A 13-week cash flow forecast is a weekly grid of money in and money out, built from committed obligations rather than from your profit and loss statement. It runs thirteen weeks because that is one quarter, which is far enough ahead to see an inventory purchase coming and close enough that your assumptions are still worth something. For an inventory business the forecast has one job: tell you which week you run out of money, before you get there.
Profit will not tell you this. A seller can post a strong quarter on the P&L and miss payroll in the same period, because inventory leaves the bank account months before the revenue it produces arrives.
Step 1: Build the skeleton, not the model
Open a spreadsheet. Thirteen columns, one per week, dated by week ending. Five row groups: opening cash, cash in, cash out, net movement, closing cash. Closing cash in week one becomes opening cash in week two. That is the whole structure, and it should take fifteen minutes.
Resist adding detail here. The most common failure in cash forecasting is a beautiful 400-row model that nobody updates after week three. Aim for something you will actually maintain every Monday morning.
Step 2: Fill cash in from payout schedules, not sales
Sales are not cash. Marketplace payouts are cash, and they arrive on a lag that varies by channel.
Pull your last eight weeks of actual deposits by channel and calculate the real gap between the sale and the deposit hitting your bank. Do not use the marketplace’s stated policy; use your own bank statement. Reserves, holds and settlement period boundaries mean your realized lag is usually longer than the documented one.
Then forecast forward using net payout, not gross sales. If you sold $100,000 on eBay last month and received $84,000, your forecast row is the $84,000 pattern. eBay’s published fee structure for most categories, listed on its selling fees help page in August 2026, is 13.6% of the total sale amount up to $7,500 per item plus a $0.40 per order fee, and eBay states that the total sale amount includes shipping the buyer paid and sales tax. Your net is always lower than your instinct.
Step 3: Load cash out in three separate blocks
Split outflows into fixed, variable and lumpy. The split matters because each behaves differently under stress.
Fixed: rent, payroll, software, insurance, loan payments. These are knowable to the dollar and the week. Enter them as actual dates. Payroll on the 15th and the last business day is not “two payrolls a month” when the last business day lands on a Friday in a five-week month.
Variable: marketplace fees, advertising, shipping, merchant processing. Forecast these as a percentage of the sales that produce them, but book them in the week they are deducted, not the week the sale happened.
Lumpy: inventory purchases, freight and duty, quarterly tax payments, annual software renewals. This block is why the forecast exists. Every purchase order you have signed belongs here with its actual payment terms, and so does the deposit you have not paid yet on the container arriving in week nine.
Step 4: Model inventory on payment dates, not receipt dates
This is the step sellers get wrong, and it is worth being pedantic about.
Suppose you place a $120,000 purchase order with a supplier on 30% deposit and 70% on shipment. The deposit is $36,000 and it leaves in week two. Production takes six weeks. The balance of $84,000 leaves in week eight when the goods ship. Ocean freight is another five weeks, so the inventory lands in week thirteen, and you begin selling it in week fourteen.
Your P&L shows nothing at all until those units sell, because inventory sits on the balance sheet. Your bank account is down $120,000 across weeks two and eight. The forecast is the only document in your business that shows this correctly.
Step 5: Run the worked example
Take a seller doing $2.4 million a year, roughly $200,000 a month across Amazon and Shopify, with a 24% net margin on paper.
Opening cash in week one: $95,000.
Weekly cash in from payouts averages $41,000, so $533,000 across thirteen weeks. Fixed costs run $18,000 a week, or $234,000. Variable costs are already netted out of the payout figure. Lumpy costs are the two inventory payments above, $36,000 in week two and $84,000 in week eight, plus a $19,000 estimated tax payment in week seven.
Total out: $234,000 plus $120,000 plus $19,000, which is $373,000. Total in: $533,000. Net positive $160,000 across the quarter, ending at $255,000. On a quarterly view this business looks fine.
Now look at week eight. Opening cash that week is roughly $85,000 after the week seven tax payment. Cash in is $41,000. Cash out is $18,000 fixed plus the $84,000 supplier balance, so $102,000. Closing cash: $24,000.
Twenty-four thousand dollars against an $18,000 weekly fixed cost is a business with one week of runway, in a quarter that ends up $160,000 ahead. One late payout, one chargeback batch, one delayed marketplace deposit and week nine is a phone call to a supplier asking for terms you did not negotiate in advance.
Step 6: Decide what you do about week eight
A forecast that only shows you the cliff is half a tool. The remedies are limited and all of them are easier eight weeks out than two days out.
Renegotiate the supplier split from 30/70 to 50/50 so more of the cost lands in week two when you are flush. Pull the tax payment forward or back a week where the schedule permits. Slow advertising for two weeks in weeks six and seven, accepting lower week nine revenue in exchange for surviving week eight. Draw on a line of credit you arranged before you needed it, which is the only time anyone will give you one.
None of these are clever. They are all available, and they are only available with notice.
Step 7: Reforecast every week, and keep the misses
Every Monday, replace week one’s forecast with actuals, roll the grid forward one column and add a new week thirteen. Keep a column recording what you predicted versus what happened.
After six weeks that variance column is the most valuable part of the model. It tells you your payout lag is nine days rather than seven, that your fixed costs are $19,400 rather than $18,000, and that you consistently forget freight. A forecast built on your own measured error beats a forecast built on your own optimism.
Where the numbers come from
The inputs live in three places: your bank feed for actual timing, your marketplace settlement reports for gross-to-net, and your accounting file for committed obligations. If those three do not reconcile, the forecast inherits the discrepancy.
Sellers running multiple marketplaces usually need the settlement data posted into the accounting file automatically before this is workable by hand, which is the problem ConnectBooks writes about at length in its own treatment of why profit and cash diverge.
Two references worth having open while you build this. IRS Publication 538 covers accounting periods and methods, including the point that a taxpayer carrying inventories generally uses an accrual method for purchases and sales, which is precisely why your P&L and your bank balance disagree. And the Small Business Administration’s guidance on managing your business finances is a plain-language check on which records you are obligated to keep before you start building anything on top of them.
Build the grid this week. Thirteen columns and about an hour, and at the end of it you can name the specific week your bank balance gets uncomfortably close to zero, which is the one thing your profit and loss statement will never tell you.
