What Is Wrong With Last-Year-Plus-Ten-Percent
The most common budgeting approach is extrapolation: take the prior year actuals, apply a growth assumption to revenue, apply an inflation assumption to costs, and call it a budget. This is fast. It is also nearly useless as a management tool.
When revenue comes in below budget at the end of Q1, an extrapolation budget cannot tell you why. Was the growth assumption wrong — was the market less receptive than expected? Did market conditions change — did a competitor launch something that took share? Did execution fail — did the sales team underperform against a realistic target? The extrapolation budget does not have enough structure to distinguish between these three explanations, so the management conversation circles around the variance number without reaching a conclusion about what to do differently.
A drivers-based budget built with an assumptions register has the answer immediately. Revenue was built from leads × conversion rate × average deal size. Leads came in 15% below plan. That narrows the root cause to one of three questions: was the lead volume assumption wrong, did marketing underperform, or was the market more competitive than expected? The conversation is now about a specific, addressable driver rather than a general revenue miss.
Margaret — the KissMySkills budget planning agent — builds budgets from drivers, not extrapolation. Every number traces back to a stated assumption that can be monitored, updated, and used as the basis for a precise management conversation when variance occurs.
What a Good Budget Actually Enables
A budget is not primarily a prediction. It is a decision-making framework. A well-built budget answers the questions that management faces throughout the year: can we afford to hire ahead of the revenue that justifies the headcount? What happens to the cash position if the new product launch is three months late? If revenue is 15% below plan by June, which costs can be cut without damaging the growth trajectory? What level of marketing spend is justified at the current customer acquisition cost?
None of these questions can be answered from an extrapolation budget. All of them can be answered from a drivers-based budget with documented assumptions and scenario analysis. That is the difference between a budget as a compliance exercise — something submitted to the board and filed — and a budget as a management tool used every month to make better decisions.
The Assumptions Register: The Most Underused Budget Tool
An assumptions register is a document that records every material assumption behind the budget — the growth rate and its basis for each revenue stream, the hiring plan with timing and fully-loaded cost, the variable cost ratios, the fixed cost changes, the key operating metrics being targeted. Each assumption has a stated basis (market research, historical trend, sales pipeline visibility) and a confidence level.
Margaret builds the assumptions register before building the budget numbers, because the assumptions are the budget. The numbers in the spreadsheet are the mathematical output of the assumptions. Getting the assumptions explicit and agreed first also prevents the common situation where the CEO is planning for 40% revenue growth, the CFO has modelled 25%, and the operations team has hired for 30% — all from the same budget process, with nobody having explicitly aligned on the growth assumption.
The assumptions register also makes the budget a living document rather than an annual filing. When market conditions change mid-year, the register shows exactly which assumptions need to be updated — and a revised forecast can be produced from the updated drivers without rebuilding the model from scratch.
Revenue Forecast Built from Business Drivers
A revenue forecast for a SaaS business is built from ARR at the start of the period, new ARR from new customers by segment, expansion ARR from upgrades and seat growth, and gross churn from cancellations — each driven by the metrics the sales and marketing teams actually control. A revenue forecast for a professional services firm is built from available billable days, expected utilisation rate by seniority level, and average day rate by service type. A revenue forecast for an e-commerce business is built from traffic by channel, conversion rate, and average order value.
Each is a different driver model — and the right one depends entirely on how the business generates revenue. Margaret asks how revenue works and builds the forecast from the specific drivers of that business. The result is a forecast the commercial team can understand, own, and update — because it is built from the metrics they manage, not from a finance model that requires a financial analyst to interpret.
Cost Budget: Fixed, Variable, and Headcount
Cost budgets built as a single line per category — "salaries," "marketing," "office" — cannot support the management decisions that arise during the year. Margaret builds cost budgets with the structure that makes them useful: fixed costs (those that do not change with volume, and that represent committed cash outflows regardless of revenue performance), variable costs (those that scale with revenue or volume, expressed as a ratio so they flex automatically with the top line), and headcount costs treated separately from other fixed costs because headcount decisions are the highest-leverage cost decisions most businesses make.
The headcount plan includes the timing of each planned hire, the fully-loaded cost (salary plus employer NI plus benefits plus equipment), and the revenue assumption that justifies the hire. This makes the relationship between headcount decisions and financial performance explicit — so the board can evaluate whether hiring ahead of revenue is a justified investment in capacity or a risk to the cash position.
Three Scenarios: The Budget Is Never Just One Number
A budget presented as a single set of numbers creates a false certainty. Every assumption has a range of plausible outcomes, and the compounding effect of several assumptions moving against plan simultaneously is often worse than any individual assumption suggests.
Margaret builds three scenarios for every budget: base case (the plan, built from the agreed assumptions), downside (what the P&L and cash position look like if key assumptions prove 20–30% more optimistic than reality), and upside (what becomes possible if execution is strong and market conditions are favourable). The scenarios answer the operational questions: what decisions would we make differently in each scenario, what are the early warning signs that the downside is playing out, and what costs can be reduced quickly if the downside materialises?
How to Start a Budget Session with Margaret
Load the Margaret skill file into Claude Projects. Paste the activation prompt. Margaret asks intake questions about the business model, the revenue streams and their drivers, the current cost structure, the headcount plan, and the key strategic assumptions behind the coming year. Answer specifically — the more detail provided about how the business actually works, the more useful and accurate the budget structure. Margaret works with Claude, ChatGPT, or any AI chat that accepts system prompts. The output is a complete budget framework with assumptions register, driver-based revenue forecast, scenario analysis, and headcount plan — ready for review and refinement before the board submission.
The agent behind this guide. Margaret builds a driver-based budget — assumptions register, revenue model, fixed/variable/headcount costs, and three scenarios — you can manage against all year.