There is a specific way small businesses die, and it is not the one people picture. It is not a bad quarter or a lost customer. It is a Tuesday where payroll is $41,000, the bank account holds $28,000, and the $60,000 invoice that would have covered it is sitting in a client’s accounts-payable queue on net-45 terms, exactly where the owner put it.
Nothing was wrong with the business. The margin was fine. The pipeline was fine. What failed was timing, and timing is invisible in a profit-and-loss statement because a P&L records the sale on the day you invoice it, not the day the money lands. Your accountant is not going to catch this for you. By the time the quarterly statements arrive, the Tuesday already happened.
The tool that catches it is the 13-week cash flow forecast. It is what restructuring advisors build in the first 48 hours of any engagement, and it is the single most useful spreadsheet a small operator can own. It takes about two hours to build and twenty minutes a week to keep alive.
Why 13 weeks, specifically
Thirteen weeks is one quarter. That length is not a tradition, it is a compromise between two failure modes.
Shorter than about eight weeks and the forecast is useless as a decision tool: every lever that fixes a cash gap — collecting early, renegotiating a vendor, drawing a credit line, delaying a hire — needs more runway than that to work. You would be reading a diagnosis you can no longer treat.
Longer than about 16 weeks and the accuracy collapses. You are guessing at deals that have not closed, and the model starts producing precise-looking numbers built on fiction, which is worse than no model at all.
Thirteen weeks is far enough out to act and close enough in that most of what you are forecasting already exists as a signed contract, an outstanding invoice, or a recurring bill. That last point is the whole trick: a good 13-week forecast is mostly a record of commitments already made, not a prediction.
The structure
One sheet. Columns are the next 13 weeks, labeled by the Monday date. Rows are the following, in this order:
- Opening cash — the previous week’s closing balance.
- Inflows, itemized: collections from existing invoices, deposits on new work, recurring or subscription revenue, other (tax refunds, grants, rebates).
- Total inflows.
- Outflows, in four buckets (below).
- Total outflows.
- Net movement — inflows minus outflows.
- Closing cash — opening plus net.
- Cash floor — a fixed line you set once (below).
- Headroom — closing cash minus floor. This is the only number that matters.
Weekly buckets, not monthly. Monthly buckets hide the entire problem: payroll on the 15th and a big receipt on the 28th net out to a comfortable month and conceal two weeks of insolvency in the middle.
Step 1: Opening cash, honestly
Start with the bank balance, then subtract anything in it that is not yours to spend: sales tax and VAT collected on behalf of the government, payroll taxes withheld and not yet remitted, client deposits for work not yet started, retainers held in trust. What is left is your actual cash.
Do not include an undrawn credit line here. A credit line is a lever you pull when the forecast goes red, and if you count it as cash you will never see the red. Note the available limit in a cell off to the side, labeled as what it is: a lever.
Step 2: Inflows, invoice by invoice
This is where every bad forecast is born. Do not model revenue. Model collections — the specific dates specific money is likely to arrive.
Pull your accounts-receivable aging report and place each open invoice in the week you actually expect payment, not the week the terms say. Then apply a haircut based on how late it already is. A reasonable default set to start from and calibrate against your own history:
- Current, not yet due: 95% of face value, in the week after the due date. Almost nobody pays on the due date.
- 1–30 days past due: 85%, spread across the next two weeks.
- 31–60 days: 70%.
- 61–90 days: 50%.
- Over 90 days: zero. Not because it is uncollectable, but because a forecast that depends on it is a forecast that will be wrong. If it lands, it is a happy variance. Chase it separately — the get-paid playbook covers how without torching the relationship.
For unsigned work, use a stricter rule than your sales optimism allows: include it only if there is a signed contract or a written commitment, and only at the deposit amount, in the week the deposit is contractually due. Verbal “we’re definitely doing this in September” is worth zero in this sheet. Put it in your pipeline tracker instead, where it belongs.
Step 3: Outflows, in four buckets
Bucket 1 — Payroll and contractors. Exact dates. Include employer taxes and benefits, which run 10–25% above gross depending on jurisdiction. If you pay semi-monthly, three of your 13 weeks contain two payroll runs. Find them now.
Bucket 2 — Fixed recurring. Rent, insurance, software, loan payments, utilities. Boring, predictable, and the easiest bucket to get right.
Bucket 3 — Variable and cost of delivery. Materials, subcontractors, ad spend, merchant fees, shipping. Tie these to the work actually in the calendar, not to an average. If Q3 delivery is front-loaded, so is this line.
Bucket 4 — The landmines. This is the bucket that actually kills people, because these items appear only once or twice a year and therefore never make it into anyone’s mental model of “normal monthly spend.” Go find them. Estimated quarterly tax payments. Annual insurance renewal. Annual software contracts that auto-renew. Equipment deposits. Professional fees at year-end. Bonus accruals. The security deposit on the new lease.
Open your bank statements for the last 12 months, sort by amount descending, and look at every single transaction over $1,000. You will find three or four things you had completely forgotten. Put them in the sheet.
Step 4: Set the floor
Your cash floor is the balance below which you are no longer running a business, you are managing a crisis. A workable default is four weeks of fixed outflows — payroll plus bucket 2. If that number is $52,000, your floor is $52,000 and closing cash of $60,000 does not mean you have $60,000. It means you have $8,000 of headroom.
Headroom is the number you carry in your head. Not revenue, not the bank balance. Headroom, and the week it goes negative.
The point of the sheet is not accuracy. It is lead time. A forecast that says “week seven is short about $30,000” is doing its entire job even if the real gap turns out to be $19,000 or $44,000. You now have six weeks to work the problem, which is the difference between a phone call and a fire sale. Operators chase decimal-point precision and miss this: the value is in the warning, not the estimate.
The worked example
A six-person design agency. Roughly $1.6 million a year in billings, healthy 18% net margin, and the owner has never missed payroll. Opening cash: $84,000. Floor: four weeks of fixed costs, $58,000. Headroom on day one: $26,000.
Weeks 1 through 5 look ordinary. Payroll runs $31,000 semi-monthly, fixed costs about $8,500 a week, collections arrive at a rough $40,000 clip. Closing cash bounces between $70,000 and $95,000. Nobody is worried.
Then the sheet does its job. Three things stack into a nine-day window in weeks 6 and 7:
- The quarterly estimated tax payment: $34,000, due the 15th.
- The annual professional-liability renewal: $11,200, which the owner had genuinely forgotten because it was last paid 51 weeks ago.
- The largest client, who represents 31% of billings, moved to net-60 terms in a contract renewal the account lead signed in April without flagging it. Two invoices totaling $96,000 that the owner mentally filed in week 6 now land in week 10.
Closing cash in week 7: $19,000. Against a $58,000 floor, that is headroom of negative $39,000 — and a payroll run on the following Monday that the account does not cover.
The business is profitable the entire time. It is profitable in week 7. Profitability is not the question the sheet asks.
Four levers, ranked by what they cost you
When a week goes red, you have exactly four moves. Work them in this order, because the order is the cost order — cheapest first.
1. Accelerate inflows. Cheapest, because it uses money you have already earned. Call the three largest open invoices personally, not by email. Offer a 2% discount for payment inside five days on anything over $25,000 — on $96,000 that is $1,920 to move six weeks of timing, which is a bargain against any form of borrowing. Invoice work-in-progress immediately rather than at project completion. On new contracts, restructure to 40% deposit rather than 100% on delivery.
2. Delay outflows. Nearly free if you do it honestly and in advance. Call vendors before the due date, never after. Ask for 30 extra days, once, with a specific date attached. Vendors say yes far more often than people expect, because a customer who calls proactively is a customer they are not going to lose. Move discretionary spend — ad tests, equipment, conferences, that hire — to week 11 or later.
3. Cut. Costs real capability. Freeze new spend, pause a marketing channel, defer the hire. Do this only after the first two levers are exhausted, because cuts made in a cash panic are almost always the wrong cuts.
4. Finance. Most expensive, and the only lever with a permanent tail. Draw the credit line, take an equipment loan, factor the receivable. If you get here, borrow against a gap you can see in the forecast and can articulate to the lender — that conversation goes very differently than the one where you need money by Friday.
Scripts that do the work
Early payment, to a client: “We’re closing out our quarter and I’d rather give you the discount than the bank. If invoice 2214 clears by the 8th, take 2% off it. Same offer stands on the next one if it’s useful to you.”
Extension, to a vendor: “I want to give you a heads-up rather than have you chase me. Our largest client moved us to net-60 and it’s put a bubble in August. I can pay the September 3rd invoice on October 3rd, in full, and everything after that stays on schedule. Does that work?”
Credit line, to a banker: “We’re profitable and growing, and a payment-terms change created a six-week working-capital gap. I have a 13-week forecast that shows the trough and the recovery. I’d like to put a facility in place now, while nothing is on fire, and I expect to use it for about eight weeks.”
Note what all three have in common: a specific number, a specific date, and a reason that is not an apology. That posture comes directly from having the sheet.
The Friday ritual
The forecast is worthless as an artifact and valuable as a habit. Twenty minutes, same time every week, ideally Friday morning when the week’s banking has settled.
- Enter actuals for the week that just ended, next to the forecast. Keep both.
- Calculate variance on the two lines that matter: total collections and total outflows. Anything over 10% off, write one sentence explaining why. That sentence is how the model learns.
- Roll the window forward — drop the completed week, add a new week 13 at the far end. The forecast always looks a quarter ahead, permanently.
- Re-age receivables and reapply the haircuts. Invoices slide down the aging buckets whether you look or not.
- Read the headroom row and find the lowest week. If it is negative, or within 20% of the floor, pick a lever today.
After six or eight weeks of doing this, your collection haircuts stop being generic defaults and become your actual, observed payment behavior. That is when the forecast gets genuinely good — and it is why starting with imperfect assumptions today beats waiting for perfect ones.
Five ways people wreck it
- Forecasting revenue instead of collections. The most common and most fatal error. Revenue is a promise, cash is a fact, and the gap between them is measured in weeks.
- Averaging. “We collect about $160,000 a month” spread evenly across four weeks produces a smooth line that has never once matched reality. The whole point is the lumpiness.
- Missing the landmines. If your outflow rows look identical for 13 straight weeks, you have not found the quarterly tax payment or the annual renewal yet. Go back to the bank statements.
- Building it alone and hiding it. Whoever signs checks and whoever signs contracts both need to see the sheet. The agency in the example lost $96,000 of timing because an account lead accepted net-60 without knowing what net-60 does to week 7. That is an information problem, not a competence problem — and it is the reason the sheet belongs in a shared folder, reviewed with the person who does your books.
- Letting it go stale the moment things improve. The forecast feels essential during a crunch and pointless during a good quarter. The good quarter is exactly when it should be telling you whether you can afford the next hire or a price change — both of which hit cash months before they hit profit.
Start this afternoon
Open a spreadsheet. Thirteen columns, Monday dates across the top. Opening cash from your bank, minus the money that is not yours. Your AR aging pasted in with the haircuts applied. Payroll dates. Fixed bills. Then one hour on 12 months of bank statements hunting for the over-$1,000 items you forgot.
Set the floor at four weeks of fixed costs. Read the headroom row. Find the lowest week.
Most people who do this for the first time find something in weeks 5 through 9 that they did not know was there. That discovery, six weeks early, is the entire return on the afternoon.
This is a general operating framework, not accounting, tax, or legal advice. Terms, tax dates, and obligations vary by jurisdiction and by business — check yours with your accountant before you act on a number.