Your Forecast Is Not a Plan. It Is a Test of Management Honesty.

# Your Forecast Is Not a Plan. It Is a Test of Management Honesty. **Direct answer:** A forecast is not a promise, a budget, or a motivational target. It is a structured estimate of what may happen, based on assumptions, evidence, uncertainty, and operating signals. When a founder asks for a number and the team learns that optimism is safer than accuracy, the forecast becomes theatre. The spreadsheet looks precise. The business is not. McKinsey surveyed 130 CFOs and found that less than half used any given form of nonfinancial internal data in their forecasts. Only 35% used external market data, and only 18% used extra signals such as weather or traffic. About 40% said their forecasts were not particularly accurate and the process took too much time [1]. That is not only a finance problem. It is a management-design problem. ## The forecast is where honesty becomes measurable A forecast says what leaders currently believe will happen. A plan says what the company intends to do. A budget allocates resources. A target sets ambition. These are different instruments. Confusing them creates bad decisions. If a target is presented as a forecast, the board cannot see the gap between ambition and evidence. If the budget is presented as a forecast, management may defend last year’s assumptions instead of responding to new information. If the plan is presented as a forecast, leaders may hide the choices and dependencies required to make the number real. A decision-grade forecast makes the assumptions visible: * What must be true for the number to happen? * Which operational signal will show whether it is happening? * What range is plausible? * What would make us change the view? * Who owns the response if the signal moves? A number without assumptions is not precision. It is concealment. ## Why finance-only forecasts fail McKinsey describes a common failure pattern: managers are held to earnings targets, so they find one or two ways to make the number. Quality, customer retention, and operating efficiency deteriorate. The bottom line looks strong for a while, then performance becomes stuck in low gear [1]. This is how a forecast can become a self-fulfilling problem. People optimise the metric used to judge them, even when the metric is a lagging indicator. Revenue is late. Churn is late. Margin is late. Cash is late. The leading signals often live elsewhere: qualified pipeline, activation, usage, renewal intent, delivery capacity, defect rates, collection days, support volume, hiring time, and supplier reliability. A CEO who sees only the financial output is not seeing the business. They are seeing its delayed reflection. ## Forecast theatre versus decision-grade forecasting | Question | Forecast theatre | Decision-grade forecast | |---|---|---| | Purpose | Defend a number | Improve a decision | | Number | One precise point | Base, upside, and downside range | | Assumptions | Hidden in the model | Written, owned, and reviewable | | Data | Mostly financial history | Financial, operational, and external signals | | Horizon | Annual static view | Rolling updates plus scenario tests | | Incentive | Reward optimistic commitment | Reward useful accuracy and early warning | | Variance | Explained after failure | Used to change action before failure | | Ownership | Finance owns the file | Operators own drivers; finance integrates | | Challenge | Political consensus | Independent challenge and evidence | | Output | A report | A decision, trigger, or resource move | ## The four ways founders corrupt a forecast ### 1. They ask for commitment before evidence “Give me the number you will deliver” is a target question, not a forecast question. It tells the team to negotiate. Ask two questions separately: “What do you believe will happen?” and “What do we want to make happen?” The difference is the management gap. ### 2. They reward optimism and punish updates If a manager who raises a risk is labelled negative, risk data disappears. If a manager who misses a forecast is punished more than one who hides a problem until the end of the quarter, the company gets silence instead of accuracy. Harvard Business School research identifies intentional bias from incentive misalignment and power, as well as unintentional bias from information and procedural blind spots [3]. The forecasting process must be designed to surface both. ### 3. They use the business plan as the forecast McKinsey calls this an echo chamber: the plan is rolled up, repeated, and mistaken for an unbiased view of what the market is doing [1]. A forecast should contain a momentum case based on internal and external evidence, then show how initiatives change that base case. ### 4. They review variance too late A variance report after the quarter is useful for learning but too late for steering. The forecast must name early indicators and trigger points. If qualified pipeline falls below a threshold, hiring changes. If activation falls, the product plan changes. If collections slow, spending changes. A forecast without a trigger is a diary entry. ## What the evidence says about better forecasting McKinsey recommends rolling forecasts that update inputs predictably as conditions change [1]. In its forecasting research, companies using more real-world operating data improved visibility into bottom-line issues before they became large problems. McKinsey also reports that AI-driven forecasting can reduce errors by 20% to 50% in cited supply-chain examples. One call-centre approach improved volume accuracy by almost 10%, reduced costs by about 10% to 15%, and improved service levels by 5% to 10% [2]. These numbers are not a licence to buy software. They are evidence that forecasting improves when the process uses better signals, tests models, and prepares for uncertainty. HBS research reaches the same management conclusion from another angle: an independent group responsible for managing the forecasting process can stabilise political conflict, while clear roles, information exchange, and explicit assumptions help manage bias [3]. ## The founder’s forecast operating system A small company does not need a large FP&A department. It needs discipline. Start with a **momentum case**: what is likely if the company changes nothing? Use internal operating data and external market signals. Add a **management case**: what changes if the company executes the plan? Name each initiative, owner, timing, dependency, and expected effect. Add **scenario bands**: what happens if the key assumption is wrong? Use base, downside, and upside cases. Do not hide uncertainty inside one decimal point. Define **leading indicators**: the few weekly or monthly signals that move before the financial result. Link each indicator to a decision trigger. Run a **variance review**: compare forecast to actual, but also compare assumptions to reality. Was the price wrong? Was the conversion rate wrong? Was the timing wrong? Was the owner late? Each answer needs a different response. Protect **forecast independence**: the person who owns the target should not be the only person who controls the forecast. A challenge function is not a lack of trust. It is a defence against blind spots.

## Frequently asked questions ### What is the difference between a forecast and a budget? A budget allocates resources for a period. A forecast estimates what is likely to happen based on current evidence. The budget may stay fixed while the forecast changes. ### What is the difference between a forecast and a target? A target is an ambition or performance objective. A forecast is a belief about the likely outcome. Treating the target as the forecast hides the gap between desire and evidence. ### How often should a startup update its forecast? Update it at the pace of the business. A company with fast-moving sales, cash, or usage signals may update weekly or monthly. The important rule is predictable updates tied to real operating changes, not random revisions to make the number look good. ### Should a founder use AI for forecasting? AI can help when it is connected to clean operational data, tested against reality, and used with scenario analysis. It cannot fix unclear definitions, biased incentives, or missing ownership. Better software does not replace a better process. ### What makes a forecast honest? An honest forecast separates what leaders want from what evidence supports. It shows assumptions, ranges, leading indicators, uncertainty, and the actions that will change the result. ### Who should own the forecast? Operators should own the drivers they control. Finance should integrate the model, challenge assumptions, and explain the variance. The CEO owns the decisions made from the forecast. ### What should the board ask? Ask what changed since the last forecast, which assumption is weakest, what leading indicator will move first, what downside trigger is active, and what management will do before the quarter ends. ## Final verdict A forecast is not a promise to the board. It is a test of whether the leadership team can separate evidence from ambition. **If your forecast cannot show assumptions, ranges, leading signals, and decision triggers, it is not planning. It is organised optimism.** ## References [1] [McKinsey — Bringing the real world into your forecasting process](https://www.mckinsey.com/capabilities/strategy-and-corporate-finance/our-insights/bringing-a-real-world-edge-to-forecasting) [2] [McKinsey — AI-driven operations forecasting in data-light environments](https://www.mckinsey.com/capabilities/operations/our-insights/ai-driven-operations-forecasting-in-data-light-environments) [3] [Harvard Business School — Managing Functional Biases in Organizational Forecasts](https://www.hbs.edu/faculty/Pages/item.aspx?num=22756) [4] [Harvard Business Review — Sales Teams Aren’t Great at Forecasting. Here’s How to Fix That.](https://hbr.org/2019/03/sales-teams-arent-great-at-forecasting-heres-how-to-fix-that)