A friend has spent the better part of a year trying to talk me into opening a padel club. The pitch is genuinely good: the sport is everywhere, the courts near us are booked solid by seven in the morning, and — his clinching argument — “everyone’s making a fortune.” I believe roughly half of that. The half I believe is that padel is popular. The half I do not is that popularity and profitability are the same thing, which happens to be the precise confusion that empties bank accounts.
So rather than argue over a beer, I did what I always end up doing: I opened FinModeler and built the thing in numbers. What follows is that session, more or less as it happened, because it is a fair illustration of the gap between liking an idea and knowing whether it pays. I have rounded the figures and kept them illustrative; the point is the process, not my friend’s business plan.
The decision is narrower than the idea
The first thing the exercise does is shrink the question. “Should I open a padel club?” is unanswerable, because it smuggles in a dozen hidden choices. The answerable question is much smaller and far more useful: does this club — four indoor courts, at this rent, built for this much, filled to this level — return the capital it eats, and how badly does it suffer when one of those numbers moves against me? Everything in the model is an attempt to make those last few words precise.
The assumptions are the actual work
FinModeler starts by letting me describe the idea in plain language. I type something close to “four-court indoor padel club with memberships, court rentals and coaching”, and it reads that and pre-selects a sensible business model for me to confirm or override. Pleasant, but not the point. The point arrives at the assumptions.
I work in detailed mode, because a padel club is mostly a building and a utilisation rate, and I want to see both. The general assumptions come first — tax rate, inflation, the risk-free rate and risk premium that together set my discount rate, the interest rate on the debt I will need, and the equity I am prepared to put in. Then the specifics: the build cost for the courts and fit-out, the monthly rent, the staff, and the revenue drivers that decide everything — court price per hour, how full the courts actually are, the membership base, the coaching cut, the bar.
Typing these out is quietly clarifying, because each one is a claim about the world rather than a fact. “65% utilisation” is not a number I know; it is a bet I am making, dressed up as a number. “Everyone’s making a fortune”, I should note, is not an assumption you can enter into a model at all, which is the first honest thing the exercise does to the pitch.
The model takes about a minute
With the assumptions in, FinModeler generates the full model and returns a downloadable Excel file inside the app in roughly a minute. The workbook is the real thing — income statement, balance sheet, free cash flow, a feasibility study with NPV, IRR, payback and break-even, the base, optimistic and pessimistic scenarios, a Monte Carlo simulation, and the charts — all built on native formulas, the way a careful analyst would have done it by hand. The spreadsheet is a genuinely useful output. It is not, however, the reason I am here. I am here for the verdict.
Reading the verdict
I do not need to open Excel to read it, because the same statements are there in the app. The feasibility study is where I look first. The base case comes out positive — a respectable IRR, a payback of around four and a half years, a break-even occupancy of roughly 58%. That last figure is the one that matters, and it is the one a glossier pitch would never have surfaced: the whole venture lives or dies on whether four courts stay 58% full, every month, for years. Suddenly my friend’s “booked solid by seven” looks less like evidence and more like the only hour of the day that works.
The AI reads it back to me
FinModeler’s Analyse with AI produces a written reading of the model — its strengths, its main risks, the trends in margin and cash, and a few decision-oriented recommendations. The thing I value is that it is generated from my own assumptions and outputs, not from market averages, so it reflects the model I actually built, inconsistencies and all. In this case it does exactly what a good analyst would: it names utilisation as the fragile driver, points out that the pessimistic scenario turns cash-negative in the second year, and suggests I pressure-test the rent and the build cost before anything else. It is reading my numbers, not improvising new ones, which is the only version of AI I am willing to let near a cash flow.
Testing the future I would rather not think about
A verdict on the base case is only half an answer, so I change the one assumption I trust least — utilisation from 65% down to a plausible, unglamorous 52% — and regenerate. FinModeler keeps the original and saves this as a new version, which means I can hold the two side by side rather than overwriting the future I preferred. The swing is sobering: the same club, with one realistic adjustment, slides from a sound investment to a slow bleed. For the full picture — what happens when utilisation, price and rent all wander at once rather than one at a time — the Monte Carlo simulation in the workbook does the heavy lifting, and there is a separate piece on why a single base case hides exactly that kind of risk.
Comparing the two versions is the part that changes how I think rather than what I see. The version history is a record of how my judgement moved and why, which is worth as much as the numbers themselves when, six months from now, I have forgotten why I ever believed in 65%.
What I knew at the end that I did not at the start
The session does not end in a yes or a no, and it should not. It ends in a condition: the club works if utilisation clears about 58% within the first eighteen months, the rent holds below a level I can now state exactly, and the build comes in on budget — and below that line it quietly loses money for years. I walked in with a feeling and walked out with three things to verify before I sign anything: real local utilisation data, a firm rent, and a builder’s quote rather than a guess.
There is a particular relief in finding the soft spot of an idea on a screen, in an afternoon, rather than in month fourteen with a lease and a personal guarantee. It is cheaper to be wrong here. That, more than any chart, is what the exercise is for.
Create your first model for free — start with your own idea on FinModeler.
FAQs
Do I need a finance background to do this?
No. FinModeler guides you through the business model and assumptions step by step, and generates the financial statements for you. The judgement you bring is about your business, not about accounting mechanics.
How long does it actually take?
The model itself is generated in about a minute once your assumptions are in. The honest answer is that the thinking — deciding what your assumptions really are — is the part worth spending time on, and the part that pays you back.
Is the Excel file the deliverable?
The workbook is a useful output you can keep, share or hand to an accountant, with all the formulas intact. But the deliverable is the decision: knowing which assumptions your idea depends on, and what to validate before committing capital.