How accounting actually works

One equation, honestly kept

The whole system rests on a single identity: assets = liabilities + equity. In founder language: What the company has equals who funded it — lenders and suppliers on one side, owners on the other. Every transaction moves at least two numbers so the equation always balances.

Three statements, one system

One deal, three views

You invoice a client €50,000 in March, pay a €30,000 supplier in April, and collect in May. Watch the same deal land differently in each view:

March looks brilliant, April looks frightening, May is when the bank finally agrees with the P&L. None of the three views is wrong — they answer different questions. Trouble starts when founders read one and think they've read all three.

The chart of accounts is your reporting schema

The general ledger is the timestamped record of every transaction; the chart of accounts decides which questions it can answer. A 400-account template gives you noise; 12 vague accounts give you nothing.

Common mistakes

Cash vs accrual accounting

Two ways to tell the same story

Under cash basis, revenue and expenses are recorded on the date money enters or leaves the bank. It's intuitive — and it's how most founders think by default. Under accrual basis, revenue is recorded when you deliver what you promised, and expenses when you receive the benefit. Invoice and payment dates become secondary.

One contract, two growth charts

Your SaaS startup sells a €12,000 annual subscription in January, paid upfront:

Multiply this by fifty customers on different renewal dates and a cash-basis P&L becomes noise — impossible to steer by, impossible to defend in diligence.

When cash basis stops being acceptable

Common mistakes

Reading the three financial statements

The P&L: a story told top to bottom

The P&L reads as a cascade — each subtotal answers a different question on the way down:

The balance sheet: where diligence problems live

A snapshot of what you own (assets), what you owe (liabilities), and the residual owned by shareholders (equity). It's the statement founders skip — and the one where surprises hide. Receivables, deferred revenue, loans, unpaid taxes: all here.

The cash flow statement: the lie detector

It reconciles profit to actual cash movement, split into operating, investing, and financing activities. A company can post profits while operating cash flow is deeply negative — usually because receivables or inventory are swallowing the cash.

Ten-second reads

The monthly ritual

Common mistakes

Revenue recognition & deferred revenue

The five steps, in plain language

Deferred revenue: good news in a liability costume

Try it: Build a recognition schedule

The €28,000 mistake

The wrong version inflates January, guts February, and — repeated across contracts — fabricates a growth curve that isn't there. Diligence teams find this in days; it's the first thing they check.

Common mistakes

Revenue recognition by business model

One principle, six shapes

Lesson 04's rule never changes: Revenue is earned on delivery — not on invoice, not on cash. What changes is what "delivery" means. A subscription delivers a little every day; a consulting project delivers in stages; a one-time sale delivers in an instant. Match the recognition pattern to the delivery pattern and your P&L tells the truth. Mismatch them and it doesn't.

Subscriptions and annual contracts

A monthly SaaS subscription is the clean case: The customer pays €500, gets a month of access, you recognize €500. The complications are upgrades, credits, and refunds. A mid-month upgrade is prorated from the change date; a credit reduces revenue in the period you grant it; a refund reverses revenue — it never becomes an expense.

An annual contract is the same engine with the cash pulled forward. Invoice €12,000 in January, bank the cash, but recognize €1,000 a month — the other €11,000 sits in deferred revenue (lesson 04). Billings spike in January; revenue stays a flat line.

Usage-based pricing

Usage models — API calls, gigabytes, seats consumed — recognize revenue as the usage happens, not when the meter is billed. The catch is timing: Usage runs all month, but you invoice in arrears. At month-end some usage is delivered but not yet billed, so you accrue it as unbilled revenue (an asset) and true it up when the real invoice goes out.

Rental and lease income

Rental income is recognized straight-line over the lease term — even when the payment schedule is lumpy. A 12-month lease with total consideration of €24,000 earns €2,000 every month, whether the tenant prepaid six months upfront or the first month was a rent-free incentive. The cash schedule and the earning pattern are two different things.

A security deposit is not income — it's cash you're holding and will return. Book it as a liability. It becomes revenue only if and when you're contractually entitled to keep it.

Consulting and services

Services split by contract type. Time-and-materials work is recognized as billed — you delivered the hours, you earned the fee. Fixed-fee projects recognize by percentage of completion: A €90,000 project that's 40% delivered has earned €36,000, whatever you've invoiced. Work delivered but not yet invoiced is unbilled revenue, or WIP — an asset; cash collected ahead of delivery is deferred revenue — a liability. Most services months carry a little of both.

One month, three engines at once

Common mistakes

Accounts receivable

AR is an asset that decays

You hold a legal claim to the cash — but collection probability drops sharply as invoices age. The aging report buckets open invoices by how overdue they are, and it's the most actionable report in your finance stack:

DSO: collections as a single number

Dunning is a process, not an apology

Companies that automate this typically pull weeks out of their collection cycle without damaging a single relationship — the machine is politely relentless so you don't have to be.

Common mistakes

VAT & sales taxes

You're a collector, not a payer

When you add VAT to an invoice, you're charging the customer a tax on the state's behalf and holding it until your return is due. It never touches your P&L as income. A €10,000 invoice at an illustrative 22% VAT collects €12,200 — but your revenue is €10,000. The €2,200 is a liability you owe the tax office.

Inclusive vs exclusive pricing

B2B prices are quoted exclusive of VAT — "€10,000 + VAT" — because your business customer reclaims the input VAT, so the tax is invisible to them. B2C prices are quoted inclusive — the shelf price already contains it. The mistake that hurts is quoting a consumer an exclusive price, then discovering the €12,200 you banked is really €10,000 of revenue and a €2,200 bill.

Cross-border: reverse charge and OSS

Sell a service to a VAT-registered business in another EU country and you generally don't charge VAT — the customer accounts for it themselves under the reverse charge. It runs both directions: When you buy from a US or EU SaaS vendor, you self-assess the VAT, reporting it as both output and input on your return. The net cash effect is usually zero, but skip it and your return is wrong.

For B2C digital sales across the EU, the One-Stop Shop (OSS) lets you charge each customer their own country's VAT rate and file one consolidated return, instead of registering in each member state separately.

VAT vs US sales tax

They aren't the same tax under two names. VAT applies at every step of the chain, with businesses reclaiming input VAT so only the end consumer bears it. US sales tax applies once, at the final retail sale, and is triggered by nexus — a threshold of sales or physical presence in a state. Sell into the US and you may owe sales tax in a state you've never set foot in. Economic-nexus thresholds — commonly around $100,000 of sales or 200 transactions, though they vary by state — are the trigger to watch.

Try it: what's actually yours

Common mistakes

Prepaid & accrued expenses

Prepaids: paid now, expensed later

Accruals: consumed now, invoiced later

The lawyer's Q3 invoice arriving in November belongs in Q3's P&L. Book the cost when the benefit happened — as a liability — and let it reverse when the real invoice lands:

Materiality: Don't spread the €40 subscription

Common mistakes

COGS & gross margin

The line that changes everything

COGS rises roughly in step with revenue: costs incurred to deliver what customers bought. Operating expenses are the costs of running and growing the company — sales, marketing, R&D, admin — chosen largely independent of this month's volume. The distinction feeds gross margin: the percentage of each euro left to fund everything else.

What belongs in COGS, by model

Try it: honest vs flattering

Split payroll or stay blind

The single change that unlocks real margin analysis: Allocate salaries across COGS / sales & marketing / R&D / G&A. One undifferentiated "personnel" line makes gross margin meaningless and benchmarking impossible.

Common mistakes

Accounts payable

The purchase-to-pay flow

Payment terms are free financing

Net-30 means the supplier lets you hold the cash for 30 days after invoice, interest-free. Paying on receipt throws that financing away. On €50,000 of monthly supplier spend, moving from pay-on-receipt to net-30 keeps roughly a month of that spend — about €50,000 — working in your business instead of the supplier's.

Early-payment discount, or hold the cash?

Suppliers sometimes offer "2/10 net 30" — a 2% discount for paying within 10 days instead of 30. Taking it trades 20 days of financing for a 2% saving. Annualized, that works out to an effective return of about 37%, so it's almost always worth taking when you have the cash. When runway is the binding constraint, though, holding the cash can matter more than the discount.

Common mistakes

Payroll accounting

Net, gross, and total employer cost

The number in the offer letter is gross. From it, the employee's income tax and their share of social contributions are deducted to reach net — what actually lands in their bank account. On top of gross, the employer pays its own social contributions, commonly 25–35% of gross across the EU, though the exact rate varies by country. A €60,000 gross salary typically costs the company around €78,000 all-in.

Accruals — the money you already owe

Much of payroll isn't paid monthly. A 13th-month salary, accrued vacation, annual bonuses, and statutory severance funds — Italy's TFR, for example — are earned a little each month but paid, or owed, later. If you don't accrue them monthly, they land as a lump and blow a hole in a month that looked fine. Set aside one-twelfth of each every month, and the payment is just a transfer out of a liability you already built.

Contractor or employee?

Misclassifying an employee as a contractor to save on contributions is one of the most expensive shortcuts in startup finance. If you control how, when, and where someone works, tax authorities will likely treat them as an employee — and back-charge the contributions, plus penalties. Classify by the substance of the relationship, not the label on the invoice.

Payroll is not one line

A single "salaries" line makes gross margin meaningless. Split payroll the way lesson 09 splits costs: staff who build the product and serve customers into their operating buckets, salespeople into S&M, finance and ops into G&A. The same €78,000 tells a completely different story depending on which bucket it lands in.

Try it: what a hire really costs

Common mistakes

Depreciation & capitalization

CapEx vs OpEx

Operating expenditure is consumed now — salaries, rent, cloud hosting — and hits this month's P&L in full. Capital expenditure buys something that will earn for years — servers, equipment, a fitted-out office. You don't expense it at once; you capitalize it as an asset and depreciate it over its useful life. The test is simple: Will this still be delivering value in three years? If yes, it's probably CapEx.

Straight-line depreciation

The simplest method spreads cost evenly. €120,000 of servers with a four-year useful life depreciates at €30,000 a year — €2,500 a month for 48 months — after which its book value is zero. The cash left on day one; the P&L feels it slowly.

Try it: Spread the cost

The software-capitalization choice

Under IFRS, and under US GAAP, qualifying software development can be capitalized — the engineers' salaries become an intangible asset, amortized over the software's life, instead of hitting the P&L now. That lowers reported expense and lifts EBITDA today. It's legal and sometimes correct. But many startups deliberately expense all development anyway: Capitalized dev inflates today's profit and creates an asset investors will discount toward zero. Expensing everything is the more conservative, more credible choice — and it keeps burn honest.

EBITDA and the fixed asset register

EBITDA — earnings before interest, taxes, depreciation and amortization — strips out depreciation precisely because it's a non-cash allocation of past spending. It's useful for comparing operating performance, and misleading when depreciation reflects real, recurring capital needs. Every capitalized asset lives on a fixed asset register: what you bought, when, its cost, its depreciation to date, and its current book value. Assets scrapped but never removed — "ghost assets" — overstate what you own.

Common mistakes

Inventory accounting

Cash → asset → COGS

When you buy stock, cash converts into an asset — you're no poorer, just less liquid — and nothing hits the P&L. Only when the item sells does its cost release from the balance sheet into COGS, matched against the revenue from that sale. Buy €100,000 of stock and expense it on receipt, and you've understated this month's profit and overstated next month's, when the goods actually sell.

Landed cost — the price is not the cost

The purchase price is only part of what a unit costs you. Freight, customs duty, insurance, and handling all belong in inventory cost — this is landed cost. Leave them out and your gross margin looks better than it is, right up until those bills land in a later month and the truth arrives late.

The €12 of freight and duty doesn't vanish because you booked it elsewhere — it just resurfaces as a nasty surprise in a later month.

Costing methods

When units cost different amounts over time, which cost do you release at sale? FIFO — first-in, first-out — assumes the oldest stock sells first; in rising prices it reports lower COGS and higher profit. Weighted average blends every unit into one cost. LIFO — last-in, first-out — exists in US GAAP but is banned under IFRS, so treat it as a US curiosity, not a European option. Pick one, apply it consistently, and disclose it.

Write-downs, counts, and turns

Inventory is valued at the lower of cost and net realizable value. When stock ages, gets damaged, or goes out of season, it's worth less than you paid — and you must write it down, taking the loss now rather than pretending the asset is whole. Physical and cycle counts catch shrinkage — theft, breakage, miscounts — the gap between what the system says you hold and what's actually on the shelf. Inventory turns (COGS ÷ average inventory) show how fast stock moves; low turns mean cash is sitting still.

Common mistakes

The month-end close

Why closing matters more than founders think

Un-closed books are permanently provisional: Last quarter's numbers keep shifting after the board saw them, metrics drift between decks, and every analysis starts with "depending on which version…". The period lock — freezing a month so no entry can silently change it — turns bookkeeping into reporting people can rely on.

The close week, visualized

The checklist itself

Fast beats perfect

A 95%-right close delivered in five days beats a 100%-right close delivered in six weeks, every time. Estimates (accruals) are allowed — that's what they're for. Set a materiality threshold, estimate the rest, ship the pack on schedule. Speed is itself a quality signal: It means the process works.

Common mistakes

Accounting controls

Segregation of duties, pragmatically

The principle: No single person should control a transaction end to end. Whoever can create a supplier shouldn't also approve the invoice and release the payment — because someone who can do all three can pay themselves. In a five-person company you can't have five approvers, but you can always split the chain across at least two people: one raises and records, another approves and pays. That single split closes most of the risk.

Approval thresholds and individual credentials

Not every payment needs the same scrutiny. Set thresholds: anyone can approve up to €1,000, a manager to €10,000, a founder above that. And every person logs in as themselves — never a shared "finance@" bank login. Shared credentials destroy the audit trail: When everyone is the same user, no one is accountable and nothing can be traced.

Reconciliation is the master control

Reconciliation — matching your books to the bank, to supplier statements, to the payment processor — is the one control that catches what every other control misses. If the books say €480,000 and the bank says €390,000, something is wrong: an error, a duplicate, or a theft. Done monthly, reconciliation is how fraud surfaces in weeks instead of years.

Common mistakes

Burn rate & runway

Burn is a cash concept

Gross burn is total monthly cash outflow — your cost engine. Net burn is cash out minus cash in — the actual monthly decline in your bank position. Neither equals the P&L loss: Accrual profit includes revenue you haven't collected and excludes cash you've prepaid.

Normalize, or the average lies

Try it: naive vs forward-looking runway

Default alive or default dead

On current growth and burn, do you reach breakeven before cash runs out — no new funding needed? Neither answer is wrong, but you must know which you are: It determines how much risk you can take, and who has leverage when you raise.

Common mistakes

Unit economics

The unit and its contribution margin

Pick the unit that drives your business: a customer, an order, a seat. Its contribution margin is revenue minus the variable cost to serve it — using the gross-margin definition from lesson 09, not a flattering one. A customer paying €400 a month at an 80% gross margin contributes €320 a month. Everything else builds on this number, so it has to be honest.

Fully-loaded CAC

Customer acquisition cost is total sales and marketing spend divided by new customers won — and "total" means fully loaded: media and ad spend, the tools, and the salaries of the people doing the acquiring. Counting only ad spend is the most common way founders flatter CAC. With €40,000 of media, €20,000 of S&M salaries, and 30 customers won, CAC is €2,000 — not the €1,333 the ad spend alone suggests.

Honest LTV

Lifetime value is the contribution margin a customer delivers across their life — contribution per period divided by churn, never a naive multi-year extrapolation of today's revenue. At €320 monthly contribution and 5% monthly churn, LTV is €320 ÷ 0.05 = €6,400. Use observed retention, not hope; a cohort that's six months old can't tell you its five-year value.

The ratios — and reading them by segment

Two headline ratios: LTV:CAC (is a customer worth more than they cost to win?) and CAC payback (how many months of contribution to earn the CAC back?). Common heuristics put a healthy LTV:CAC around 3:1 and payback inside 12–18 months — but treat those as rules of thumb to verify against current benchmarks, not laws. And always read them by segment: A blended 3:1 can hide an SMB segment losing money and a mid-market segment that's excellent.

Try it: Load CAC, then read LTV:CAC

Common mistakes

Working capital

What working capital is

Working capital = accounts receivable + inventory − accounts payable. Receivables (lesson 06) and inventory are cash you've laid out but not recovered; payables (lesson 10) are cash you're holding via supplier terms. The difference is what running the business ties up. Positive and growing means cash is being consumed; negative means customers and suppliers are funding you.

The cash conversion cycle

The cash conversion cycle turns that into days: DSO (days to collect) + DIO (days inventory sits) − DPO (days you take to pay). It's the number of days between paying for something and getting paid for it — the days each euro is trapped. A 90-day cycle on €600,000 of monthly revenue locks up about €1.8 million.

Growth multiplies the trap

Here's the trap founders miss: A profitable, growing company can run out of cash. Double revenue and you double the receivables and inventory the business must carry — often faster than payables grow to offset them. The profit is real; it's just locked in working capital instead of the bank.

Financing the gap

When the gap is real, you can fund it — but know the true cost. Invoice factoring and receivables financing advance cash against unpaid invoices for a fee; a bank line of credit is cheaper but slower to arrange; stretching payables is free but limited before suppliers push back. The cheapest fix is usually operational: Collect faster (lesson 06), hold less stock, and use every day of supplier terms (lesson 10).

Try it: your cash conversion cycle

Common mistakes

The founder's KPI dashboard

The curated set

Founders don't need forty metrics; they need a dozen that are always defined the same way. Cash, net burn, and runway (lesson 16) tell you how long you live. MRR/ARR, growth rate, and NRR tell you how fast you're compounding. Gross margin (lesson 09), churn, and CAC payback (lesson 17) tell you whether the growth is healthy. DSO (lesson 06) and the headcount cost ratio tell you how efficiently you run. For an SME, EBITDA belongs on the list; for a pre-seed startup it rarely does.

Leading vs lagging, and definitions as governance

Lagging indicators tell you what happened — revenue, churn, margin. Leading indicators hint at what's coming — pipeline, activation, trial conversions. You need both: lagging to keep score, leading to steer. And every metric needs one written definition, owned by one person. "ARR" that means recurring-only in January and includes services by June isn't a metric — it's a story. Locked definitions are governance: They're what let a board compare this quarter to last.

Stage-appropriate dashboards

The right dashboard changes with the company. A pre-seed startup watches cash, runway, and a single growth signal — little else is stable enough to matter. A Series B company adds NRR, CAC payback, and cohort retention. A €15M SME cares about EBITDA, working capital, and margin by line. Reporting everything at every stage buries the two or three numbers that actually decide the next move.

Common mistakes

Budgeting & forecasting

Drivers, not outcomes

"Revenue +10% a month" is not a forecast — it's a wish with a spreadsheet. A driver-based model starts from things you decide or can measure: how many salespeople you hire and when they ramp, how much pipeline you create and what share of it closes, what you charge. Costs are built the same way — every planned hire carries salary, employer load (lesson 11), tools, and a laptop; hosting scales with usage; rent steps up when you move. When reality changes a driver, the whole model reprices honestly — and when the model is wrong, you can see which assumption was wrong.

The 13-week cash forecast

The rolling 13-week cash forecast is the survival tool. One quarter ahead, week by week: opening cash, expected receipts invoice by invoice (lesson 06), expected payments run by run — payroll, VAT, rent, suppliers — closing cash. Beneath it, a minimum-cash line: the level you've agreed never to cross. Its power isn't precision; it's early warning. A breach that's visible in week one gives you ten weeks to chase collections, slow payments, or start a raise. The bank balance gives you none.

The annual budget and the rolling reforecast

The annual budget is a contract with the board — set once, then left alone, so that accountability means something. The rolling reforecast is the living version: Every month, or at minimum every quarter, you re-run the next twelve months with actuals loaded and drivers updated. You measure against both — budget for accountability, reforecast for decisions. A company that only budgets once a year spends the fourth quarter steering by a map drawn in January.

The variance ritual

A budget you never compare to actuals is decoration. The ritual is monthly, an hour, right after the close (lesson 14):

Scenarios: base, bear, bull

One forecast is one opinion. Keep three: base (the plan), bear (pipeline slips, churn ticks up, the round takes longer), bull (the upside — with the hiring and hosting costs it drags along). The discipline isn't the three columns; it's the triggers agreed in advance: "if MRR growth is under 4% for two consecutive months, we are in bear — hiring freezes." Deciding the trigger before the downturn removes its hardest step: admitting you're in one.

Common mistakes

Deferred tax assets & liabilities

Two rulebooks, two profits

Book profit follows accounting standards — the accrual picture from lesson 02. Taxable profit follows the tax code, which has its own opinions about what counts and when. The differences come in two kinds. Timing differences are recognized by both rulebooks, just in different years — the tax code lets you depreciate a machine faster than your books do (lesson 12), or refuses the bonus accrual until the bonus is actually paid. Permanent differences never converge — fines aren't deductible in most systems, and some R&D incentives deduct more than you spent. Only timing differences create deferred tax, because only they reverse.

DTA — tax you'll save later

The biggest deferred tax asset in startup life is the loss carryforward. Years of losses accumulate into a shield: When profits finally arrive, the old losses offset them and the tax bill shrinks. At an illustrative 25% rate, €2M of usable losses is up to €500k of tax never paid. But accounting is strict about when that value may appear on the balance sheet: only when future taxable profits are probable. A loss-making startup with no reliable path to profit keeps the asset off the books entirely.

DTL — tax you'll owe later

The mirror image. When the tax code lets you deduct sooner than your books do — accelerated depreciation is the classic — your cash tax bill is lower today, and a deferred tax liability records that the advantage reverses. A DTL isn't a penalty or a hidden debt to panic about; it's a scheduling note: Some of today's tax saving is a loan from future years.

Expiry, ownership changes, and R&D cash

Loss carryforwards are not eternal everywhere. Depending on jurisdiction, they can expire after a set number of years, be capped at a share of each year's profit — and, critically, be restricted or forfeited after a change of ownership. A large funding round can qualify. The survival of your losses is a diligence question with a cheap answer before the round and an expensive one after. Adjacent to all this: R&D tax credits are often cash, not just deductions — in many systems they're refundable even while you're loss-making, one of the few tax lines that pays a startup before it earns a cent of profit.

Common mistakes

Equity, SAFEs & the cap table

Share capital and share premium

Shares carry a nominal value — often cents. Investors pay far more. The nominal value of the new shares goes to share capital; everything above it goes to share premium (additional paid-in capital). An investor wiring €2M for 100,000 shares of €0.20 nominal creates €20,000 of share capital and €1,980,000 of premium — and exactly zero revenue.

The round's transaction costs — legal, notary, advisory — are typically netted against the premium raised rather than shown as operating expense. Treatment varies by framework, so agree it with the accountant before the entries are booked, not after.

SAFEs and convertibles: a classification question

A SAFE is money received today for shares to be issued at a future round. Between signature and conversion it lives on the balance sheet — and where it lives is a genuine classification question: equity-like instrument or liability, depending on its terms (caps, discounts, repayment features) and your accounting framework. Reasonable people can debate that line with an accountant. What is not debatable: It is not revenue and not "other income." Nothing was sold and nothing was delivered — the recognition principle from lesson 04 has nothing to recognize.

Options are compensation

Equity given to employees is pay. An option grant has a value, and accounting frameworks expense it over the vesting period — even though no cash leaves. Startups routinely ignore this because it feels free, until the first audit restates several years of P&L at once. You don't need the valuation math in year one; you do need every grant documented, board-approved, recorded in the option ledger, and known to your accountant so the expense can be booked when your framework requires it.

One story: cap table, registry, ledger

Three documents describe your ownership: the cap table (the spreadsheet you negotiate with), the shareholder registry (the legal record), and the equity accounts in the ledger. Diligence reads all three side by side, and every divergence is a finding — an unregistered SAFE, an option pool that exists in the deck but not the ledger, a converted note nobody ever booked. After every equity event — round, SAFE, grant, conversion — reconcile the three the way you reconcile the bank (lesson 14).

Common mistakes

Fundraising-ready financials

What diligence actually examines

Analysts don't read your deck and nod; they rebuild your numbers from the ledger and compare. Four questions do most of the work. Is revenue real — recognized on delivery, per revenue recognition & deferred revenue (lesson 04), and genuinely recurring where the deck says it is? Are margins fully loaded, or is COGS flattered? Is every obligation on the balance sheet — VAT collected, payroll accruals, bonuses, loans? And do the metrics mean the same thing everywhere — the locked definitions of the founder's KPI dashboard (lesson 19)?

The financial data room

A data room isn't assembled in a panic the week the term sheet lands — it's the natural output of the month-end close (lesson 14) and accounting controls (lesson 15), maintained continuously. The checklist:

Restate before, never during

Every company has something — revenue recognized early, a missing accrual, a SAFE in the wrong account. Found and fixed before the process, it's housekeeping with a footnote. Found during the process, the identical fix reads as concealment: Momentum stalls while every number is re-checked, and the other side reprices the deal with the leverage you handed them. The accounting is the same; the timing decides whether it costs a footnote or the round.

Common mistakes

Scaling the finance function

The maturity ladder

There's no prize for building a finance department early — and a real cost to building it late. The pattern that works moves in three snapshots:

Around €1M, the founder plus an external accountant is right: bookkeeping, payroll, and filings outsourced, the founder watching cash weekly (lesson 16) and running the monthly variance conversation (lesson 20). Around €6M, the volume breaks the arrangement — the first in-house hire is a controller who owns the close (lesson 14) and the controls (lesson 15), with a fractional CFO borrowed for fundraising. Around €18M, finance is a function: controller, FP&A, AR/AP, payroll — a full-time CFO when capital strategy is a weekly job, and a voluntary audit already behind you.

Bookkeeper, controller, FP&A, CFO

The titles blur in job ads; the jobs don't. A bookkeeper records what happened. A controller makes the record trustworthy — the close, reconciliations, controls. FP&A looks forward — budget, reforecast, the board pack (lesson 20). A CFO manages capital and strategy — the raise, the banks, the model behind the model.

What to outsource — and what never

Outsource the repeatable and regulated: bookkeeping, payroll runs, tax filings, VAT returns (lesson 07). Specialists do them cheaper and with fewer errors. Never outsource the things that steer the company: cash and the 13-week forecast (lesson 20), the board narrative, metric definitions (lesson 19), and the decisions the numbers exist to inform. When the founder spends more than a day a week on finance ops — or complexity like multi-entity, inventory, or 20+ people outgrows the external accountant — that's the signal to bring the first hire in-house.

The first audit — before it's forced

Sooner or later an audit becomes mandatory — by statute as you cross size thresholds, or by a lender or acquirer who demands one. A first audit under deal pressure is slow and expensive: Every finding lands mid-negotiation. A voluntary audit a year earlier is a dress rehearsal — the same findings surface cheaply, get fixed calmly, and the second audit (the one that matters) is quiet. Companies planning a raise or exit within two years should treat it as preparation, not compliance.

From doer to reader

The founder's finance job changes shape three times. At €1M you touch everything — every invoice, every payment run. At €6M you own decisions and reviews: the variance meeting (lesson 20), the hiring plan, the raise. At €20M your job is to read the monthly pack in thirty minutes, trust it, and ask the three questions that matter. Building the system that earns that trust — the close, the controls, the definitions, the team — is the work in between, and it's what the previous twenty-three lessons were for.

Common mistakes