Variance Analysis with AI

Budget vs actual variance analysis is a core monthly deliverable for most finance teams — and writing the narrative to explain those variances is often the most time-consuming part. AI can dramatically speed up the commentary without sacrificing quality.

This lesson covers how to use AI to explain variances clearly, accurately, and in the right tone for different audiences.

The Variance Commentary Problem

Most finance professionals are good at identifying variances. The time sink is translating them into coherent narratives:

• The same variance needs different explanations for the CFO, the board, and the operational manager
• Commentary needs to be specific, not generic ("costs were higher" isn't analysis)
• It needs to distinguish between one-off items and structural issues
• Month-end deadlines mean there's rarely enough time to write it well

AI doesn't remove the need for your financial judgment — but it takes the writing labour off your plate.

The Core Variance Commentary Prompt

"Act as a senior finance manager writing management commentary. Here are the key variances for [Month] [Year]:

• Revenue: £4.2m actual vs £4.5m budget (£300k adverse, -6.7%) Reason: Key customer order delayed to next month. Volume shortfall of 1,200 units.

• Gross Margin: 39.1% actual vs 41.0% budget Reason: Product mix shift toward lower-margin product lines. Raw material prices 3% above budget.

• Overheads: £1.1m actual vs £1.0m budget (£100k adverse) Reason: Unplanned agency staff costs to cover sick leave. IT maintenance contract renewal above budget.

Write a 200-word variance commentary for the monthly management report. The audience is the senior management team (finance-literate). Distinguish between one-off items and recurring issues. End with a 1-sentence outlook comment."

Adapting Commentary for Different Audiences

Once you have a base commentary, use AI to reformat it:

"Rewrite this commentary for a non-finance board member. Remove technical jargon. Use plain English. The key message is that we're £400k behind plan but the main driver is a delayed order which recovers next month.

[paste your commentary]"

"Condense this into 3 bullet points for an email update to the CEO. Maximum 30 words per bullet."

Structured Variance Table Analysis

If you have a full variance table, you can paste it and ask for prioritisation:

"Here is our full budget vs actual for the month:

[paste table]

Rank the variances by materiality and business significance. Identify which are:

1. One-off / timing
2. Recurring / structural
3. Require management action

Then write a priority list of the 3 variances that need the most explanation in the board report."

Before/After Example

Before AI — typical draft commentary: "Revenue was below budget due to lower volumes. Costs were higher than anticipated in several areas. Margins were impacted by mix."

After AI — refined commentary: "Revenue of £4.2m was £300k (6.7%) below budget, driven entirely by a delayed order from [Customer X] rescheduled to April — underlying volume from other customers was on track. Gross margin of 39.1% fell 190bps below budget, reflecting a temporary shift in product mix toward lower-margin custom orders and raw material prices tracking 3% above budget assumptions. Overheads exceeded plan by £100k due to short-term agency costs to cover operational sick leave (expected to self-correct) and an IT contract renewal that came in above estimate. The revenue shortfall is expected to reverse in April; the margin and overhead pressures are being monitored."

Your Turn

Take the last month-end variance report you produced. Copy the key variances and reasons into the prompt above. Run it and compare the AI commentary to what you submitted. What did it improve? What did it miss that required your specific knowledge?