Budget and Cash Flow Analysis

Gains:

• Ability to keep budget-realization deviations and calculate and interpret them as percentages
• Ability to establish weekly projections from cash flow data and determine short-term liquidity risks
• Ability to ask scenario-based 'what-if' questions with a clear prompt

The health of a business is measured by two questions: “Are we going according to plan?” and “Is there enough money in the safe?” The first is budget vs actual analysis; The second is cash flow management. Even a company that seems profitable can struggle on payday if cash isn't managed. In this unit, we will learn to use artificial intelligence (AI) as an analysis assistant in both, but retain the decision.

What is Budget vs Actual?

A budget is the plan you put in place at the beginning of the semester; Realization is what actually happens. The difference between the two is called variance. The variance is expressed in two ways: as an amount (actual − budget) and as a percentage (difference / budget × 100). Whether it is positive or negative depends on the item: It is good if the actual income exceeds the budget, and vice versa for expenses.

The AI's job here is twofold: to calculate deviations (which you need to verify) and to interpret possible causes of large deviations (which you need to confirm).

Hint: Have the model say "sort the 5 largest deviations by absolute percentage". Thus, the overlooked but most critical items in the 100-line budget come to the top.

Step by Step: Deviation Analysis

1. Give the budget and the actual together. Item, budget, actual are three columns.
2. Keep the deviation and calculate it as a percentage. Let it also show the formula.
3. Specify the direction. Say whether it comes or goes so that "good/bad" is interpreted correctly.
4. Put a threshold. Like "Highlight deviations greater than 10%".
5. Why comment, but confirm. The reasons suggested by the model are hypotheses; You find the source.

Weak Prompt / Strong Prompt

Weak prompt:How's my budget going? [digits]

This gives a useless answer like "generally good" which doesn't indicate which pen is deviated by how much.

Strong prompt:Your role: FP&A (financial planning and analysis) specialist.Task: Analyze the budget-to-realization table below.Steps:1) For each item, calculate deviation = actual - budget (amount) and deviation% = deviation / budget * 100 (show formula).2) Label positive deviation in revenue items as "positive" and positive deviation in expense items as "negative".3) Rank the 5 largest deviations by absolute percentage.4) Each suggest 1-2 POSSIBLE reasons for the large deviation and mark them as "hypothesis".Constraint: Use only figures from the data.<data>Pencil | Budget | Realized Net Sales | 5,000,000 | 4,300,000Marketing | 400,000 | 610,000Staff | 1,200,000 | 1.180.000Logistics | 300,000 | 470,000</data>

This prompt clearly shows that sales are under 14%, marketing is over 52%, and logistics are over 57%; It opens the question that management should ask, such as "marketing spend increased but sales decreased."

Cash Flow: Difference from Profit

Profit is an accounting concept; Cash is real money in the safe. A sale is recorded as revenue when an invoice is issued, but money may arrive 60 days later. Cash flow analysis asks, "Is cash in surplus in the coming weeks?" looks for an answer to the question. AI helps in establishing a weekly cash projection from collection and payment items and marking open weeks.

Derive the weekly cash projection from the following 8-week collection and payment plan:- Each week: opening cash + collection - payment = closing cash- Mark weeks in closing cash negative as "LIQUIDITY RISK"- Summarize the earliest week of risk and the amount of the shortfall at the topStarting cash: 250,000<data> ... </data>

Caution: Cash projections are only as good as the collection assumptions in the input. If the assumption of "we will collect it in 60 days" does not come true, the project will collapse. Explicitly print the assumptions into the model and compare them with reality.

Scenario (What-if) Analysis

Management often asks "what if sales drop 10%?" he asks. AI generates these scenarios quickly:

Calculate three cases based on the following base scenario:- Optimistic: sales +10%, collection period -5 days- Baseline: no change- Pessimistic: sales -15%, collection period +15 daysFor each scenario, tabulate the 8-week minimum closing cash and risk week. List the assumptions clearly.

Scenario

sales change

lowest cash

risk week

optimistic

+10%

180,000

None

base

0

40,000

week 6 (critical)

pessimistic

−15%

−220,000

week 4 (open)

Reading Deviation Correctly: Permanent or Temporary?

Not every deviation carries the same importance. Just because an item exceeds the budget for a month does not mean that that item is permanently broken. Evaluate deviation to AI on two axes: magnitude (amount and percentage) and persistence (one-time or recurring). For example, a consultancy fee paid once may exceed the budget, but it does not affect the next month; However, increasing personnel costs repeat every month and their annual impact is much greater.

Classify each item in the deviation list below on these two axes: - Persistence: "one-off" or "recurring" (write your justification) - Annual impact: if this deviation lasts for 12 months, what will be the impact on the annual amount? Mark deviations that are recurring and have a high annual impact as "priority". Classification is a hypothesis; If there is no definitive information, add "confirmation required".

This distinction directs management's attention to what really matters. Rather than a one-time deviation of $50,000, a recurring leak of $8,000 per month ($96,000 per year) is much more critical; but when you look at the amount, the first one seems larger.

Tip: Update the cash projection every week (rolling forecast logic). Comparing last week's prediction with what actually happened will show how accurate your assumptions are and improve your prediction muscle.

Mini Cases

Case 1 — Silent logistics leak. In an e-commerce company, the logistics expense was 57% above the budget, but it was overlooked because it was a small item in total. When AI made an absolute percentage ranking, the pen came out on top; The investigation revealed that the cargo company charged additional fees outside the contract without any increase. Annual earnings are approximately 180,000 TL.

Case 2 — Payweek crisis averted. The cash projection showed that in week 6 the closing cash would drop to 40,000, whereas there was a salary payment of 320,000 TL that week. The CFO negotiated an early payment discount with the customer to bring a large collection forward one week and averted the crisis.

Case 3 — The cost of wrong assumption. One team panicked, thinking the "-220,000" gap in the AI's pessimistic scenario was real. However, the scenario was based on the assumption that collection would be delayed by 15 days, and this was unrealistic for the company. In the past 12 months, collections have never been delayed by more than 5 days. When the assumption was updated with real data, the lowest cash in the pessimistic scenario turned from -220,000 to +15,000 and the panic turned out to be unfounded. Lesson: scenario output is hypothetical; Do not take action without testing the assumption with historical data.

Common mistakes

• Not stating the income/expense direction. The model may mistake positive bias as “good” everywhere; It is the exact opposite in expenses.
• Considering the deviation only as amount. A large percentage deviation in a small item is overlooked; Sort by absolute percentage.
• Putting the reasons in the report without confirming them. The reason suggested by the model is the hypothesis; Human finds its source.
• Leaving cash assumptions private. If critical assumptions, such as collection period, are not written clearly, the projection will be misleading.
• Confusing profit with cash. A profitable month may end with a cash deficit; These are two separate analyses.

In summary

• In the budget-realization analysis, give the budget and the actual together and calculate the deviation as a percentage; You showed the formula.
• Specifying the income/expense direction and sorting by absolute percentage highlights the most critical deviations.
• The causes of deviation suggested by the model are hypotheses; The human confirms before putting it in the report.
• Cash flow is different from profit; Catch liquidity risk weeks early with weekly projections.
• Scenario analysis is powerful but entirely speculative; Print out your assumptions and compare them with reality.

Application task

Prepare a budget-realization table of 10-15 items. Have the deviations calculated and the top 5 listed with the powerful prompt template; Manually verify at least two percentage deviations. Then set up a simple 8-week cash projection, run base/pessimistic scenarios and determine the risk week. Finally, investigate and confirm or refute for yourself the reason for a deviation suggested by the model.

checklist

• [ ] I have given the budget and actuals together, item by item.
• [ ] I had the deviation calculated as a percentage and its formula.
• [ ] I specified the income/expense direction and sorted it by absolute percentage.
• [ ] I manually verified at least two percentage deviations.
• [ ] I have clearly printed the assumptions in the cash projection.
• [ ] I marked the reasons for the model's deviation as "hypothesis" and confirmed them.