Skip to content

Excel Data Analysis: Monthly Statistics, Charts, and Conclusions in One Go ​

💡 One-sentence summary: Use WorkBuddy to build a spreadsheet analysis pipeline—one instruction produces monthly statistics, charts, and conclusions; fine-tune with follow-ups and paste into your weekly report.

Searching for "AI data analysis" mostly turns up two kinds of results: function tutorials teaching you pivot tables and VLOOKUP, and online BI websites offering "upload your spreadsheet, get charts in one click." The former is endless to learn, and the latter requires uploading your entire business data to someone else's server. This lesson walks through a complete spreadsheet analysis workflow: drag the spreadsheet into WorkBuddy, use one instruction to compute monthly sales and profit by product line, with bar charts, line charts, and conclusions generated at once; follow up for sorting, add a pie chart, export to PDF—just keep talking; finally, copy the chart as an image and paste it straight into your weekly report.

First, the factual boundaries: the monthly statistics example instruction, the four-element prompting method, and the "start with a basic version, then add details round by round" advice in this lesson come from the official practice case "Practice 3: Data Analysis and Visualization"; the follow-ups for sorting, adding a pie chart, exporting to PDF, and copying charts as images come from the official task conversation documentation; the model selection advice and the consumption figure of "about 200-500 credits for data analysis" come from third-party testing by Tencent Cloud community authors and are not official commitments.

In the previous lesson, B1 Excel Data Cleaning, we got the spreadsheet clean; in this lesson, we make the clean data speak. This is lesson 2 of Route 2 in "Scenario Practice," based on WorkBuddy v5.9.3.

What You Can Do After This Lesson ​

  • One instruction to compute monthly sales and profit by product line, with the summary table generated directly (the official example wording can be copied as-is)
  • No manual chart insertion in Excel: bar charts, line charts, and conclusion notes generated at once
  • Keep talking to sort, add a pie chart, or enlarge fonts—the context stays intact, no need to re-describe the spreadsheet
  • One-click "copy as image" for charts, pasted directly into weekly reports, PPT, and email

Why WorkBuddy Instead of a Web-Based AI? ​

CapabilityWeb-Based AIWorkBuddy
Read an entire spreadsheet❌ Paste hundreds of rows—truncated, formatting breaks✅ Drag Excel / CSV / TSV directly into the conversation area to read (officially supported formats)
Statistics and charting❌ Replies with a paragraph of analysis text or Python code—you still have to run the chart yourself✅ Reads data, completes statistics, and generates charts within one task (official practice case)
Multi-round adjustments❌ Once the context breaks, re-paste the data and re-describe it✅ Sorting, adding a pie chart, exporting to PDF—just keep talking (official example scenario)
How to take charts with you❌ Screenshots—uneven edges and white borders✅ Chart component's "copy as image," pasted into weekly reports and PPT (official feature)
Internal data security❌ The entire business data uploaded to the cloud✅ File processing done locally by default, raw data never leaves your machine (official statement)

This is not marketing talk: the WorkBuddy Bench evaluation benchmark, officially open-sourced by Tencent, includes a dedicated Office subset where all 50 questions require producing a completely correct target deliverable on files in mixed xlsx / csv / pdf / docx formats—office file handling is a foundational capability seriously tested by this official benchmark.

Your Current Pain Points ​

  • End-of-month business analysis: hundreds of rows of sales detail, pivot tables learned and forgotten and relearned, formulas erroring out halfway through
  • You want AI to draw charts, but the web version replies with a wall of text plus a matplotlib snippet—and the chart still never appears
  • Charts are all inserted manually: select the data range, switch chart type, adjust colors—and when the data changes, redo everything
  • The numbers are there, but the conclusions—"what does this mean, what to do next month"—you still have to squeeze out yourself staring at the screen

When to Use This Technique ​

When you need to:

  • Answer business questions from a spreadsheet: which month was best, which product line is dragging, up or down month over month
  • Produce statistics and charts by month / quarter / product line, ready to use in the end-of-month report
  • Get charts and conclusions in one go, without repeatedly tweaking things by hand in Excel

And you don't want to:

  • Learn pivot tables, VLOOKUP, and chart styling
  • Upload business data to an online BI tool
  • Stare at a blank document squeezing out conclusions yourself

🎒 Preparation Before Starting ​

  • Completed B1 Excel Data Cleaning, with the spreadsheet deduplicated and formats unified (if not cleaned yet, go back and clean it first)
  • Have a spreadsheet to analyze (.xlsx / .csv / .tsv all supported)
  • Know clearly which business question you want to answer: which month, which product line, which metrics
  • Have credits remaining: all conversations consume credits, see the rules at 2.6 Credits and Pricing
  • 15 minutes

Follow Along With Me ​

Step 1: Drag the spreadsheet into the conversation area (AI data analysis starts with reading the file) ​

Why: WorkBuddy is a desktop AI workbench that reads files directly on your computer—no need to paste hundreds of rows of data into the chat box, and no need to upload to online tools.

How to do it: Drag the spreadsheet to analyze directly into the input box area, or click the upload button; if the file is elsewhere, you can also just tell it the path. Officially supported formats are Excel, CSV, TSV (plus PDF, Word, PPT, and other common formats), with up to 50 files draggable at a time.

You should see: the file attached above the input box. After sending, WorkBuddy reads the spreadsheet, and the reply shows the row/column counts and column names—proof that it actually read it in.

  • WorkBuddy's reply shows my column names, and the row count matches

💡 Clean before you analyze

Don't skip the cleaning workflow from the previous lesson: with duplicate rows not removed and formats not unified, neither the statistics nor the charts can be trusted. If not cleaned yet, go back to B1 Excel Data Cleaning.

Step 2: Spell out metrics, charts, and dimensions in a single instruction ​

Why: Official practice cases recommend stating four things at once in your analysis request: the metrics to analyze, the chart types you want, the statistical dimensions or time range, and whether a report is needed. The more complete your request, the closer the first version of the result will be to usable.

Action: The official example instruction, verbatim (replace the metrics and dimensions with yours):

text
Please read this sales data, tally sales revenue and profit by month for each product line, generate a bar chart and a line chart respectively, and add a concluding summary.

Broken down against the four elements: metrics = sales revenue and profit; chart types = bar chart + line chart; statistical dimensions = by month × each product line; report = add a concluding summary.

You should see: WorkBuddy automatically reads the data, completes the statistics, and generates charts — a monthly summary table, a bar chart, a line chart, and a concluding summary appear in the conversation.

💡 This is the unique value of WorkBuddy

When a web-based AI makes charts, what it gives you back is a block of matplotlib code; when you ask WorkBuddy to draw charts, it directly reads the file, computes the statistics, draws the charts, and writes the conclusion — all done within one task. What's more, file processing is done locally by default and raw data never goes to the cloud (per the official statement), so your business data never needs to be uploaded to any website.

  • All four elements are written out in the instruction, not just a vague "help me analyze this"

Step 3: Verify the statistics before trusting the conclusions ​

Why: Whether AI data analysis conclusions can be trusted depends on whether the statistical methodology is correct. Are there gaps in the months? Was profit calculated the way you intended? Verify first, then cite — don't carry flawed numbers into your reports.

Action: Ask WorkBuddy to lay out the statistical process:

text
List the complete monthly summary table, and explain how sales revenue and profit were each calculated (which columns were used, and whether any rows were excluded or patched), so I can verify.

You should see: a complete month × product line summary table plus an explanation of the calculation methodology. Spot-check a few numbers: add up a few rows from the original table for a given month and product line, and if they match the summary table, the charts and conclusions that follow can be trusted.

  • Spot-checked 2-3 numbers, and the statistical methodology matches my expectations

Step 4: Follow up on sorting, adding a pie chart, exporting to PDF — just keep talking ​

Why: The first version of a chart is rarely perfect right away. The context is still there, so you don't need to re-describe the table — the follow-up examples given in the official task conversation documentation are exactly this scenario.

Action: Official examples, verbatim — use as needed:

Follow-up scenarioOfficial example
Additional requestAdd a pie chart as well; make the fonts a bit bigger
Adjust existing resultsSort the table by sales revenue; add an executive summary to the report
Go deeper in the analysisHelp me export it as a PDF; also analyze the month-over-month changes

Just send it directly, for example:

text
Sort the table by sales revenue
text
帮我导出成 PDF

Not sure what to follow up with? After completing a round of conversation, suggested next-step prompts appear above the input box — click to fill one in, edit it, and send. You can also send messages while a task is running — new messages go into a queue during execution and are processed in order once the current round of output finishes, without interruption.

You should see: the table sorted by sales revenue, an additional pie chart, and an extra PDF in the artifacts area. After each round of follow-up, the charts and conclusions update accordingly.

  • After follow-ups the context isn't lost, so no need to re-describe the table background

Step 5: Copy charts as images and paste them into weekly reports and PPTs ​

Why: Analysis results ultimately need to go into weekly reports, PPTs, or emails. Charts generated in the conversation are widgets that can be turned directly into images — no screenshots needed.

Action: Hover over a chart widget, and the top-right corner offers three entry points: "Add to conversation", "Download locally", and "More actions (⋯)"; "Copy as image", "Save as image", and "View code" are all in the "More actions" menu (per the official statement):

ActionDescription
Copy as imageCopies the chart's current content to the clipboard, with the confirmation "Image copied"; paste directly into other apps
Save as imageSaves as an image file, with the confirmation "Image saved"
Download locallySaves the widget file
View codeSwitches to showing the widget's code; click "Show UI" to switch back to the chart view

You should see: the "Image copied" confirmation; paste it directly into your weekly report or PPT — the image is clean with no white borders. If you want to turn the whole analysis into a presentation, continue to A4 Generate a PPT in one sentence.

  • Chart copied as an image and pasted into my document

📋 Magic Prompts ​

Turn the workflow above into three copyable templates: first a basic version, then progressively deeper rounds, and finally a report. Placeholders are marked with 【】 — replace them with yours and they're ready to run.

📝 Monthly statistics + dual charts + conclusion ​

Expected result: statistics, charts, and a conclusion done with one instruction — a template version of the official practice case example; just fill in the four elements and it works.

text
## Role
You are a data analysis expert, skilled at finding trends and anomalies in sales data.

## Task
Please read this 【sales data】, tally 【sales revenue and profit】 by 【month】 for each 【product line】, generate 【a bar chart and a line chart】 respectively, and add a concluding summary.

## Input information
### Required fields
- Data file: 【the dragged-in .xlsx / .csv file】
- Analysis metrics: 【e.g., sales revenue, profit】
- Statistical dimensions: 【e.g., by month × product line】
- Chart types: 【e.g., bar chart + line chart】
### Optional fields
- Time range: 【e.g., only 2026 year 1-9?】
- Whether to output a report: 【e.g., attach a one-page conclusion report?】

## Output specifications
- A summary table by dimension
- Charts of the specified types
- A concluding summary pointing out trends and months worth noting

## Constraints
- ✅ All numbers come from the table — no fabrication
- ✅ Produce the basic version first; add more only after I confirm
- ❌ Do not modify the original file
- ❌ Do not introduce external data not present in the table

📝 Month-over-month deep analysis (second-round follow-up) ​

Expected result: dig deeper after the basic version is out — a template version of the official example "also analyze the month-over-month changes", finding the months and product lines with the biggest rises and falls.

text
## Role
You are a data analysis expert, skilled at finding anomalies and inflection points in time series.

## Task
Based on the monthly statistics just produced, further analyze the month-over-month changes: calculate each month's month-over-month growth rate, identify the 【month / product line】 with the largest increase and the largest decrease, and explain possible reasons.

## Input information
### Required fields
- Comparison metric: 【e.g., sales revenue month-over-month】
### Optional fields
- Focus scope: 【e.g., only product line A?】
- Comparison basis: 【e.g., month-over-month / year-over-year?】

## Output specifications
- A month-over-month summary table + one trend chart
- A list of the top 3 risers and fallers, with reason clues visible in the table

## Constraints
- ✅ Only use existing data for calculations
- ✅ Clearly mark reason speculation as speculation
- ❌ Do not fabricate market information not in the table
- ❌ Do not alter the summary numbers confirmed in the previous round

📝 Analysis report version (conclusions + recommendations + audience) ​

Expected result: make full use of the "whether to output a report" element — beyond the charts, also request a shareable visual analysis report. The official guidance recommends that report requests simultaneously specify: report format, section structure, chart preferences, and the audience.

text
## Role
You are a business analyst, skilled at turning data into conclusions management can use directly.

## Task
Based on the monthly statistics from this 【sales data】, generate a visual analysis report including 【core conclusions, key data charts, trend judgments, and action recommendations】, suitable for 【management】 to read.

## Input information
### Required fields
- Data file: 【the dragged-in file】
- Audience: 【e.g., department head / management】
### Optional fields
- Report format: 【e.g., PDF / Word?】
- Section structure: 【e.g., conclusions first, charts at the end?】
- Chart preferences: 【e.g., line charts for trends, pie charts for proportions?】

## Output specifications
- A complete report file, with charts embedded in the corresponding sections
- Core conclusions up front, supporting charts at the end

## Constraints
- ✅ Conclusions cite their basis (which table, which number)
- ✅ No more than 【3】 action recommendations, all actionable
- ❌ Do not fabricate numbers not in the table
- ❌ Conclusions must not go beyond what the data can support

Checkpoint ✅ ​

  • The table has been dragged into the conversation area, and WorkBuddy's reply shows the column names and row count
  • The instruction covers all four elements: metrics, chart type, statistical dimension, and whether to output a report
  • The bar chart and line chart have been generated, and 2-3 summary figures were spot-checked
  • After follow-ups on sorting and adding a pie chart, the context stayed intact and the charts updated accordingly
  • The chart has been "copied as an image" and pasted into the weekly report or PPT; the original file was never touched

Pitfall Reminders ​

❌ Requesting too much at once ​

Asking for statistics, dual charts, a pie chart, month-over-month comparison, and a report all in one breath—the first version tends to drop the ball on some of them, and every round of changes means rework.

The official usage advice, verbatim: "Avoid requesting too much at once: it is recommended to produce a basic version first, then add charts and conclusions round by round."

Correct approach: for the first version, only ask for the summary + chart + conclusions; bring up the pie chart, month-over-month comparison, and report in the second round (step 4).

❌ Choosing a chart type before the business question ​

Seeing a cool chart in someone else's report and ordering the same—no matter how pretty a chart is, if it can't answer your question, it's useless.

Official usage advice: "Clarify the analysis question first: state clearly what business question you want to answer, then decide the chart format."

Correct approach: first write down the question you want to answer (which month was the worst?), then decide the chart type (line chart for trends, bar chart for comparing sizes, pie chart for proportions).

❌ Clicking "Copy as Image" before the chart finishes loading ​

You click and nothing happens, so you think the feature is broken—in fact, the component is still generating. Official wording: when the content has not finished loading, "Copy as Image" and "Save as Image" are temporarily unavailable; these two options are also not yet supported on mobile.

Correct approach: wait until the chart finishes loading before taking action; to save an image on mobile, go back to the desktop.

❌ Generating charts straight from dirty data ​

Duplicate rows not removed, formats not unified, months missing—the statistics are untrustworthy at the root, and the prettier the chart, the more likely it is to mislead whoever you're reporting to.

Correct approach: first go back to B1 Excel Data Cleaning for a health check and cleaning, then come back to analyze.

💡 How many credits does it cost

Every conversation consumes credits. The officials haven't given specific consumption figures for data analysis tasks; third-party testing puts it at "data analysis + chart generation costs about 200-500 credits," for reference only. Each round of follow-up counts as a conversation, so once you're happy with the first version, don't keep adding charts endlessly. For credit rules, see 2.6 Credits and Pricing.

Advanced Tips (Optional) ​

  • Try switching models: Third-party testing suggests using MiniMax M2.5 for Excel processing and data analysis tasks (fast execution), and reminds users to "switch models flexibly based on task type; don't stick with one model for everything." For how to switch, see 2.5 Model Selection.
  • Cite a table for follow-up questions: Hover over the summary table, and an "Add to conversation" icon appears in the top-right corner; click it and a table reference is generated in the input box—ask "why did this month drop" about a specific table without re-describing it. See 2.1 Conversation Advanced.
  • Hand recurring monthly analyses to scheduled tasks: Is the end-of-month business analysis the same every week/month? Scheduled tasks support 7 recurrence rules—set it once and it runs automatically on time. For details, see 3.3 Scheduled Tasks.

Lesson Summary ​

In this lesson, you completed the full workflow of using WorkBuddy for AI data analysis:

  • ✅ One instruction: state the four elements clearly—metrics, chart type, statistical dimension, and report—and get monthly statistics + dual charts + conclusions in one go
  • ✅ Follow-up questions: sorting, adding a pie chart, month-over-month analysis, exporting to PDF, with no break in context
  • ✅ Take charts with you: "Copy as Image" to paste directly into weekly reports and PPTs, no screenshots needed
  • ✅ Verify before trusting: spot-check the numbers, clarify the statistical criteria, and go back to B1 to clean dirty data first
Key PointsCriteria
Core MethodFour elements in one instruction → basic version → add round by round (official usage advice)
Official ExampleTally sales and profit by product line by month, bar chart + line chart + conclusion notes
Follow-up ExamplesSort the table by sales / add a pie chart / export it as a PDF for me (official documentation)
Taking Charts AlongComponent "Copy as Image / Save as Image," available after loading finishes, not yet supported on mobile
Data SecurityFile processing is done locally by default; raw data is not uploaded to the cloud (official wording)

Next Lesson Preview ​

Squeezed everything out of the tables at hand? In the next lesson, let WorkBuddy find the data itself—B3 Research Reports: From Data Collection to a Visualized Report in One Delivery.

You will learn:

  • Data not yet organized into files: have WorkBuddy collect market information and organize it into structured data with a single sentence
  • Two-step prompt: first ask for the data, then ask for a visualized report suitable for management reading
  • Have the AI self-check the quality of its output, with data sources and time ranges stated up front
gaodong.aiAboutPrivacyContact

Independent tutorials, not affiliated with any product mentioned. Product names and trademarks belong to their owners. Check official pages for prices and policies.