For two and a half years, I managed cash at OutcomeCatalyst with a monthly cash flow statement. In between months, I would look at my bank account, and if the number was low I would start spinning through everything that could go wrong.
We're almost entirely bootstrapped, and managing cash this way nearly ended the company. A little over a month ago, I finally looked closely at my daily cash balance. I had about two weeks of runway left, with more people on payroll than ever and my first baby due in two weeks. I almost went bankrupt, and it was the most stressful stretch I've had running this company.
The problem was timing. Buyers who said yes took months to sign, and clients didn't all pay on the due date. A monthly cash view hides all of that.
I studied engineering and I've always been good at math, but I don't have a background in finance. I didn't even know the difference between cash and accrual accounting when starting the company.
Then I remembered something a client told me. He runs a turnaround firm and has led successful turnarounds of companies you'd recognize, all of them far bigger than mine.
When I was first selling to him I asked what the turnarounds come down to, and he said I'd laugh at how simple it is. He just interviews the executive team and builds a 13-week daily cash flow statement. With those two things he can figure out how to turn around any company.
Everything I needed to build my own 13-week cash flow statement already existed in three places: my QuickBooks, my bank account, and my credit cards.
So I gave all three to Claude and had it build a 13-week statement with a daily position. It was easier than I expected, and it needed no engineers and no data pipelines. It shows me my balance, my projected low point, and the first day I'd go negative.
It immediately helped me identify which vendor payments to negotiate and which financing to use to close the short-term gaps, and I avoided running out of cash. Now my bank balance could read one dollar and I wouldn't care, as long as the forecast shows me surviving.
Here is the framework to build one for your own business, so you can see a cash shortfall months before it happens. Plus, I'll show you where to look for cash shortfalls in your specific industry.
The framework
A 13-week daily cash position takes four steps:
Pull: Get the last 90 days of your books, your bank activity, and your credit card statements into one place, so all revenue and expenses are accounted for.
Project: Have Claude analyze your cash flow and project it forward, day by day, for the next 13 weeks, so you can see the lowest your cash gets and the day it happens.
Test: Before you commit to a new expense or new terms, add it to the forecast and see what the scenario does to your low point.
Decide: When the forecast shows a day you go negative, act while you still have options.
Let's take a look at exactly how I did each step below:
The walkthrough
Step 1: Pull
I only needed three sources. My books live in QuickBooks, which connects to Claude directly. My bank and credit cards don't have a connector, so I just manually export the statements as files.
Your systems will be different, but you need the same three records: the books, the bank activity, and the cards. If your accounting system doesn't connect, an export works the same way. Check your company's data policy before you upload bank files to any AI tool.
Give Claude the last 90 days for every account that cash moves through. That should be enough history for it to find what repeats monthly or more often, including payroll, vendor payments, subscriptions, card autopay, and how long each client takes to pay.
Ninety days does have a blind spot, however. Anything that happens once a year or once a quarter may not appear in that window at all.
Make sure you account for those and give them to Claude with the files: insurance premiums, quarterly tax payments, annual software renewals, bonuses, and any large purchase you already know is coming.
After the first build, upkeep is light. Every few days I export the most recent bank and credit card activity and add it, which takes me about 20 to 30 minutes.
Step 2: Project
I never built a spreadsheet model for this. Claude reads my transaction history and projects it forward day by day for 13 weeks using a Claude artifact, which is a small interactive page Claude builds inside the chat.
At the top are four numbers: my balance today, my projected low point, the first day I go negative, and where I end week 13. Below that is a chart that visualizes my daily balance.
Two details decide whether the forecast is right. The first is timing on receivables. An invoice should appear on the forecast on the date that client usually pays, which can be later than the due date.
The second is credit cards. If you load both your card statements and your bank statements, every card purchase can be counted twice, once on the card and again inside the payment that leaves your bank. Count only the payments from the bank as cash going out.
Start every review with one check. Before you look at anything else, compare the forecast's balance for today with your bank's. If the two numbers don't match, something was missed or counted twice, and common causes are a card payment or a transfer between your own accounts. Do this on every refresh.
The instruction I gave Claude to build the cash flow statement looks like this:
“Using my QuickBooks data, bank statements, and credit card statements from the last 90 days, plus my list of annual and quarterly payments, build a 13-week cash flow with a daily cash position. Find every recurring inflow and outflow and project it forward. Put each receivable on the date that client has historically paid, not the invoice due date. Count only the card payment leaving my bank account as cash going out, and use the card statements only to estimate its size. Ignore transfers between my own accounts. Show me my balance today, the projected low point and its date, the first day the balance goes negative, and the week-13 balance. Then give me a chart and every line item with its source or the assumption behind it.”
Step 3: Test
If you have a finance team, a 13-week forecast may already exist in a spreadsheet somewhere. The difference with this one is that you can question it yourself using Claude, in plain English, and get the answer right away without waiting for someone to rerun a model.
Once the forecast exists, test any scenario by asking it common questions, such as:
The new commitment: “If I add this vendor or this hire starting on this date, what happens to my low point?”
The late payer: “If my largest customer pays 30 days later than usual, is there a day I go negative?”
The delay: “If I push this payment out a month, how much runway does it buy me?”
When I want to bring on a new vendor, I go to the forecast before I sign anything. I tell it what I want to do, what it costs, and when each payment goes out, then I ask it to update and tell me when I go negative.
If no negative day appears, I move forward. If one does, I know how big the gap is and how many weeks I have to close it, before I've committed to anything.
$118,000
What I raised and borrowed to close the gap: $48,000 in short-term loans plus a $70,000 SAFE. The SAFE bridges us to the seed round I'm raising now.
When the forecast showed a gap I needed to cover, I ran my options through it: short-term loans, a draw on my $50,000 revolving line, or a raise on a SAFE. I ended up taking $48,000 in short-term loans to cover the near weeks and raising $70,000 on a SAFE for the rest.
Your options will look different, maybe a larger line or a term loan. Run each one through the same forecast and look at every week after the money arrives, including the weeks the repayments are due.
Step 4: Decide
I look at the daily cash position almost every day. If a negative day appears anywhere in the next 13 weeks, I see it within days, with months to fix it.
Seeing it early is a huge advantage. With two weeks of notice you take whatever option you can get. With two or three months you get to compare.
The first place I looked was vendor terms, which is a common place to start when turning around a company. Turnaround operators will go to a vendor and say they can't pay for 90 days.
I did a softer version. I went to vendors we had long-standing relationships with and asked for an extra 30 days. Those conversations let me keep payroll covered without needing to take a larger loan.
The forecast tells me which conversations are worth having. I can look at one contractor payment and see that moving it out 30 days buys me a full extra month of runway.
It works the same way on money coming in. The overdue invoice that would raise the low point most is the first collections call to make.
At a larger company, catching a bad week shouldn't depend on one person looking every day. At that point, it's best to automate it with live connections and alerts.
The outcome
I took $48,000 in short-term loans and raised $70,000 on a SAFE after comparing my options against the cash flow statement. Choosing the right mix instead of the first thing available is what a daily position gives you.
Vendor terms extended my runway without new money. Asking long-standing vendors for an extra 30 days kept payroll covered, and the forecast told me exactly which conversations were worth having by showing which delayed payment bought the most time.
I went from finding out I had two weeks of cash to knowing the exact day it would run out with months to act instead of days.
It costs me 20 to 30 minutes every few days and no engineering time. Every new commitment or scenario is run through it first, and I'm not stressed about cash anymore.
■ The industry audit
A daily cash position finds the week your cash gets tight and shows you what's causing it. Here's where to look in your specific industry:
If you're in commercial real estate: Put your debt service, property tax, and insurance dates against your actual rent roll for the next 13 weeks. Rent arrives monthly and those obligations don't, so the mismatch alone creates tight weeks a monthly statement hides. Then model what one major tenant going vacant does to the weeks that follow, because the income disappears while the mortgage, taxes, and insurance keep coming. That gap is where CRE operators actually go negative.
If you're in industrials: Pull your last ten large orders and count the days between your first material purchase and the customer's payment. You pay for materials and labor long before you get paid, so your biggest order can be the one that takes you negative. That count is how long you finance each order yourself. Put it in the forecast before you accept the next big PO.
If you're in healthcare: Pull the last 90 days of collections and count the days from date of service to cash, by payer. Payroll goes out on a fixed schedule, and reimbursement arrives after each payer's own delay, minus denials. Project collections on those real delays, then count the payroll weeks that land before the cash that funds them.
Where it breaks
A verbal yes is not cash. The forecast will project any revenue you tell it to expect, and I've had a buyer commit out loud and still not sign two months later. Keep unsigned deals out of the forecast entirely, and add them the day the contract is signed, not the day you feel good about the call.
The forecast is only as current as your last export. Between updates the position is a few days stale, so a payment that clears early or a client who pays late won't show until you refresh. And anything that fires once a year or once a quarter won't appear in a 90-day window at all, so list your annual and quarterly payments.
I keep that forecast current by hand, which works at my size. At $100 million in revenue there are too many invoices, vendors, and accounts for anyone to export manually. At that point, you'd want to automate it with data pipelines and alerts, so every invoice and payment updates the forecast on its own.
Zach
Founder & CEO, OutcomeCatalyst
Book 30 minutes with me →
