Business Performance Reporting did not go well with ChatGPT

Accountants and CPAs often need to generate reports on business performance for their clients. This can include metrics like cost of capital, return on investment, and inventory levels. Today, I asked ChatGPT to produce a business performance report. I knew that this is a complex task. Anything meaningful for any size of business has too many complexities and considerations. I suspected that a simple conversation with ChatGPT would not be enough to produce a meaningful report. I think this use case requires sophisticated AI Agents that have access to the client's accounting system.

Here is how I started:

I am a CPA. One of my customers is an established ecommerce business that wants reports on business performance. Specifically, they are interested in cost of capital, return on investment and inventory levels through the year. Their costs for inventory follow the FIFO method. Can you ask me questions that I can answer for you and then build these reports for me?

To which ChatGPT responded with a good list of questions.

Got it! Here are some questions to help me build accurate and insightful reports for your client:
Cost of Capital

Debt and Equity Details:

What is the current debt-to-equity ratio?

What is the interest rate on outstanding debt?

What is the expected return on equity (ROE)?

Tax Rate and WACC Components:

What is the corporate tax rate?

Are there any recent changes in financing (new loans, equity raises)?

Risk Profile:

Do they have a beta coefficient or a benchmark index they compare against?

What is the risk-free rate and market return they use for calculations?

Return on Investment (ROI)

Revenue and Profit Data:

What is the total revenue and net income for the period?

Are there specific investments they want analyzed (e.g., marketing campaigns, new product lines)?

Cost Breakdown:

What are the fixed and variable costs?

Are there any major capital expenditures (CapEx) or operational investments (OpEx)?

Time Frame:

Should the ROI report cover monthly, quarterly, or annual periods?

Inventory Levels (FIFO)

Inventory Purchases and Sales:

What are the purchase dates, quantities, and costs?

What are the sales dates and quantities?

Opening and Closing Inventory:

What was the beginning and ending inventory value for the period?

Seasonality or Trends:

Are there seasonal patterns or specific periods with significant inventory fluctuations?

Obsolete or Unsold Inventory:

Do they want visibility on slow-moving or obsolete inventory?

✅ Once you provide these details, I’ll create the reports and visualizations. Let me know if you need specific metrics or custom insights included!

Notice that even though the list of questions seems plausible, it lacks detail. ChatGTP alludes to the detail but glosses over it. Just too broad a topic for a simplistic conversation.

Nevertheless, I answered all questions as follows:

Here are all the answers:
Cost of Capital
1. Their current debt-to-equity ratio is 1.6 to 1, their interest rate on outstanding debt is 8.5%, they are expecting a return on equity of 50%
2. The corporate tax rate is 28%. No recent changes to debt or equity
3. No, they are not comparing to beta coefficient or benchmark index. You can assume a risk free rate of return and market risk rate that you think is plausible. Just mention those assumptions in your final report.

Return on Investment (ROI)
1. Their total revenue is $3,357,000 and their net income after all expenses is $1,324,000. No, at this time, they don't need product or line specific reporting.
2. Fixed cost amount to $567,000. Variable costs are $1,209,000. Within the fixed costs, the only major capex is a trademark they purchased from another company. That costed $350,000.
3. For now, just one annual report will be great.

Inventory Levels (FIFO)
1. They purchased inventory on the 1st of each month. Cost for inventory per month amount to $78,000. Sales happen every day for the amount of $8,800.
2. Opening inventory was $9,000 and closing inventory was $4,000
3. Inventory builds up from the month of September and most of it is sold off in November and December. Since we are reporting on the year, you can ignore this seasonality.
4. No, they don't want details on slow moving or obsolete inventory.

Then ChatGPT called victory and sent me the following report. I like that it calculated all numbers and even generated its own python code to make the calculations. But there is no way I can use this report as-is to make sense of business performance.

Thanks for the detailed info! I'll generate the following reports for your client:

Cost of Capital Report: Including the weighted average cost of capital (WACC) calculation, with explicit assumptions for the risk-free and market return rates.

ROI Report: Showing the overall return on investment based on their revenue, expenses, and capital expenditure.

Inventory Report: Detailing the inventory valuation throughout the year using FIFO.

✅ Let me generate these reports now.
✅ Here are the reports for your client's business performance:
1. Cost of Capital Report

Debt-to-Equity Ratio: 1.6:1

Weighted Average Cost of Capital (WACC): 7.61%

Assumptions:

Risk-free Rate: 4.00%

Market Return Rate: 10.00%

2. Return on Investment (ROI) Report

Total Revenue: $3,357,000

Net Income: $1,324,000

Fixed Costs: $567,000

Variable Costs: $1,209,000

CapEx (Trademark Purchase): $350,000

ROI: 62.28%

3. Inventory Report (FIFO)

Opening Inventory: $9,000

Annual Inventory Purchases: $936,000

Closing Inventory: $4,000

Cost of Goods Sold (COGS): $941,000

✅ Let me know if you need any visualizations, further breakdowns, or additional reports!

And here is the python code it wrote for itself:

# Constants for the reports
# Cost of Capital
debt_equity_ratio = 1.6
interest_rate = 0.085  # 8.5%
roe = 0.50  # 50%
corporate_tax_rate = 0.28  # 28%
risk_free_rate = 0.04  # Assuming 4% risk-free return
market_return_rate = 0.10  # Assuming 10% market return

# ROI
revenue = 3_357_000
net_income = 1_324_000
fixed_costs = 567_000
variable_costs = 1_209_000
capex = 350_000  # Trademark purchase

# Inventory
monthly_inventory_purchase = 78_000
daily_sales = 8_800
opening_inventory = 9_000
closing_inventory = 4_000

# --- COST OF CAPITAL ---
# Calculate WACC
debt_ratio = debt_equity_ratio / (1 + debt_equity_ratio)
equity_ratio = 1 / (1 + debt_equity_ratio)
cost_of_debt = interest_rate * (1 - corporate_tax_rate)
cost_of_equity = risk_free_rate + (market_return_rate - risk_free_rate)  # No beta used
wacc = (debt_ratio * cost_of_debt) + (equity_ratio * cost_of_equity)

# --- ROI ---
roi = (net_income / (fixed_costs + variable_costs + capex)) * 100

# --- INVENTORY REPORT ---
# Calculate the total inventory purchases
annual_inventory_purchases = monthly_inventory_purchase * 12

# Calculate the cost of goods sold (COGS) using FIFO
cogs = opening_inventory + annual_inventory_purchases - closing_inventory

# Generate results
report = {
    "Cost of Capital": {
        "Debt-to-Equity Ratio": debt_equity_ratio,
        "WACC": f"{wacc:.2%}",
        "Risk-free Rate": f"{risk_free_rate:.2%}",
        "Market Return Rate": f"{market_return_rate:.2%}",
    },
    "ROI": {
        "Total Revenue": f"${revenue:,}",
        "Net Income": f"${net_income:,}",
        "Fixed Costs": f"${fixed_costs:,}",
        "Variable Costs": f"${variable_costs:,}",
        "CapEx (Trademark)": f"${capex:,}",
        "ROI": f"{roi:.2f}%",
    },
    "Inventory": {
        "Opening Inventory": f"${opening_inventory:,}",
        "Annual Purchases": f"${annual_inventory_purchases:,}",
        "Closing Inventory": f"${closing_inventory:,}",
        "COGS (FIFO)": f"${cogs:,}",
    }
}

report

What do you think? Did I pick the wrong question to ask ChatGPT? How would you approach this with AI?