A SaaS that uses AI has a cost that grows with every customer: model calls, data, payment fees. The margin per customer is calculated as price − variable cost − payment fees, and the number of customers needed to cover fixed costs is fixed costs ÷ margin per customer. Example: at €49 excl. VAT per month, €6 of AI cost and 3% payment fees, each customer brings in €41.53; with €400 of fixed costs, you need 10 customers to break even.
Send these figures to your assistant to challenge the assumptions. Nothing is sent until you click: the button opens the assistant you choose with the figures entered here, subject to its own terms.
The formulas
| Indicator | Calculation |
|---|---|
| Monthly revenue | price × customers |
| Margin per customer | price − variable cost − price × payment fees |
| Monthly profit before your pay | margin per customer × customers − fixed costs |
| Break-even point | fixed costs ÷ margin per customer, rounded up to the next customer |
The example, step by step
- Price: €49 excl. VAT per customer per month; 60 customers: €2,940.00 in monthly revenue.
- Margin per customer: 49 − 6 − 49 × 3% = €41.53.
- Monthly profit: €41.53 × 60 − €400 = €2,091.80, before your pay, social charges and tax.
- Break-even: 400 ÷ €41.53 = 10 customers.
The example values are assumptions to replace with your own, not reference prices.
What the calculation also watches
The calculator projects annual revenue (monthly × 12). Above a threshold, it shows "to check": this is a warning about the trajectory, not your current tax situation.
- Micro-entreprise ceiling (€83,600 for services): one year above it does not take you out of the scheme; after two consecutive calendar years above it, the standard regime applies from the following 1 January.
Service-public
- VAT franchise (franchise en base): above €37,500 for services, the franchise ends on 1 January of the following year; above €41,250, from the day the threshold is exceeded.
Service-public
What the calculation does not include
Your pay, social contributions, tax, refunds and unpaid invoices, customer acquisition cost. AI usage that varies widely from one customer to another is better modelled with two scenarios: average usage and heavy users.
Written by Razvan Toma, founder of LegalRapide. How we check our figures · LinkedIn