Quick answer: The projection mistakes that cost sellers the most credibility are hockey-stick growth with no driver, margins that expand without a cause, working capital and capital spending left out, statements that do not agree with each other, false precision in the outer years, a forecast that contradicts the CIM, and no link to history. Each has a fix a buyer can check, such as drivers backed by documents, margins held at the normalized level, integrated statements and one version of the numbers.
Key points
- A projection loses credibility when a buyer finds one place where it disagrees with the company’s own records, and the doubt then spreads to figures that were sound.
- The common failures are growth with no driver, margins that rise without a cause, cash costs left out, statements that do not tie, false precision in outer years, a forecast that contradicts the CIM, and no link to history.
- Each failure has a mechanical repair, such as drivers with documents, margins held at the normalized level, linked statements, coarser outer years and one version of the numbers.
- A forecast year that grows faster than the best year in the company’s history needs a driver-by-driver bridge from LTM, and what the bridge cannot support belongs in an upside case.
Why one weak assumption discounts the whole model
A projection loses credibility when a buyer finds a place where it disagrees with the company’s own records. The cost is rarely limited to the line in question. A buyer that finds one assumption built backward from a target price assumes the others were too, and it replaces the seller’s case with its own, as the guide to how buyers test projections describes. The seller then loses the credit that its sound assumptions would have earned.
The mistakes recur because they share a habit, which is starting from the price the owner wants and working back to a forecast. The table pairs each mistake with what a buyer concludes from it and the repair.
| Mistake | How it shows up | What a buyer concludes | Repair |
|---|---|---|---|
| Hockey-stick growth with no driver | Growth steps up after the sale year and keeps rising, with no customer, price or capacity change behind it | The forecast was built backward from a target | Bridge each step from LTM with a document, and move unsupported growth to an upside case |
| Margins that expand with no cause | Gross or EBITDA margin rises every year as operating leverage or efficiency | Costs were set to reach a margin | Hold margin at the normalized level and move it only for a dated, documented cause |
| Working capital and capital spending left out | EBITDA grows with no balance sheet, and spending equals depreciation or the low level of recent years | The forecast hides the cash that growth uses | Forecast receivable, inventory and payable days, and schedule replacements from the asset register |
| Statements that disagree | The balance sheet does not balance, cash does not tie, formulas are typed over | Management does not control its own numbers | Link the three statements, label hard-coded cells and run implied-ratio checks |
| False precision in the outer years | Distant-year figures by line item, to the dollar, from one growth formula | Detail the company cannot defend, used for show | Forecast later years annually with fewer lines, round the figures and state the basis |
| A forecast that contradicts the CIM | The CIM calls the market mature and customers loyal while the model shows rapid growth from new accounts, or figures differ between documents | Management tells different readers different stories | Write the CIM from the model and check every figure that appears in both |
| No link to history | The first forecast year starts from a base that is not LTM, or uses different policies and adjustments | The buyer cannot test the forecast and assumes the worst | Start from normalized LTM, show history and forecast in one format and back-test the model on the past |
The guides to driver-based forecasting and to projecting margins, capital spending and working capital show how to build the repairs in the first three rows. The sections below cover the rest.
A hockey stick measured against history
The test for a hockey stick compares each forecast year’s growth with the best year in the company’s history. A forecast year that grows faster than any year the company has delivered needs a bridge from LTM, driver by driver, with a document for each step. The bridge does not have to prove the growth is certain. It has to name a cause, such as a signed contract, a price increase already in effect or capacity already bought, and it has to size it.
Illustrative example: A hockey stick against the historical trend
A company’s revenue was $9,000,000 two years ago, $9,400,000 last year and $9,800,000 in the LTM period, a gain of $400,000 a year, or about 4 percent. The management case grows far faster, with no new customer, price increase or capacity added to explain it. The trend case continues the historical gain.
| Year 1 | Year 2 | Year 3 | |
|---|---|---|---|
| Management case revenue | $10,800,000 | $13,000,000 | $16,000,000 |
| Growth over prior year | 10.2% | 20.4% | 23.1% |
| Trend case revenue | $10,200,000 | $10,600,000 | $11,000,000 |
| Revenue with no stated cause | $600,000 | $2,400,000 | $5,000,000 |
Nearly a third of third-year revenue has no stated cause. At a placeholder EBITDA margin of 15 percent, the $5,000,000 is worth $750,000 of third-year EBITDA, which a buyer usually prices as upside at best. The numbers are invented to show the mechanics and are not benchmarks.
What the bridge cannot support moves to the upside case, as the guide to base, upside and downside cases explains. A buyer may agree to pay for that growth if it arrives, through an earnout, but it usually does not pay for it at closing.
Statements that do not agree with each other
A forecast whose revenue, expenses and balance sheet do not tie tells a buyer the model was assembled to produce an answer. The quickest way to find a break is to compute implied ratios and compare them with history. A ratio that moves with no stated cause is either an error or an assumption nobody has defended.
Illustrative example: Implied ratios that expose a model
A forecast lifts revenue from $10,000,000 to $12,000,000. It keeps receivables at $1,500,000 and headcount at 40 people. Two ratios computed from the model show what it assumes.
| Check | History | Forecast | What it implies |
|---|---|---|---|
| Receivable days (receivables divided by revenue, times 365) | 55 | 46 | Customers pay faster, with no change in terms or collections behind it |
| Revenue per employee | $250,000 | $300,000 | Each person produces 20% more, with no stated reason |
At the historical receivable days, forecast receivables would be $1,800,000, so the model leaves out $300,000 of cash absorbed by growth. The figures are invented to show the mechanics and are not benchmarks.
Other breaks are mechanical. Totals that were typed over, a balance sheet that balances only because of a plug, historical columns that do not tie to the financial statements and interest lines that ignore the debt a buyer will use all send the same signal. Labeling every hard-coded input, and asking someone outside finance to trace revenue through the model to the statements, finds most of them before a buyer’s analyst does.
Outer years and the story told in the CIM
Precision in distant years suggests knowledge the company does not have. Every expense line forecast to the dollar for the last year, from one growth formula, is a rule applied for show, and a buyer reads it that way. Evidence reaches furthest where there are contracts, backlog and renewal history, so the first forecast year can be monthly and built from orders and plans. Later years can be annual, with fewer lines and rounded figures, and each should state its basis. Buyers often build their own long-range view in any case, so detail beyond the evidence adds little and invites questions.
The forecast must also agree with the story. A CIM that describes a mature market and loyal long-term customers while the model shows rapid growth from new accounts asks a buyer to believe two things at once. The reverse fails the same way, as when the CIM promotes a new product that the model leaves out. Write the narrative from the model, and give each growth claim in the CIM a matching line in the forecast. The numbers in the model, the CIM and the management presentation must be the same, including the bridge from reported to adjusted earnings.
The order in which to repair a model
Repair the model in the order a buyer tests it. First, tie it to history. Start from normalized LTM earnings, as the guide to LTM, run-rate and forward EBITDA explains, and make the model reproduce past results. Second, replace growth percentages with drivers and documents. Third, add the cash lines and link the statements. Fourth, coarsen the outer years. Last, rewrite the CIM and the presentation so that all three documents agree. A seller who works in that order has answered, before the buyer asks, the questions that the complete guide to building financial projections raises.
Frequently asked questions
What is a hockey stick projection?
A hockey stick projection is one in which results are flat or gently rising in the historical period and then bend sharply upward in the forecast, usually starting in the year after the sale. Buyers treat the bend as the point to test, because nothing in the company’s records explains it unless a contract, a price change or new capacity does.
How far out should a sale projection go?
Only as far as the evidence reaches. The first year can be built monthly from orders, contracts and plans. Later years can be annual and coarser, with fewer lines and a stated basis. Buyers often build their own long-range view, so extra years add little credit for the seller and a good deal of surface for questions.
Should a sale projection be conservative?
It should be defensible, which is different. A forecast whose drivers all have support earns credit for the growth it shows, while a deliberately low one gives away value the evidence would justify. The usual approach is a base case the owner will stand behind, with the stretch placed in an upside case that is kept separate.
What should a seller do if the projection and the CIM disagree?
Decide which one is right and change the other before any buyer sees both. The CIM narrative should be written from the model, and every figure that appears in both should match, including the bridge to adjusted EBITDA. Buyers compare documents line by line, and a mismatch tells them that management is not working from one set of numbers.
Preparing a company for sale or a buyer's diligence?
The Impetus Group offers fractional investment banking support, including earnings normalization, financial projections, CIM preparation and help navigating buyer diligence, with an engagement scoped to what the business needs.
Related guides
- How Buyers Test Your Projections in Diligence, Test by Test: How buyers evaluate financial projections in a sale: the bridge from LTM, forecast accuracy, backlog and pipeline checks, customer calls, capacity and sensitivities.
- Driver-Based Forecasting for a Business Sale: Build a Model Buyers Can Test: Driver-based forecasting builds a sale projection from units, price, utilization and retention. See driver trees for a manufacturer and a services firm.
- What Goes in a Confidential Information Memorandum (CIM): Section by Section: What is in a CIM: each section of a confidential information memorandum, from the executive summary to normalized financials, and what to hold back until later.
- Projecting Margins, Capital Spending and Working Capital for a Sale: How to forecast gross margin, operating expenses, capital spending and working capital in a sale model so a buyer can follow the path from EBITDA to free cash flow.
- Base, Upside and Downside Cases: Scenario Analysis for Sale Projections: Scenario analysis for a sale: what moves between base, upside and downside cases, how to label assumptions and how lenders and buyers use the downside.
- Management Case vs Buyer Case: How Buyers Adjust Your Projections: How a buyer turns a management case into its own case: the adjustments it makes, a reconciliation of the gap, and how sellers defend assumptions with evidence.
This guide is general education. It is not a valuation opinion, appraisal, investment advice, or legal or tax advice. How these guides are made.