The US Government Accountability Office publishes a guide on how to work out what something will cost before you commit to it. The Cost Estimating and Assessment Guide, updated in March 2020, lists the steps in order: the purpose, scope and schedule of the estimate, a technical baseline description, a work breakdown structure, ground rules and assumptions, data collection, estimating methodologies, sensitivity and risk analysis, documenting and presenting results, and updating estimates with actual costs.
Read that last one again. The process does not end when the decision is made. It ends when somebody goes back and compares the estimate to what the thing actually cost.
That is the step almost no small business runs. The forecast gets built, the decision gets made, the money gets spent, and nobody ever returns to score it. Which means that every business case an owner writes is, in the way that matters, the first one they have ever written. The arithmetic never improves, because nothing ever tells it it was wrong.
An estimate with no error rate is just a number
Here is what the missing step actually costs you.
When a bank, a partner or your own nervous gut asks “how confident are you in this?”, the honest answer depends on a fact you do not have: how far off your last three estimates were. A business that has scored its past decisions can say something real, like “I run about 20 percent optimistic on volume and close on cost.” A business that has not can only say “pretty confident,” which is not information.
This is why forecasts feel like advocacy. Without a track record, the number is just the most persuasive figure you could defend, and everyone in the room knows it. The fix is not a better spreadsheet. It is a history.
The split most cases get wrong
Before the loop can be closed it has to be worth closing, which means counting the right things in the first place. The most common error is not optimism about revenue. It is counting only the purchase.
The SBA’s own guidance splits the cost of any new undertaking into two buckets: one-time expenses and monthly expenses. One-time expenses are the initial outlays, the equipment, the designer, the permits and fees. Monthly expenses are the ones that keep arriving: salaries, rent, utilities, subscriptions. The SBA’s instruction on how far to count them is specific and more demanding than most owners expect: count at least one year of monthly expenses, and five years is ideal.
A new machine is not a $6,000 decision if it also carries $180 a month in software, servicing and floor space. It is a $6,000 decision plus $2,160 a year, forever, and the second number is the one that decides whether it pays.
The one line in the case that is not a forecast
Almost every figure in a business case is a guess. Break-even is the exception, because it is arithmetic on numbers you already know.
The SBA states the formula plainly: fixed costs divided by price minus variable costs gives the break-even point in units.
Run it on a hypothetical, invented here rather than taken from a real business. A small print shop is considering a wide-format printer at $6,000, with $180 a month in software, servicing and space. Posters sell at $45 and cost $18 in materials and ink, so each one contributes $27. First-year fixed cost is $6,000 plus $2,160, which is $8,160. Divide $8,160 by $27 and the break-even is 303 posters in year one.
Three hundred and three posters a year is about six a week.
Notice what happened to the question. “Will this printer pay for itself?” is unanswerable and invites a story. “Can we sell six posters a week?” is a question the owner can answer honestly in about four seconds, because they already know what walks through the door. Break-even converts a forecast into a fact about your own business, which is why it belongs near the top of the case rather than in an appendix.
Write the range, and write what would stop you
GAO’s list includes sensitivity and risk analysis for a reason. A single number cannot be wrong in any useful way, so it never gets tested and never teaches you anything.
The small-business version of that step is two lines, and neither requires software. First, write every uncertain figure as a range rather than a point: six to eleven posters a week, not “about eight.” Second, write down in advance what result would make you stop, and when you will look. Something like: if we are under four a week at the ninety-day mark, we sell the printer rather than buy more marketing for it.
That second line is the one people resist, because writing it feels like planning to fail. It is the opposite. A decision with no stated stop condition cannot fail, it can only be defended, and the defending is what turns a $6,000 mistake into a $20,000 one. Sizing that exposure before you commit is its own discipline, and we worked through how to size a bet you can afford to lose separately.
The half page that makes the next one better
Now the step GAO ends on, in the form a small business can actually run.
Put a date in the calendar for twelve months after the decision. On that date, write half a page with four things on it: what you estimated the one-time cost would be and what it actually was, what you estimated the monthly cost would be and what it actually was, what volume you forecast and what volume you got, and one sentence on which assumption was furthest off.
That is the whole exercise. It takes twenty minutes and it is the most valuable twenty minutes in the process, because it is the only part that compounds.
The first one you write will feel useless, and this is the part worth arguing for hardest. It will read like a report card on a decision you cannot change. But its value was never in the decision it scores. It is in the fact that your next business case now has something no competitor’s has: a measured tendency. You will know whether you run hot on volume or on cost, in which direction and roughly by how much, and you can correct for it before you show the number to anybody. Three of these and you are no longer guessing about your own guessing.
Keeping the volume side honest over time is easier when you are already tracking something countable, which is the argument in picking one growth metric you can already count, and the cost side is steadier when your prices were built on real margins to begin with, covered in pricing for profit and cash flow.
What to build it in
The estimate and the score need to live in the same place, or the comparison will not happen.
A spreadsheet is genuinely sufficient, and for most owners it is the right answer. Google Workspace Business Starter lists at $7 per user per month and Business Standard at $14, with Business Plus at $22, all excluding tax, and Google is currently running 50 percent off those rates for the first three months, a promotion its own pricing page shows running through January 5, 2027. Checked September 21, 2026. One tab for the estimate, one for the score, and a calendar reminder does the rest.
If the case is going to a lender and you would rather hand over formatted statements than a spreadsheet, LivePlan publishes Standard at $20 a month paid monthly or $15 paid annually, and Premium at $40 monthly or $30 annually, with a 35-day money-back guarantee, on its pricing page, also checked September 21, 2026. Premium is the tier that carries scenario comparison, which matters only if you are genuinely modeling more than one version.
If you would rather talk the assumptions through before committing them to a grid, the Innovation ROI Calculator prompt at BusinessPrompter.com is built to walk a non-specialist through the inputs this article assumes you have gathered: the one-time cost, the recurring cost, the unit economics and the probability you are willing to defend. It structures the thinking. It cannot tell you what your posters cost to print, and it has no way to score last year’s decision for you, which remains the part only your own records can supply.
That boundary is worth stating plainly. None of this replaces an accountant on the tax treatment of a capital purchase, or a lender’s own underwriting. What it replaces is the habit of deciding on a feeling and never finding out whether the feeling was any good.
There is also a reason this matters beyond the arithmetic. Owners who cannot tell a good bet from a bad one stop making bets at all, and the business quietly narrows to whatever is already safe. A case you can actually check is what makes the next investment defensible, including the ones that add a person rather than a machine. Knowing your own error rate is what lets a small team commit to something ambitious instead of talking itself out of everything.
Start here this week
Take a decision you made last year that cost real money. Write down what you thought it would cost, what it actually cost, and what you thought you would sell against what you sold. Do not fix anything and do not show it to anyone. That half page is the first data point you have ever had about how you estimate, and it makes every case you write from here on measurably less of a guess.
Frequently Asked Questions
What should a small business case actually contain?
Four things, at minimum: the one-time cost, the recurring cost counted over at least a year, the break-even in units, and the condition that would make you stop. The SBA’s guidance on splitting costs into one-time and monthly expenses is the part most owners shorten, and it recommends counting at least one year of monthly expenses, with five years ideal.
How do I calculate break-even?
Divide fixed costs by the selling price minus the variable cost per unit, which the SBA publishes as the break-even point in units. Include both the one-time outlay and a year of recurring cost in the fixed figure, or the result will flatter the decision.
Why bother reviewing a decision I cannot reverse?
Because you are not scoring the decision, you are calibrating yourself. The federal Cost Estimating and Assessment Guide ends its process on updating estimates with actual costs for the same reason. Once you know whether you typically run optimistic on volume or on cost, and by roughly how much, every future estimate can be corrected before anyone else sees it.
Do I need software to write a business case?
No. A spreadsheet with one tab for the estimate and one for the twelve-month score is enough, and the discipline matters far more than the tool. Paid planning software is worth considering mainly when the output is going to a lender who expects formatted financial statements.
How long should the forecast run?
Long enough to include the recurring costs, which is the part short forecasts hide. One year is the working minimum on the SBA’s guidance. Beyond about three years the variance grows faster than the insight for most small businesses, so a longer horizon is worth building only when the commitment itself is long, such as a lease or financed equipment.
