Budgets, forecasts and finance are where cloud cost leaves engineering and enters the company's planning. This chapter explains how a budget for cloud spend is set and evaluated, how a forecast is produced and what it cannot know, and how billed cost arrives in accounting, procurement and the KPIs. In FinOps terms this is the point where finance and engineering have to read the same numbers: one side plans in amounts and periods, the other in usage and architecture.

What this chapter covers

The Budgets section answers what a cloud budget is and how it is held against reality: an amount for a scope and a period, compared as the period runs with the cost incurred so far and the cost projected to its end. It explains the scopes a budget can take, why a budget per owner does more than one total, how thresholds and alerts work on AWS, Azure and Google Cloud, how a variance is broken down into price, volume and mix, and how an annual budget is built from plans and commitments.

The Forecasts section answers how the cost so far becomes an expectation for the end of the period: run-rate extrapolation, trailing averages, trend with seasonality, driver-based models of planned usage times a unit cost, or a bottom-up plan per team. It shows what each method needs as input, where it fails, what no method can know, and why a forecast is a range rather than a single number. The exchange-rate question belongs here too: a company that budgets in euros and is billed in US dollars needs a rule for the rate at which budget and actual are compared.

The Finance section describes where cloud cost meets accounting: the distinction between capex and opex and the question of when cloud cost is capitalised; where cloud cost sits in the P&L, in cost of goods sold and in gross margin; accruals and the month-end close when the invoice arrives after the books are closed; procurement, contracts and renewals; and the KPIs with which finance and engineering talk about the same numbers. The pages explain the mechanism and what the providers document. Which treatment is right in a given case is for the tax adviser or the auditor to decide.

All three sections assume the same monthly routine: reconcile the numbers, explain the variance, compare budget with forecast, and record actions with an owner. How that meeting runs is the subject of Monthly cloud cost review: agenda and templates.

Key terms

  • Budget: the amount this chapter sets for a scope and a period and holds against actual and forecast.
  • Forecast: the estimate for the end of the period, with a range, recomputed with every new day of data.
  • Cost report: the view a budget reads its actuals from and a forecast reads its history from.
  • Effective cost: the cost basis teams budget on, because it spreads commitments over the usage they cover.
  • Amortized cost: the same spreading seen from the prepayment, which a budget on billed cost would otherwise see as a spike in the month of purchase.
  • Unit economics: cost per unit, the factor a driver-based forecast multiplies planned usage by.

Neighbouring chapters

The chapter sits between reporting and optimisation. Reporting and analysis, before it, supplies the reports a budget reads its actuals from and a forecast reads its history from; without knowing a report's cost basis and data cut-off, neither a budget nor a forecast can be judged. Optimisation and savings, after it, supplies the savings a plan assumes: an annual budget counts on commitments and rightsizing that are explained only there. Europe: currency, tax and sovereignty holds the currency and VAT questions the finance side of this chapter touches without answering them here. The foundations are in FinOps and cloud cost management, the mechanics of the invoice in How cloud billing works and Pricing models and commitments, and the overview of every chapter in the handbook.