30 AI Prompt Examples for Sales, Finance, and Operations Data Analysis

AI prompts for data analysis work best when they contain four elements: business context, a description of the data, a specific question, and an output format. This article provides 30 prompt examples for sales, finance, and operations analysis that you can copy right away, plus an executive summary template and ways to check the results.

Using AI at work has become commonplace. In McKinsey's 2025 global survey of 1,993 respondents, 88% reported that their organization uses AI in at least one business function, and 79% already use generative AI. Yet nearly two thirds have not started scaling it across the enterprise. One obstacle looks trivial: everyone writes prompts their own way, so analysis results are never consistent.

What Is an AI Prompt for Data Analysis?

An AI prompt for data analysis is a written instruction that asks AI to read, process, and interpret business data, such as sales tables, profit and loss statements, or customer service ticket logs. The clearer the goal, metric definitions, and report format you request, the sharper the findings.

Large language models (LLMs), the technology behind chatbots and AI assistants, do not know your company and do not know which metrics the board cares about most. So treat a prompt like a brief for a new analyst: anything you leave out, the AI will fill in with its own guesses.

Anatomy of an AI Data Analysis Prompt: Context, Data, Question, Output

A good data analysis prompt has four elements. If one is missing, the AI fills the gap with its own assumptions.

Context. Example: "I am Head of Sales at an FMCG distribution company with 12 branches. This analysis is for the Q4 target-setting meeting."

Data. Besides the source, period, columns, and units, state the limitations, for example "September data does not yet include returns" or "sales values exclude VAT." Small details like these prevent wrong conclusions.

Question. Stick to one prompt, one analytical goal. A sharp question names the metric, period, segment, and comparison, as explained in our guide on how to analyze data with AI without coding. "Why did the East Java branch's gross margin fall 3 points in Q3 2026 compared with Q2?" is far easier to answer than "how are our finances doing?".

Output. Specify the format and the reader. A summary for the CFO is different from working notes for an analyst.

Example: before and after

Incomplete prompt:

Analyze this sales data.

Complete prompt:

I am a Sales Manager at a consumer electronics company. The attached data contains daily sales from January to September 2026 by SKU and by city (columns: date, sku, city, units, revenue_idr). Identify the three SKUs with the largest revenue decline in Q3 compared with Q2, then explain the likely causes based on patterns by city. Present it in a table and close with three short recommendations for the sales team meeting.

The first prompt usually produces only a description of numbers already visible in the table. The second produces causes, unusual findings, and recommendations.

10 AI Prompt Examples for Sales Analysis

Sales teams usually have the most data but the least time to read it. The ten prompts below help VPs of Sales, sales representatives, and analysts read revenue trends, product performance, customer behavior, and pipeline health from POS data, CRM, or spreadsheet exports.

How to use them: the prompts below focus on the data, question, and output elements. Open with one sentence of context about your company, as in the anatomy section example. Replace the text in square brackets with your own data; numbers in brackets, such as [10]%, are sample values you can adjust.

Sales Trends and Performance

Prompt 1: Monthly revenue trend

I am Head of Sales at an FMCG distributor. The attached data contains monthly revenue for [period] by product category. Calculate month-over-month and year-over-year growth for each category. Flag months with an increase or decrease above [10]%, then explain any seasonal patterns. Present the results in a table with a two-paragraph summary.

Prompt 2: Target achievement by branch

Compare sales target achievement by branch for [quarter]. Rank branches from highest to lowest achievement and calculate the gap against target in rupiah and percent. Also flag branches whose three-month trend is declining even though they hit their quarterly target.

Prompt 3: Transactions that deviate from the pattern

Review the daily transaction data for [period]. Find days or stores with sales values more than two standard deviations from their average. For each finding, show the date, location, value, and suspected cause (promotion, public holiday, or input error), along with your confidence level in that suspicion.

Prompt 4: Retail promotion impact

We ran a [promotion type] promotion in [number] stores on [date]. Compare unit sales and margin during the promotion with the four weeks before and after. Calculate the sales uplift, the change in margin, and signs that the promotion cannibalized sales of other products in the same category.

Customer and Product Analysis

Prompt 5: RFM customer segmentation

Use customer transaction data (columns: customer_id, transaction_date, transaction_value_idr) to build an RFM segmentation, which groups customers by how recently (recency), how often (frequency), and how much (monetary) they transact. Divide customers into five segments, name each segment, describe its characteristics, and suggest one retention tactic for each.

Prompt 6: Products frequently bought together

From transaction data per receipt at our retail store, identify the 10 product pairs most frequently bought together. For each pair, show the percentage of receipts containing both and how likely the second product is to be bought when the first one is. Recommend the three bundles most worth testing.

Prompt 7: B2B customers who stopped buying

The attached data contains B2B customers who have not made a repeat purchase in [90] days. Compare their characteristics (industry, company size, contract value, number of complaints) with customers who are still active. Name the three factors that most distinguish the two groups, then list 20 active customers whose profiles resemble those who stopped buying.

Pipeline and Forecasting

Prompt 8: Car dealership sales funnel

I am a Sales Manager at a car dealership. The attached data contains prospects for [period] (columns: prospect_id, source, entry_date, test_drive_date, spk_date, delivery_date, salesperson). Calculate conversion rates from prospect to test drive, test drive to SPK (vehicle order form), and SPK to delivery, by prospect source and by salesperson. Flag the stage with the longest wait time.

Prompt 9: Next-quarter sales forecast

Based on monthly sales data for the last 24 months, build a three-month sales forecast in three scenarios: conservative, base, and optimistic. Explain the assumptions behind each scenario, the method used, and the uncertainty range. Present the forecast as an estimate, along with the factors that could change it.

Prompt 10: Win and loss reasons

From notes on won and lost deals during [period], group the loss reasons into price, features, timing, and client relationship. Calculate the share of each category, show representative notes, then recommend two changes to the sales process.

10 AI Prompt Examples for Financial Analysis

In finance, one wrong number can change a decision. The ten prompts below speed up the review of financial statements, budget variances, cash flow, and profitability for CFOs, controllers, and financial analysts. Still reconcile every figure with your accounting system before using it.

Financial Statements and Variance

Prompt 11: Budget variance

Compare actual costs for [month/quarter] with the budget by department and by cost item. Show variances in rupiah and percent, then highlight variances above [5]%. Based on patterns from previous months, separate variances that are only a matter of timing from variances that are likely to persist.

Prompt 12: Profit and loss summary

Summarize the profit and loss statement for [period] for the board meeting. Explain the changes in revenue, gross profit, operating expenses, and net profit compared with the same period last year. Limit it to five key points and one table of gross margin, operating margin, and net margin.

Prompt 13: Financial ratios

From the balance sheets and income statements of the last three years, calculate liquidity ratios (current ratio, quick ratio), solvency (debt to equity), and profitability (ROA, ROE). Explain the trend of each ratio and what it means for the company's ability to take on new loans.

Cash Flow and Profitability

Prompt 14: 13-week cash flow forecast

Based on the attached opening cash balance, receivables collection schedule, payables schedule, and weekly fixed costs, build a 13-week cash flow forecast. Flag the weeks when the balance is expected to fall below the minimum threshold of [Rp X], then propose options to cover the shortfall.

Prompt 15: Receivables aging

Build an accounts receivable aging report from invoice data (columns: customer, invoice_date, due_date, amount_idr, status). Group it into 0 to 30, 31 to 60, 61 to 90, and over 90 days. Show the 10 customers with the largest overdue receivables and calculate days sales outstanding (DSO).

Prompt 16: Profitability by product or customer

Calculate the contribution margin per [product/customer], taking into account cost of goods sold, discounts, shipping costs, and allocated service costs. Rank from highest to lowest margin and flag those with a negative margin.

Prompt 17: QSR outlet food cost

I am a Finance Manager at a quick service restaurant chain with [number] outlets. The attached data contains sales, raw material purchases, and opening and closing stock per outlet for [period]. Calculate the food cost percentage for each outlet, compare it with the [30]% target, then show the five outlets with the largest gap along with the ingredients that contribute most to it.

Risk and Compliance

Prompt 18: Transactions to review

Review the following expense and reimbursement data. Flag recurring round-number transactions, transactions just below the [Rp X] approval limit, duplicate invoice numbers, and weekend transactions. Present them as a list for internal audit to review, without concluding that any violation occurred.

Prompt 19: Sensitivity analysis

Using the attached profit and loss model, calculate the impact on net profit if raw material prices rise 5%, 10%, and 15%, and if the rupiah weakens 5% and 10% against the US dollar. Present the results in a two-way sensitivity table.

Prompt 20: Audit preparation

Based on the attached financial statements and notes to the financial statements, list the 10 questions external auditors are most likely to ask, along with the supporting documents to prepare for each one.

10 AI Prompt Examples for Operations Analysis

Operational problems rarely show up in a single report because the data is spread across warehouses, suppliers, the helpdesk, and HR. The ten prompts below help COOs and supply chain, customer support, and HR teams find process bottlenecks, waste, and service risks. State the data source and period in every prompt.

Inventory and Supply Chain

Prompt 21: Inventory turnover

Calculate inventory turnover and days of inventory on hand per SKU for [period]. Group SKUs into fast moving, slow moving, and dead stock using [describe criteria], then estimate the working capital tied up in slow moving and dead stock.

Prompt 22: ABC inventory classification

Run an ABC analysis on the inventory data based on annual usage value. Show the number of SKUs and the value contribution of each class, then propose different stock count frequencies and safety stock levels for classes A, B, and C.

Prompt 23: Supplier performance

From purchase order data (columns: supplier, order_date, promised_delivery_date, received_date, qty_ordered, qty_received, qty_rejected), calculate the on-time delivery percentage, order fill rate, and rejection rate per supplier. Rank suppliers and flag those below the [95]% standard.

Prompt 24: Delivery delays

Analyze delivery data for [period] to find the routes, warehouses, or logistics partners with the highest delay rates. Separate delays caused by internal processes (picking, packing) from delays caused by external parties, then show which stage contributes most to the delays.

Customer Service and Quality

Prompt 25: Ticket SLA compliance

From service ticket data (columns: ticket_id, category, priority, created_at, first_response_at, resolved_at, agent), calculate the percentage of tickets that meet the response SLA and resolution SLA by priority and by category. Flag the hours and days when SLA breaches occur most often.

Prompt 26: Customer complaint themes

Group the attached [500] customer complaints into no more than eight themes. For each theme, show the count, percentage, one representative complaint, and the department best placed to follow up. Merge themes below 2% into an "other" category.

Prompt 27: Root cause of product defects

The defect rate on the [line name] line rose from [X]% to [Y]% over the last two months. Based on the attached production, shift, operator, and raw material data, build a root cause analysis using the 5 Whys method and flag the factors most strongly linked to the increase.

People and Productivity

Prompt 28: Employee turnover

Analyze data on employees who left in the last 12 months (columns: department, tenure, level, exit_reason, last_performance_score). Calculate the turnover rate per department, find the tenure at which employees most often leave, and summarize the top three reasons. The data has been anonymized.

Prompt 29: Team workload

From project assignment data and logged working hours per team member for [period], identify members with utilization above 110% and below 60%. Propose a reallocation of tasks that still protects the deadlines of priority projects.

Prompt 30: Process efficiency

Below are time logs for each stage of [process name, for example procurement approval] for the last [200] cases. Calculate the average and median time per stage, find the stage with the greatest time variation, and recommend the stage most worth automating.

Is your operational data scattered across ERP, POS, and CRM? EDIC runs prompts like these directly on your data sources, with no need to merge files manually. See how EDIC works.

Prompts for Executive Summaries and Recommendations

A good executive summary starts with the conclusion, limits the number of points, states the impact in numbers, and separates facts from recommendations. Use the three templates below once the main analysis is done and the numbers have been checked.

One-page executive summary

Based on the analysis above, write an executive summary for the [CEO/COO/CFO] of no more than 250 words. Open with one sentence stating the main conclusion, followed by the three most important findings with their numbers, two risks, and three recommendations ranked by impact and ease of execution.

Prioritized recommendations

Organize the recommendations from this analysis into a table with these columns: recommendation, expected impact, estimated cost or effort (low/medium/high), owner, and success indicator. Separate recommendations directly supported by the data from those that are still hypotheses.

Meeting slide outline

Turn this analysis into a five-slide presentation outline for a management meeting. For each slide, write the title as a conclusion sentence, three supporting points, and the most suitable chart type.

How Do You Chain Several Prompts into One Analysis Workflow?

Prompt chaining is a technique that breaks a large analysis into several sequential prompts, with the output of one prompt feeding the next. It is usually more accurate than one very long prompt, because the result of each step can be checked first.

Example case (illustration): the Head of Sales at an FMCG distributor needs to explain why Q3 revenue came in below target.

1. Check the data. "Review this dataset. Report empty columns, duplicate records, inconsistent date formats, and extreme values before the analysis begins."
2. Describe the situation. Use Prompt 2 to see which branches are furthest from target.
3. Find the causes. "For the three branches with the largest gap, explain the most likely contributing factors, with supporting data."
4. Project forward. Use Prompt 9 to build Q4 scenarios.
5. Package it for the board. Use the executive summary template above.

The first step is the one most often skipped, even though data quality determines every step that follows. Gartner research (2020) estimates that poor data quality costs organizations an average of at least US$12.9 million per year, while 59% of organizations do not measure their data quality.

Tips for Writing Prompts That Produce Consistent Output

Consistent AI output comes from standardized prompts. That way, monthly reports prepared by different analysts can still be compared.

• Write metric definitions. Example: "Net revenue is gross revenue minus discounts and returns." Without a definition, the AI may use a different formula each time.
• Lock the output format. If a routine report always uses a table with specific columns, write the column names in the prompt.
• Ask for calculation steps and assumptions. The instruction "show the formula, calculation steps, and assumptions used" makes the results easier to check.
• Provide an example result. One sample table or paragraph is often more effective than a long description.
• Build a team prompt library. Store tested prompts in one place along with sample outputs and the date they were last updated. Assign an owner for each group, for example the finance team for finance prompts, and review them regularly so metric definitions stay in line with current policy.

How Do You Make Sure AI Answers Are Accurate?

AI can produce answers that sound convincing but do not match the data. That is why the AI Risk Management Framework from the National Institute of Standards and Technology (NIST) lists validity and reliability as core characteristics of trustworthy AI systems. Before analysis results are used for decisions, run these four checks:

1. Reconcile totals with official reports.
2. Recalculate a few samples manually.
3. Ask the AI to explain its assumptions.
4. Distinguish correlation from causation.

The strictest checks are needed for financial figures and reports to external parties. Add the following prompt at the end of every analysis:

Review your analysis. Show which numbers were calculated directly from the data, which are estimates, and which come from assumptions.

One more thing before you upload data: remove or mask identifying columns such as customer names and phone numbers, in line with Indonesia's Law No. 27 of 2022 on Personal Data Protection.

When Should Teams Move from a General Chatbot to an AI Workspace?

A general AI chatbot is enough for one-off analysis on small files. Once analysis becomes routine, draws on many data sources, needs the context of internal documents, and is done across teams, an enterprise AI workspace is a better fit, because data, prompts, and results are kept in one place with controlled access rights.

For examples of AI use by division, read our article on AI for data analysis and how businesses apply it.

Conclusion

The thirty prompts above can be used one at a time or chained into a single workflow, from data checks to the executive summary. Start with one routine report, standardize its prompt, then expand to other functions once the team is comfortable.

EDIC is an enterprise AI workspace that connects company knowledge, business data, and workflows in one place. Prompts like the ones in this article can run directly on data from ERP, POS, CRM, data marts, or uploaded Excel and PDF files. The results are turned into charts, reports, meeting-ready presentations, and auto-updating dashboards. See how EDIC works, from data source to report and slides, for an overview of the workflow.

Frequently Asked Questions about AI Prompts for Data Analysis

What is the simplest example of an AI prompt for data analysis?

"The attached data contains 2026 monthly sales by product. Show the three products with the highest and lowest month-over-month growth in a table." This prompt already includes a data description, a specific question, and an output format. Add your role and business goal to make the results more relevant.

Can these prompts be used in ChatGPT, Gemini, or Claude?

Yes. The context, data, question, and output structure works on any generative AI platform, including ChatGPT, Gemini, and Claude. What usually needs adjusting is how the data gets in, for example a file upload, a database connection, or access to internal documents.

Can AI replace data analysts?

No. AI speeds up cleaning data, recalculating figures, and drafting reports. Analysts are still needed to frame the right business questions, check the results, and interpret findings in the context of the company. Their work shifts from processing numbers to evaluating and communicating findings.

Why can AI give different answers to the same prompt?

Generative AI works on probabilities, so the wording and details of its answers can change every time it runs. This variation can be reduced with firm metric definitions, a locked output format, and a request to show the calculation steps. For important analysis, run the prompt twice and compare the numbers.

What data should not be put into an AI prompt?

Avoid personal data such as names, national ID numbers (NIK), phone numbers, and employee salaries without a legal basis and adequate safeguards. Use IDs or aggregated data if the analysis does not need individual identities. Personal data processing in Indonesia is governed by Law No. 27 of 2022 on Personal Data Protection.