Why subset testing creates risk
A common approach is to test on a subset of the portfolio: test 10,000 cards, find the issues, fix them, then migrate a million. The problem is that dormant and older accounts often contain the anomalies that cause load failures. They are not represented in a clean subset sample. Those issues only appear when the full portfolio is loaded, which is the worst possible moment to discover them.
Full-portfolio testing is the minimum standard for a migration that will not produce surprises on live day.
Where the switch itself can go wrong
Even well-prepared migrations can encounter problems in the final moments. The switch, when live traffic shifts to the new processor, has to be timed precisely within narrow network windows set by the card schemes. A minor overrun extends downtime. Incorrect chip settings cause point-of-sale rejects. Missed field mappings result in wrong card statuses. Any of these, at scale, means cardholders facing declines.
Booking a contingency network window alongside the primary migration slot is a basic risk management step. Without one, if a dress rehearsal uncovers a minor anomaly, the project can stall for months while the next available slot is found.
What the failures have in common
Across migration failures, the recurring theme is that assumptions were made where verification was needed: assumptions about what data existed, about what had been tested, and about how four parties, the issuer, the existing processor, the card networks, and the new processor, would coordinate across different interests and timelines.
The antidote is a structured, gated process that surfaces problems early, tests at full scale, and leaves nothing to chance on the day that counts.
Thredd's migration methodology is built around removing those assumptions at every stage. That means defining data formats before extraction begins, testing the full portfolio rather than a subset, running full-scale dress rehearsals, and booking contingency network slots as standard. The approach has been developed across decades of migrations and tens of millions of cards. The goal throughout is the same: a cutover that cardholders never notice.
The approach is built around removing those assumptions at every stage. That means defining data formats before extraction begins, testing the full portfolio rather than a subset, running full-scale dress rehearsals, and booking contingency network slots as standard. The approach has been developed across decades of migrations and tens of millions of cards. The goal throughout is the same: a cutover that cardholders never notice.
Want the complete framework? Download the playbook for invisible transitions
