say what it assumed.
CDecide — “help me choose”I have to make a decision from this data: [state the decision, e.g. which region to invest…+
I have to make a decision from this data: [state the decision, e.g. which region to invest in]. Using only what the numbers support: - lay out the two or three real options - for each, the evidence for and against, with the figures - name the risk in the option that looks best - tell me the one extra number you'd want before committing.
when the reply comes backPush once: ask it to sharpen the weakest part, and to say what it assumed.
DBecome — “help me grow”Teach me to do this analysis myself next time, using this data as the example. - walk me…+
Teach me to do this analysis myself next time, using this data as the example. - walk me through the steps you took, in order - show the one Excel formula or pivot that does each step - point out the mistake a beginner usually makes at each stage - give me a short checklist I can reuse on any dataset.
when the reply comes backPush once: ask it to sharpen the weakest part, and to say what it assumed.

The real tasks

8 of them
I have a messy export and no idea where to start
My sales dropped and I need to know why by tomorrow
My boss wants a one-page dashboard this morning
I need the top and bottom performers, fast
I want one clean chart I can drop into a slide
I need to compare this month against last month
I've got survey results and need the headline findings
I need to spot the outliers before the board meeting

Who does this

8 roles
Financial AnalystData AnalystOperations ManagerBusiness AnalystMarketing ManagerAccountantSales ManagerProduct Manager

Questions people actually ask

with the jobs and tasks they touch

You do not need a formula — you upload the file and ask in plain English.

  1. Open ChatGPT or Gemini.
  2. Upload your spreadsheet or paste the rows.
  3. Tell it your goal in one line — 'monthly sales by region'.
  4. Ask for the key numbers, one chart, and three takeaways.
see alsoData Analyst →

Ask the AI to clean it first — it spots the errors a human skims past.

  1. Upload the raw file.
  2. Ask it to flag blank rows, wrong types, and duplicates.
  3. Tell it to fix them and list every change it made.
  4. Check the change list, then analyse the clean version.
see alsoData Analyst →Accountant →

Describe the story you want to show, not the chart type — let the AI pick the right one.

  1. Upload the data.
  2. Say what you want to see — a trend, a comparison, a share.
  3. Ask for one clear chart with a plain-language title.
  4. Ask it to redo the chart if the message isn't obvious at a glance.
see alsoMarketing Manager →

Ask for a one-page report written for a manager with two minutes, not a data dump.

  1. Upload this month's data.
  2. Ask for the headline number and how it moved vs last month.
  3. Ask for a short top-and-bottom table and one chart.
  4. Ask for three 'so what' takeaways in plain English.
see alsoOperations Manager →Financial Analyst →

Treat it like a conversation — the first answer is a draft you push on.

  1. Read the first analysis.
  2. Ask it to explain the biggest surprise.
  3. Ask 'what would change your conclusion?'
  4. Ask it to state what it assumed and what it couldn't see in the data.
see alsoBusiness Analyst →

It is fast at the thinking around the numbers — reading a messy file, spotting patterns, drafting the summary, and making a first chart. It turns 'here is a pile of data' into 'here is what it seems to say' in minutes.

What it is not is a calculator you trust blindly; treat its numbers as a strong first draft you verify.

see alsoData Analyst →

It depends on the settings. For sensitive files, first turn off chat history and training, or use the paid business tier where your data is not used to train the model.

For anything truly confidential, keep it inside your company's own tools — Copilot in Excel or Power BI — where the data never leaves your organisation.

see alsoOperations Manager →

Most everyday business data works well:

  • Spreadsheets — sales, finance, inventory, HR
  • Exports — CSV files from any system
  • Survey results and form responses
  • Simple web or app analytics tables
  • Pasted tables from a document or email
see alsoBusiness Analyst →

Give it the data and the question, not just the data. A vague upload gets a vague answer.

Upload the last few months, say 'sales fell in March, help me find what changed', and ask it to break the drop down by region, product, and week, then name the two biggest contributors and what to check next.

see alsoSales Manager →Financial Analyst →

You do not need a BI tool for one dashboard — the AI can draft it and you polish it.

Upload the data, ask for the three or four numbers that matter and a chart for each, then ask for a one-line reading of each. Drop those into a slide or a one-page sheet, sanity-check the figures by hand, and you're done.

see alsoOperations Manager →

Big files choke the chat, but you rarely need every row for the method.

Trim to the columns you actually need, or paste a few hundred rows and ask for the exact steps, then run those steps on the full file in Excel. For large, repeating data, move it to Power BI — that's the right home for a live dashboard.

see alsoData Analyst →

Use the AI chat to explore and explain fast; use Excel when the numbers must be exact and repeatable. Most people use both — explore in chat, finalise in Excel.

SituationAI chatExcel
Explore a new dataset fastBestSlow
Explain what the data meansBestManual
Numbers that must be exactVerify itBest
Repeat the same report monthlyAwkwardBest
No formulas knownBestHard
see alsoFinancial Analyst →

Both read a spreadsheet and answer well. The tie-breaker is where your data already lives.

If your data is…OpenWhy
In Google Sheets or DriveGeminiit's already right there
Anywhere elseChatGPTstrong, general, easy
Inside ExcelCopilot in Excelno upload, stays private
see alsoData Analyst →

Trust the thinking, verify the arithmetic. AI is genuinely good at spotting the pattern and framing the 'so what', but it can miscount, so always spot-check a figure or two by hand.

For anything that must be exact — money, compliance, board numbers — do the final calculation in a spreadsheet, and use the AI for the reading, not the reckoning.

see alsoAccountant →Financial Analyst →

It replaces the grind, not the judgement. The cleaning, the first chart, the draft summary — that tedious middle is largely gone, and that is a real gift of time.

What stays human is asking the right question, knowing which number is misleading, and deciding what to do about it. The analyst who uses AI to skip the grind and spend the time on judgement pulls ahead of the one who doesn't.

see alsoData Analyst →Business Analyst →