Who we serve › SaaS & subscriptions
Payment infrastructure for SaaS and subscription businesses: involuntary churn is a stack problem
TL;DR. A subscription business loses customers two ways: they decide to leave, or their renewal fails and nobody notices. The second is a payments problem with payments fixes: credentials flagged correctly so issuers approve merchant-initiated renewals, account updater coverage so reissued cards do not fail, a retry policy that follows the decline codes and stays inside the schemes' limits, and an acquirer whose issuer relationships approve your renewals. Terms that fit recurring revenue (lower reserve, faster settlement, pre-dispute tools) come on top. We diagnose the renewal funnel the way we diagnose any approval problem: by issuer, by code, by market.
Who this is for. SaaS, membership and subscription merchants processing $100K+ a month in recurring card revenue who want to know what share of churn is declines, and what it would take to recover it.
Four places a failed renewal could have succeeded.
Credential flags
The first charge flagged as cardholder-initiated with credential on file; every renewal flagged as a merchant-initiated transaction referencing it. Issuers approve correctly flagged MITs they would decline as fresh transactions, and in Europe and the UK flagged MITs are out of SCA scope. Wrong flags are declines you caused.
Account updater
Reissued cards updated before the renewal runs, through the acquirer's enrolment in Visa Account Updater and Mastercard Automatic Billing Updater. Coverage varies by acquirer, issuer and region. Ask for the enrolment share and the update rate; raise both.
Retry policy
By decline code: never on codes the issuer says it will never approve, timed on codes that say wait, capped inside the schemes' limits (Visa: 15 in 30 days on retryable codes; Mastercard: by Merchant Advice Code). Beyond that, every attempt is fined and the merchant is flagged.
The acquirer's issuers
Approval on renewals differs by acquirer because issuer relationships differ. The cheapest acquirer per transaction is rarely the one that approves most renewals. Compare on recovered revenue, not rate. Run the numbers →
Four patterns we see in subscription stacks.
01
Churn reported as one number.
Voluntary and involuntary churn in the same metric, so the payments share is invisible and nobody owns it. Split failed renewals out by decline code and issuer and the recoverable share appears.
02
Retrying on a calendar.
Every failed renewal retried on days 1, 3, 7 and 14 regardless of the decline. Attempts on do-not-retry codes are fined; attempts beyond the cap are fined; attempts on wait codes are made too early. The schedule costs money and recovers less than a code-driven policy.
03
Renewals challenged in Europe.
Merchant-initiated transactions sent without the MIT flags, so issuers treat them as new transactions and demand authentication the customer is not there to give. A gateway configuration problem reported as regulatory friction.
04
Terms built for a one-off merchant.
Reserve and settlement timing set as if revenue were unpredictable, no pre-dispute tools, chargebacks on unrecognised renewals handled after the fact. Predictable revenue is an argument for better terms; make it, because the template will not.
Diagnose. Improve. Prove.
Diagnose
Failed renewals by decline code, issuer, card type and market. Credential flagging as actually sent. Updater enrolment and update rate. Retry schedule against scheme limits. Authentication rate on renewals in Europe and the UK. Dispute reasons. Current terms: reserve, settlement, pre-dispute tools.
Improve
Fix the flags at the gateway. Raise updater coverage. Rebuild the retry policy by code. Add the acquirer whose issuers approve your renewals, with terms that fit recurring revenue. Add local recurring methods where the card decline rate by market says so.
Prove
Move a bounded share of renewals, by market or card type. Measure recovered revenue, approval by issuer, retry cost, dispute ratio and settlement timing against the baseline. Expand as the data earns it.
The subscription review file.
- Six months of renewal attempts with outcome, decline code, issuer, card type and market
- How the first transaction and renewals are flagged at the gateway (CIT / MIT, credential on file)
- Account updater enrolment share and update rate from the acquirer
- Current retry schedule and the cost of attempts
- Authentication rate on renewals in Europe and the UK, by exemption
- Chargebacks by reason, and the pre-dispute tools in place
- Current acquirer agreement: reserve, settlement timing, pricing, change rights
What is involuntary churn, and how big is it?
Customers who did not decide to leave: the renewal failed, the card was declined, nobody noticed until the account lapsed. How big it is depends on your card mix, your ticket size and your stack, which is why we do not quote a benchmark for you. The published numbers say it is material: Stripe attributes about a quarter of lapsed subscriptions to payment failures alone, Recurly reports recovery rates above 45% on the most common decline messages, and a third or more of cards are reissued every year. The recovery lives in three places: updater coverage, retry policy, and the acquirer's issuer relationships.
What are stored-credential and merchant-initiated transaction rules?
Since 2017 Visa and Mastercard require merchants to flag transactions on stored cards: the first transaction where the cardholder is present and agrees to storage (a cardholder-initiated transaction with a credential-on-file indicator), and subsequent merchant-initiated transactions (MITs: recurring, instalment, unscheduled) referencing the original. Issuers use the flags to approve renewals they would otherwise decline, and under PSD2 correctly flagged MITs are out of scope for strong customer authentication. A stack that does not flag correctly is paying for declines it caused.
What does account updater do?
Visa Account Updater and Mastercard Automatic Billing Updater let the acquirer refresh stored card numbers and expiry dates when the issuer reissues a card, before the renewal runs. Coverage is not universal: it depends on the acquirer's enrolment, the issuer's participation and the card's region, and with a third or more of cards reissued every year the gap is not small. Ask your acquirer what share of your stored cards is enrolled and what share of updates it returned last month; raising both is one of the cheapest fixes in recurring.
How should we retry declined renewals?
By decline code, not by calendar. Visa groups decline codes into categories: category 1 (issuer will never approve, for example a closed account or a stolen card) must not be retried and each retry is fined; the other categories allow up to 15 retries in 30 days, after which each attempt is fined. Mastercard returns a Merchant Advice Code with the decline: 03 and 21 mean do not retry, 24 to 30 say how long to wait, and excessive attempts are fined. A retry policy that ignores the codes burns money on attempts that can never succeed and gets the merchant flagged for excessive authorisations.
Does strong customer authentication apply to renewals?
In Europe and the UK, SCA applies when the customer sets up the subscription (the cardholder-initiated first transaction) and does not apply to subsequent merchant-initiated renewals, provided they are flagged as MITs referencing the original authenticated transaction. Fixed-amount recurring series also have their own exemption. Renewals that are being challenged are usually a flagging problem at the gateway, not a regulatory requirement.
Which acquiring terms matter for a subscription business?
Three that mainstream templates get wrong. Settlement timing and reserve: recurring revenue is predictable, which is an argument you can make for lower reserves and faster settlement, and templates rarely make it for you. Chargeback handling: subscription disputes cluster around unrecognised renewals and cancellation friction, and the acquirer's pre-dispute tools (alerts, rapid resolution) are worth more here than almost anywhere. And decline economics: an acquirer whose issuer relationships approve your renewals is worth more than one that is cheaper per transaction and declines more. The true-cost calculator makes that trade-off explicit.
Should we add local payment methods for subscriptions?
Where a market's customers do not use cards for recurring: direct debit in parts of Europe, open banking variable recurring payments where they exist, wallets where the subscription can live inside them. Each adds a billing flow to maintain and a different dispute regime. Add where the card decline rate by market says so, not where a catalogue does.
How is Milestone paid for this?
By the acquirer or PSP you end up using, as a commission on volume. No fee to you for the review. We place with whichever of ~30 providers fits your footprint and card mix, so we have no reason to push one.
Subscriptions primer
Soft vs hard declines, retries and updater, in short.
E-commerce
What a merchant at $100K+ a month should get from its acquirer.
True-cost calculator
Fee rate vs approval rate on your volume.
Send us six months of failed renewals.
Decline codes, issuers and outcomes. We tell you within one business day what share was recoverable, where it was lost, and what the stack would need to recover it. No fee for the review.