Explain Budget vs. Actual Variances for the Period

Variance analysis is where a lot of monthly reporting time disappears: comparing every budget line to actuals, computing percentages, deciding what's favorable, and then guessing at causes. Most of those lines aren't worth discussing. This prompt does the arithmetic, then filters to only the variances large enough to matter. The favorable/unfavorable labeling is from a profit standpoint, which trips people up when a line comes in under budget but for a bad reason (like missed hiring slowing growth). By ranking on absolute dollar impact and asking for root-cause hypotheses, the prompt turns a wall of numbers into a short list of things to actually investigate. Crucially, it frames causes as hypotheses with a confirming question, not conclusions. That keeps you honest in front of line owners and prevents the AI from asserting a reason it can't know. Feed it accurate budget and actual figures and any context you already have.

What you can expect back

Material variances, Q2 FY2027 (ranked by $ impact):

1. Payroll: $1,100K budget vs $1,050K actual = -$50K (-4.5%), favorable. Hypothesis: the open engineering role. Ask: when do we now expect to backfill it?
2. Marketing: $200K vs $245K = +$45K (+23%), unfavorable. Hypothesis: unplanned conference booth. Ask: was the booth approved as incremental, and what pipeline did it generate?
3. Contractors: $50K vs $72K = +$22K (+44%), unfavorable. Hypothesis: contractor covering the open role. Ask: is this offsetting the payroll savings, and through when?
4. Travel: $30K vs $18K = -$12K (-40%), favorable. Hypothesis: less travel than planned. Ask: deferred or genuinely lower?

(Software +$8K is below threshold.)

Summary: The quarter ran modestly favorable on payroll due to an open role, but that saving was largely consumed by a contractor backfill and an unplanned marketing spend. Net spend was close to plan. The key question is whether the conference and contractor costs delivered proportional value.

Illustrative example — your results will vary by tool and inputs.

Get sharper results

• 01Net related lines mentally — an open role saving payroll while adding contractor cost is one story, not two; the prompt's questions help you connect them.
• 02Set the threshold high enough that you get 4–6 items, not 20; you can always lower it for a deep dive.
• 03Give context even if partial — naming the conference booth turns a vague 'marketing over' into an actionable question.
• 04Keep causes as hypotheses in your write-up too; confirming with owners before the meeting prevents awkward corrections.

Adapt it for your case