What's the honest churn-risk estimate for a subscription-program migration, and what benchmark should we hold the migration agency to?

Where does the 2 to 5% churn estimate come from, and what does it measure?

The 2 to 5% churn estimate comes from Deploi's subscription migration record, which cites July 2026 research on community-reported migrations (Deploi Build vs. Buy, October 2026). It measures the share of active subscribers that merchants report losing in subscription platform migrations, which the same research puts at 60 to 90 days each. The record gives no sample size or method, so read the range as a set of reports, not a rate. It is not a Deploi measurement, and it is not a forecast for your base.

The record also says where the loss sits: it concentrates where subscribers have to re-enter a card. So the honest estimate for your own program starts with one count: how many active subscribers would have to re-enter a card. That count is part of the cost-benefit case for leaving Recharge or Skio, and it is the number this answer turns into a benchmark.

Which loss channels make up a migration's subscriber loss?

Four loss channels make up a migration's subscriber loss, and payment credentials are the largest because a subscriber whose card doesn't arrive simply stops billing.

The first and third channels sit in the systems, which means the agency can be held to them. The second sits in your own messaging, so the note to long-tenured, high-value subscribers belongs in your plan, not the agency's.

What decides whether a subscriber's card transfers?

Where the card is vaulted decides whether it transfers. Shopify's token migration takes customer payment information from four processors, on every plan: Stripe, Braintree, PayPal Express and Authorize.net (Shopify Help Center, October 2026). Connected as a secondary gateway, the old processor keeps billing existing contracts, because "Your customers' payment methods are held by that processor, outside of Shopify". So a card held by one of those processors can keep billing without the subscriber re-entering it, and what happens to billing cycles and payment tokens has its own answer.

Token migration takes credit cards from Stripe, credit cards and Apple Pay from Braintree, customer profiles from Authorize.net, and PayPal billing agreements through PayPal Express (Shopify developer docs, October 2026). Card-number migration needs Shopify Plus or Enterprise and takes credit cards only (Shopify Help Center, October 2026). It can't move PayPal billing agreements, and "Accelerated wallets such as Venmo, Apple Pay, and Google Pay, and bank account information aren't supported". The re-collection cohort is every subscriber whose payment method neither route moves. A swap between third-party apps on the same store is the easy case: uninstalling a third-party subscription app keeps contracts and customer payment information (Shopify Help Center, October 2026). Count the cohort in week one, not week six, because in our judgment a share of those subscribers won't re-enter a card at all.

What benchmark should the statement of work hold the agency to?

The statement of work should hold the agency to a payment-method transfer rate, not to a churn cap. Churn is downstream of things the agency doesn't control, including your own email. Write these three in instead (Deploi's scoping practice, not an industry standard):

• Payment-method transfer rate, stated as a target share of active contracts, measured seven days after cutover.
• Billing-accuracy rate across the first full cycle: charges that fired on the right date, at the right amount, once.
• Contract-field parity: cadence, next-charge date, discount and line items matching source to destination, sampled and signed off.

Fix the denominator first, because platforms count subscriptions differently. Skio says Recharge treats every product as its own subscription, while one Skio subscription can hold several products (Skio Help Center, October 2026). Then measure against the destination's own pre-migration count; Skio's preview lists any subscriptions it can't migrate, and why. Those three numbers are measurable and attributable, so the agency can be held to them. The questions that test an agency's track record come before any of this.

Why can't anyone quote a churn number in advance?

Nobody can quote a credible churn number in advance, because the dominant variable is your gateway and vault, and the second is your own messaging. An agency that offers a guaranteed churn figure before auditing your payment stack is selling confidence, not diligence.

Vendor promises measure something else as well. Skio's App Store listing promises to move subscribers "without losing data" (Skio listing, October 2026). That promise covers records, not billing: Skio's own guide notes that "A valid payment method is not required to create a subscription contract in Shopify" (Skio Help Center, October 2026). Such a contract goes into Skio's Payment Recovery at its next billing attempt, so a clean import can still mean a failed first charge.