Recurring Billing Done Right: The Complete Setup Guide for Subscriptions and Retainers
Recurring Billing Done Right: The Complete Setup Guide for Subscriptions and Retainers
Recurring revenue is the best business model ever invented and the most operationally unforgiving: the same charge, executed perfectly, hundreds of times, against cards that expire, accounts that empty, and customers who forgot they subscribed. Done casually, it leaks several percent of revenue a month to failed payments and converts confused customers into chargebacks. Done right, it's a machine: tokenized credentials, clean consent, payday-aware retries, and a cancellation flow that protects you legally because it's easy. Here's the complete build.
What you'll learn
The recurring stack: four components
Every durable recurring system has the same skeleton. (1) Stored credentials, tokenized. Card-on-file means storing a network token — a substitute credential — never the raw card number: your PCI scope collapses (the architecture logic from the PCI guide), and network tokens survive card reissues, quietly preventing a whole category of failures. On the bank side, the equivalent is a properly authorized ACH mandate. (2) Consent capture — the legal layer, next section, stored alongside the credential because the two are one artifact when a dispute arrives. (3) A scheduling engine flagging transactions correctly as recurring to the networks (mis-flagged recurring charges authorize worse and dispute worse). (4) Failure handling — the dunning ladder, where the money actually is. Most modern processors bundle all four; the real differentiation, as we found in the involuntary churn report, is almost entirely in the fourth.
Consent and compliance: the rules that keep subscriptions legal
Subscription billing sits under three stacked rulebooks, all pointing the same direction. Card-network rules for stored-credential and recurring transactions: clear disclosure of amount and frequency at enrollment, explicit notice when a trial converts to paid, and accessible cancellation — violations feed straight into the dispute categories that fill the chargeback playbook. State auto-renewal laws, now in a substantial and growing list of states: conspicuous consent (no pre-checked boxes), renewal reminders for longer terms, and — the sharpest edge — cancellation parity: if they subscribed online, they must be able to cancel online. Federal negative-option law: the FTC's sweeping "click-to-cancel" rule was struck down in court days before taking effect — but the decision was procedural, the underlying statute it enforced remains fully live, and enforcement against subscription traps has continued under it. The strategic read: don't build to the minimum of whichever rulebook is weakest this quarter. Build to the convergence point — disclose everything before consent, remind before renewal, cancel in one step — and every current and future version of these rules is already satisfied.
Cards, ACH, or both: routing by value
The two rails fail differently and cost differently, which makes the routing decision mechanical. Cards: instant authorization, universal, frictionless signup — and 2–3% per cycle plus an expiration-and-reissue failure mode that never stops. ACH: fees in cents (the structural economics from the ACH guide), and bank accounts don't expire — the single largest involuntary-churn source simply doesn't exist on the rail — at the cost of slower confirmation and returns that arrive after the fact. The routing rule that falls out: consumer-scale subscriptions default to cards; high-value recurring migrates to ACH. A $15/month app eats the card fee happily; a $2,000/month retainer paying $58 in card fees versus under a dollar in ACH is leaving $680+/year on the table per client, twelve times. The elegant middle: offer cards by default, incentivize ACH above a value threshold (a small discount pays for itself immediately), and let each customer's rail match their economics.
The dunning ladder: rescuing failed payments
Unmanaged, involuntary failures — expired cards, insufficient funds, reissued numbers, false fraud declines — typically claim several percent of recurring revenue every month, silently outpacing voluntary churn at many businesses. The recovery ladder, in order of leverage:
- Prevention: account updater + network tokens. Automatic credential refresh when banks reissue cards eliminates the largest failure category before it happens — the highest-ROI line item in recurring infrastructure.
- Pre-billing signals. A reminder before high-value renewals reduces both NSF failures and "what is this charge" disputes in one email.
- Intelligent retries. Failed debits recover dramatically better when re-attempted on paydays (1st, 15th, Fridays) than immediately — machine-learning retry timing, the province of AI in the processing layer, outperforms fixed schedules meaningfully. Cap attempts; endless retries breed disputes and, on ACH, return-rate trouble.
- Human-tone dunning. "Your payment didn't go through — update in one click" recovers more than legal-sounding demands. Grace periods with maintained access outperform instant cutoffs for winback.
- Graceful downgrade before cancellation. A pause or cheaper tier offered at the failure point retains customers the ladder couldn't collect from.
Well-built ladders recover the majority of involuntary failures — commonly the single largest revenue-retention lever a subscription business hasn't pulled.
Cancellation as strategy
The counterintuitive rule: easy cancellation is a profit strategy. Trapped customers don't stay — they dispute: a chargeback costs the revenue plus a fee plus ratio damage, versus a clean cancel that costs only the revenue and preserves the relationship for winback. Cancellation friction is also precisely what every rulebook above targets, making it a compliance liability that compounds. The right architecture: one-step online cancellation, an exit survey after (not blocking) the cancel, a save-offer presented once, and immediate confirmation. Businesses consistently find that removing cancellation friction cuts disputes measurably while barely moving voluntary churn — the customers who leave through an easy door were already gone; the door's ease just determined whether they left through support or through their bank.
The metrics that tell you it's working
- Involuntary churn rate — failed-payment cancellations as a share of subscribers; the number the ladder exists to crush.
- Recovery rate — share of failed payments ultimately collected; the ladder's report card.
- First-attempt authorization rate — the upstream health signal; tokens, correct recurring flags, and clean data move it.
- Dispute rate on recurring charges — the consent-and-cancellation architecture's report card; rising disputes mean confusion or friction somewhere upstream.
- Effective processing rate by rail — total fees over volume, cards vs. ACH; the routing rule's savings, made visible monthly.
Recurring revenue on rails built for it
HL Hunt Pay runs subscriptions and retainers on both rails — tokenized card-on-file with account updater, ACH mandates with validation built in — plus AI-timed retries, dunning tooling, and correct recurring flagging, all in one settlement view.
Frequently asked questions
Tokenized stored credentials (or ACH mandates), a consent flow disclosing amount, frequency, and cancellation, correct recurring transaction flagging, and failure handling — retries, account updater, dunning. Modern processors bundle all four; failure handling is where they differ.
Card-network requirements (disclosure, trial-conversion notice, accessible cancellation), state auto-renewal laws (conspicuous consent, cancellation parity), and federal negative-option law — a major FTC rule was vacated pre-effect, but the underlying statutes remain live. Build to the convergence: disclose, remind, one-step cancel.
Account updater and network tokens (prevention), payday-aware intelligent retries, ACH for high-value recurring, and human-tone dunning with one-click update. Well-run ladders recover most involuntary failures — often the biggest unpulled retention lever.
Both, routed by value: cards for consumer-scale subscriptions (instant, universal), ACH above a value threshold (cents instead of percent, no expiration churn). A small ACH incentive on high-value plans pays for itself immediately.
Key takeaways
- The stack is four parts — tokens, consent, scheduling, failure handling — and the money is in the fourth.
- Build compliance to the convergence point: disclose before consent, remind before renewal, cancel in one step.
- Route by value: cards for consumer scale, ACH for retainers — cents beat percent and bank accounts don't expire.
- The dunning ladder (updater → reminders → smart retries → human dunning → downgrade) recovers most involuntary churn.
- Easy cancellation is profit strategy: trapped customers dispute; freed ones cancel cleanly and come back.
Stop losing revenue to expired cards
Sign up for HL Hunt Pay and put your subscriptions on infrastructure that prevents failures before they happen and recovers the ones that slip through.
This guide is educational and does not constitute legal advice. Network rules, state auto-renewal laws, and federal enforcement postures change over time; confirm current requirements for your jurisdictions.