The FCA's final notice of 20 December 2022 imposed a 29.75 million pound penalty on TSB Bank over its April 2018 platform migration. It found the direct causes were configuration, capacity and coding issues, but that planning, testing, risk management and governance failings let excessive operational risk reach go live.
The facts in the notice
Over the weekend of 20 to 22 April 2018, TSB migrated most of its corporate systems, customer services and data to a new platform. Customers then faced failures in internet and mobile banking, telephone banking and branches. TSB did not return to business as usual until 10 December 2018.
Between 22 April 2018 and 7 April 2019 the bank received 225,492 complaints, about 4.3 percent of its customers, and paid 32,705,762 pounds in redress. The penalty reflected a 30 percent settlement discount, which would otherwise have been 42.5 million pounds.
Testing under schedule pressure
The notice says testing phases that were meant to run in sequence ran in parallel or changed scope when they fell behind, and testing concluded only the day before the decision to migrate.
One decision, taken outside the right governance forum around the end of February 2018, reduced the scope of a type of non functional testing on the digital channels. The FCA says that had this testing been done, it likely would have found a data centre configuration problem that later made online and mobile banking unavailable.
Governance gaps
The FCA also says certain matters were not sufficiently discussed with or challenged by the board, including the ambitious timetable, why the programme fell behind, and whether the replan was achievable. A change in who attested to non functional requirements was not clearly brought to the board's attention.
Some reviews were described as point in time and qualified, and those limits were not specifically challenged at key moments.
Lessons for any large release
- Define go and no go criteria before the schedule is under pressure, and require evidence for each.
- Make any reduction in test scope a recorded risk decision, not a technical footnote.
- Report limits of reviews to decision makers alongside the findings.
- Plan contingency for the weeks after launch, including customer contact capacity.
A note on risk acceptance
The failing was not that risks existed. Large changes always carry risk. It was that some risks were not identified or reported, so nobody could choose to accept or mitigate them.
Safeguard's policies and gates can hold release criteria in the pipeline and keep evidence attached to each decision, though the go or no go call remains yours.
Quick check
Did the FCA say the migration failed because of one bug?
No. It said direct causes were configuration, capacity and coding, but the underlying failings were in planning, testing, risk management and outsourcing.
How long did disruption last?
The notice says TSB did not return to a business as usual position until 10 December 2018.
Sources
Last reviewed: 19 October 2026.
