The bill went up, finance noticed, and you have been asked to explain it. The technical investigation is one problem; the conversation is a different one, and doing the first well does not automatically get you through the second.
What they are actually asking
"Why did the bill go up" is rarely the real question. Underneath it are four:
How much, exactly? In money, against what was budgeted.
What caused it? In terms of a business activity, not a service name.
Is it permanent? A one-off migration and a new baseline require completely different responses.
What happens next month? This is the one that determines whether the conversation ends well.
Answer these four in order and you will rarely be asked a fifth.
Lead with the ratio, not the total
The single most useful reframing is to present spend relative to something that grew.
An infrastructure bill rising 40% sounds alarming. The same bill rising 40% while customers grew 55% is a margin improvement, and it is the same fact described accurately.
Pick whichever denominator your business already uses — revenue, customers, active users, transactions. Cost per unit is the number finance thinks in, and presenting it does two things: it answers the growth-versus-waste question immediately, and it demonstrates that you are thinking about the same thing they are.
If the ratio moved the wrong way, say so plainly. "Spend grew faster than customers this month, and here is why" is a much stronger position than a defended total.
Translate services into activities
"EC2 is up $8,000" means nothing to finance. It also, subtly, sounds like an excuse — a technical noun offered in place of a reason.
Translate to what the business did:
- "We onboarded three enterprise customers, and their data volume increased storage and processing cost by $8,000 a month."
- "A batch job ran for six days after a failed migration before we caught it. One-off, $4,000, fixed."
- "We moved image processing to a larger instance type to cut latency. Permanent, $3,000 a month, and it reduced support tickets."
Each names a business activity, a figure, and whether it recurs. That is a complete answer.
Separate the three types of increase
Every cost increase falls into one of three categories, and they warrant different responses.
Growth. More customers, more usage, more data. Expected, healthy, and the response is to confirm unit economics still work.
Investment. A deliberate decision that costs more and buys something — performance, reliability, capability. The response is to name the benefit.
Waste. A leak, a mistake, something left running. The response is what you have fixed and how you will catch the next one sooner.
Being explicit about which bucket each element falls into is what makes the explanation credible. A CFO who hears everything defended as growth stops believing the growth number.
When you do not know the cause yet
Sometimes the question arrives before the investigation is done. Do not guess, and do not go quiet.
Bring three things: what you know, what you are checking, and when you will have the answer. "It is concentrated in one service and one account, started on the eleventh, and I will have the cause by Thursday" is a professional answer.
The failure mode is a plausible theory offered as fact. If you say it was a migration and it turns out to be a leak, the next conversation starts from a worse position than not knowing did.
What to bring
Keep it to one page.
- Total against budget, and the variance
- Cost per customer or per unit, this month against last
- The top three drivers of change, in business terms
- Which bucket each falls into: growth, investment, waste
- The forecast for next month, with a range
- What you have changed so it is caught earlier next time
The last line matters more than people expect. What finance is really assessing is whether this will happen again without warning.
Make the next conversation easier
Most of the difficulty here is that the explanation is assembled retrospectively, from several consoles, weeks after the events. The numbers are gathered under pressure and the causes are reconstructed from memory.
The alternative is a standing daily signal with a per-service baseline, so a change is noticed and attributed on the day it happens. Then the monthly conversation is a summary of things already known rather than an investigation with an audience.
StackSpend consolidates every provider into one total with categories finance recognises, forecasts month-end with a confidence range, and flags deviations the day they appear. The conversation improves mostly because it stops being a surprise to you.
FAQ
What does a CFO want to know about a cloud bill increase?
How much it increased against budget, what business activity caused it, whether the increase is permanent or one-off, and what next month looks like. Presenting cost per customer alongside the total answers the growth-versus-waste question before it is asked.
How do I explain a cost increase I have not diagnosed yet?
State what you know, what you are checking, and when you will have an answer. A plausible guess that turns out wrong damages credibility far more than an honest "not yet".
Should I show finance the provider dashboard?
Generally no. Provider consoles use service names and structures that require translation, and handing them over shifts interpretation onto someone without the context. Bring a one-page summary in business terms instead.
How do I show that a cost increase was worth it?
Pair it with the outcome it bought and the unit economics. An increase that improved margin per customer, cut latency, or removed a reliability risk is an investment; naming the benefit converts it from a variance into a decision.
