SaaS cash flow is boring only when the plumbing is right. A modern stack sells seats and usage across dozens of currencies; mixes self-serve card payments with enterprise invoices; shares revenue with resellers and cloud marketplaces; and recognizes revenue against contracts that morph mid-term. If pricing, metering, tax, and collections don’t agree at the data level, you end up with beautiful dashboards and a general ledger you don’t quite trust. The finance win is a single commercial truth: one catalogue, one meter, one contract object, and a receivables engine that speaks both cards and bank rails without losing exchange-rate context.
Start at the catalogue, not at the invoice. Prices by edition, add-ons, and tiers need native currency points in the markets that matter and derived points elsewhere with sensible rounded endings. Quote-to-cash tools should carry the same IDs your billing system uses so a “Business-EU” annual plan in the CRM is the same thing the ledger recognizes when money lands. The tricky part is usage: you’re not charging for minutes or gigabytes so much as for outcomes—API calls, records written, seats active, compute seconds. Those events arrive from product telemetry; they require time stamps, customer IDs, and an immutable meter you can replay. Treat the meter like a sub-ledger: if you change a rate plan, you apply it to a frozen window of events; if you grant credits, you see the credit consume future events deterministically instead of hoping the system “does the right thing.”
Currency policy is a business decision with accounting consequences. Present native prices to lift conversion; lock the rate at authorization for card self-serve and at invoice issue for AR customers—then reuse that rate for credits and adjustments inside a short, published window. Store the rate source and timestamp on every monetary event. Without that metadata, you can’t explain why a March true-up in yen doesn’t quite match the February estimate or why an annual prepay in pounds creates a translation swing at quarter end. Treasury’s job is not heroics; it’s small buffers in high-velocity currencies, short-dated forwards on net exposures for your biggest non-functional currencies, and natural hedging where you can spend locally (cloud hosting regions, local payroll, marketing).
Collections should match go-to-market motion. Self-serve and SMB gift you low-touch revenue if you respect the basics: network tokens survive reissues, account updater fights card lifecycle events, and dunning follows local time zones with a couple of soft retries before you interrupt service. Enterprise customers expect invoices, purchase orders, and payment runs tied to their calendar. Give them rails they like: SEPA (or SEPA Instant) in the euro area, Faster Payments in the UK, ACH equivalents in domestic markets, and wires for the larger annual prepaids. Bank-to-bank methods cut fees and refund latency; they also deliver structured remittance so cash application is automatic if you’ve assigned virtual accounts per customer or per cohort. Measure cost per successful collection—not just processor fees but FX and the minutes your team spends chasing references.
Usage billing is where credibility is made or lost. Meter late and you under- or over-charge; meter noisily and disputes spike. A crisp pattern helps: a daily or hourly metering pass that aggregates raw events, a rating pass that applies plan rules and credits, a preview so customers can see accruing charges mid-cycle, then a statement that reads like a narrative: “You committed to X, you used Y, we applied Z credits, here’s your overage.” Prepaid commitments with draw-down are kinder to cash and to relationships than pure pay-as-you-go; minimums protect your cost base; grace on small overages reduces support volume more than it reduces revenue. Proration deserves grown-up math: seats added on the 19th of a 30-day month shouldn’t distort revenue or AR—use the same clock everywhere and document it once so auditors don’t learn your policy from support macros.
Tax is not an afterthought in SaaS anymore. Electronic services rules mean VAT/GST often follows customer location even in B2B; the U.S. continues to expand sales-tax reach for software and digital services, and more countries now require e-invoicing or real-time reporting. Encode place-of-supply logic in your billing system, not in a tax spreadsheet. Keep evidence of customer location (billing country, payment instrument country, IP or business registry) and apply reverse-charge where it belongs; treat marketplace and reseller transactions according to the deemed-supplier frameworks that apply. When e-invoicing is mandatory, don’t ship until the invoice is accepted by the local platform—or you will create DSO creep you can’t claw back with stern emails.
Partners complicate cash in two directions: upstream cloud marketplaces and downstream resellers. Marketplaces pay you on their settlement calendars and take a cut; you must mirror their entitlements and chargebacks so your ledger lines match their remittance. Resellers expect wholesale price lists, co-terming on renewals, and consolidated statements; some insist you invoice the end customer via them while paying them a margin or rebate later. Model partner-of-record and co-sell attribution at the contract object so rebate and MDF liabilities are computed rather than invented. Never net partner incentives invisibly into revenue—post them as explicit contra-revenue or expense lines with policy tags and dates, or you’ll teach your team to hide economics where no one can govern them.
Refunds and credits should read like product rules, not favors. Seat downgrades mid-term create credits; service interruptions create service credits; billing errors create credits; contract termination under convenience clauses creates refunds aligned to the notice period. Return value to the original payment method; reserve alternate destinations for edge cases with approvals. Friendly fraud in SaaS is mostly “I didn’t authorize this auto-renewal.” Clear descriptors (brand + product), pre-renewal reminders aligned to local norms, and evidence packs with login logs, 2FA events, and prior invoices will win more disputes than any lyrical prose in a chargeback portal.
Data separation is what lets all of this survive audit season. Money lives as integer minor units plus ISO currency; every external call uses idempotency keys; every monetary event carries rate, source, and timestamp. The customer object has legal entity, tax IDs, billing contacts, payment methods, and—if you sell across subsidiaries—an explicit delivering entity for intercompany. AR automation ingests ISO 20022 statements; virtual accounts map inbound funds to the right customer without a human guessing which “Acme Ltd” paid today. Reconciliation is a classification game, not a treasure hunt: FX drift, processor fee mismatch, partial pay, missing remittance, duplicate prevention. Publish those buckets with owners and SLAs and they will shrink.
What to watch changes once the plumbing is honest. DSO by segment is useful, but involuntary churn rate as a percent of MRR tells you more about dunning and rail choice than any anecdote. Net dollar retention hides sloppy upgrades if credits pile up; track credit issuance as a share of billings and the reasons behind it. Cost per successful collection, time-to-refund, and payout-failure rates to partners predict your support volume far better than NPS. FX cost in basis points of billings and of partner payouts distinguishes treasury skill from luck; covered vs. uncovered exposure removes bravado from the conversation. Auto-match rate and manual minutes per thousand payments make the month-end close a calendar appointment rather than a rite of passage.
There’s a place for a payments intermediary in SaaS even if you already have a gateway. When you need multi-currency accounts, virtual IBANs per customer or partner cohort, and local bank rails to bring down DSO in your key regions—without stitching a dozen bilateral bank connections—using a specialist can compress timelines and give finance line-level fee/FX transparency the general ledger can trust. Choose on corridor breadth, uptime, failure handling, and reporting clarity, not on the single lowest headline fee