There is a specific kind of confidence that comes from finishing a financial model. The tabs are color coded. The formulas cascade. You look at cell B14 and it says you have eleven months of runway, and for a few days you feel like someone who runs a company.
Then you hire someone. A customer pays 45 days late. You raise prices for new signups but not existing ones. Nobody updates the file, because updating the file is not urgent and everything else is. By month three that runway is fiction.
CB Insights’ analysis of startup post-mortems has consistently put running out of cash at the top of the list of reasons startups fail. Almost none of those founders lacked a spreadsheet. They had spreadsheets. The spreadsheets were just describing a company that no longer existed.
Here are the four files nearly every founder builds, and the specific way each one rots.
Spreadsheet one: the P&L you built to prove you are profitable
This one usually gets built after a good month, because a good month makes you curious about whether the business works.
It breaks in three predictable places. First, it uses blended margin instead of contribution margin, so your best customers subsidize your worst ones and you never find out which is which. Second, it mixes cash and accrual without deciding which one it is, so revenue lands when the invoice goes out but costs land when the card gets charged, and the profitable months are just months where a payment happened to clear early. Third, and this is the one founders defend hardest, it leaves out your own salary. A business that is profitable only because the founder works for nothing is not profitable. It is subsidized, and you are the subsidy.
What to fix: pick cash or accrual and label it in the tab name so future you cannot pretend otherwise. Add a contribution margin line per customer segment. Put a market rate salary for yourself in the cost column, even if you are not taking it yet. If the model still works, you have a real business. If it does not, you have found the actual problem, which is worth more than the comfort.
Spreadsheet two: the runway tab
Runway is the number founders quote most and understand least, because it gets stored as a single cell when it is actually a range.
The standard version divides cash by last month’s burn. That assumes burn is flat, and burn is never flat. It rises the month you hire, the month annual contracts renew, the month you finally pay the accountant. It also confuses gross burn with net burn, which means a founder with lumpy revenue can be off by months in either direction. And it measures months to zero, which is the wrong finish line. You do not get to operate until zero. You stop having options somewhere around three months out, because that is when a fundraise or a bridge stops being a negotiation and starts being a plea.
What to fix: model three burn scenarios instead of one, including the version where you hire the two people you already know you need. Separate gross burn from net burn on different lines. Then subtract three months from the bottom and call that your decision date. That is the real number. Put it somewhere you see it weekly.
Spreadsheet three: the growth projection
This is the most fun file to build and the least honest one, because it is where hope gets multiplied.
The classic failure is compounding off a peak. You had a 22 percent month, you type 22 percent into the growth cell, and 24 months later the model says you are at 40 million in ARR. The second failure is treating churn as one percentage applied to the whole customer base. Real churn is front loaded. New customers leave at a much higher rate than customers who made it past month six, so a single blended number will always understate near term losses and overstate long term retention. The third failure is bundling new revenue, expansion revenue, and churned revenue into one net number, which hides the case where you are losing customers as fast as you win them and the top line looks calm.
Buffer has published its revenue and metrics openly for years, and the thing you notice reading it is how unglamorous the real shape of growth is. Long flat stretches, occasional steps, the occasional bad quarter that a projection would never have drawn.
What to fix: model by cohort, not by aggregate. Break the revenue line into new, expansion, and churned so you can see all three moving. Then build the pessimistic case using your worst month rather than your best, and check whether you would still be alive in it.
Spreadsheet four: the marketing spend tracker
This one usually starts as a channel by channel list of spend and leads, which feels like measurement but is closer to bookkeeping.
Spend and leads tell you almost nothing about whether marketing works, because they say nothing about payback. A channel with a high cost per lead can be your best channel if the customers it brings stay for three years. A cheap channel can quietly bankrupt you if those customers churn in month two. The other problem is attribution theater, where every channel claims the same customer and the sum of your reported conversions exceeds your actual conversions.
What to fix: track CAC payback period per channel rather than cost per lead. Use contribution margin, not revenue, when you calculate it, so you are measuring against money you actually keep. Accept that attribution will be imperfect and pick one model you use consistently, because a slightly wrong number applied the same way every month is far more useful than a perfect number you recalculate differently each time.
The pattern underneath all four
None of these break because of bad math. They break because a spreadsheet is a photograph, and you are running something that moves. The file is accurate on the day you build it and decays every day after, and the person responsible for updating it is the person with the least time in the company.
So here is the honest test. Open your model right now and find a cell nobody has touched in 60 days. Whatever that cell says, you have been making decisions on it. That is not a discipline problem. It is a design problem, and the fix is a model that pulls from what is actually happening instead of waiting for you to remember it.
Inpaceline OS has the four of these built in, which is the point: a P&L calculator, burn rate and runway, a SaaS revenue and churn model, and marketing contribution, all working off the same numbers instead of four files that disagree with each other. If you would rather spend that hour on the business than on the spreadsheet, start at inpaceline.com.
