We use technically necessary cookies to make this website work. With your consent, we also use statistics and marketing services to improve our website, measure the success of our advertising and show you relevant ads on other platforms. Data may be transferred to countries outside the EU, such as the USA. You can change or withdraw your choice at any time via “Cookie settings” in the footer.
Banks that want to cut service costs usually look at cost per contact, FTE and outsourcing ratios. Yet the level of service costs is decided elsewhere: in confusing letters, buggy app releases, stalled ID checks or campaigns that promise more than the product delivers. Drawing on studies and case examples, this article shows why contact volume is the biggest lever, why automation alone makes errors cheaper rather than rarer – and how banks systematically reduce avoidable contacts (failure demand) at the source instead of simply cutting staff.
At a glance
– Service costs = contact volume × cost per contact + knock-on costs. Almost every efficiency programme turns the middle dial – the one with the least room to move.
– The UK tax authority HMRC cut frontline staff by 9%.1 Waiting times went from around 5 minutes (2018-19) to almost 23 minutes (April 2023 to February 2024), and failure demand climbed to 76% in 2024/25.12
– The Postbank and Deutsche Bank IT migration shows where costs really arise: complaints about banks to BaFin rose by almost 87% to 27,536 in 2023, and media reports put the extra cost of the crisis at around €80 million.34
– AI makes mistakes cheaper, not rarer. One insurer saw 26% more calls after investing in its app and website, and self-service fully resolves only 14% of issues.56
– Defection has gone quiet: depending on the country, 25 to 51% of all product purchases went to banks other than the customer’s main bank.7 A bad process costs you the next product.
– In a model-bank calculation, the root-cause lever nets €6.4m a year. Cutting staff by 10% brings in €1.6m – and only if waiting times don’t go up.8
Banks don’t steer by EBITDA, they steer by cost-income ratio – on average 52.3% in Europe (Q3 2025).[11] Customer service is one of the few areas where costs and revenue can improve at the same time: every avoided contact cuts administrative expense, every customer relationship that holds props up net interest and fee income.
The article maps the way there in five steps: understand, assign, fix, digitalise, steer. At its core is a polluter-pays principle. The login bug belongs to IT, the confusing letter to communications, the loan status query to the credit back office. A table matches six areas with their typical contact reasons and levers at the source. A cross-functional demand board sets priorities and follows through.
One large European bank shows how fast this can work: 15 pain points found, four fixes rolled out in six months, 10% fewer service calls a month. Around €1.3m saved a year – with better CX scores.[12]
New KPIs for the board
Written for boards, divisional heads and service leaders in banking. Reading time about 15 minutes.
The most useful part in practice is the side-by-side of classic and outcome-based steering. Service cost per customer instead of cost per contact, first-contact resolution instead of AHT, share of failure demand by cause owner instead of service level. The UK’s Financial Conduct Authority calls exactly this switch good practice under the Consumer Duty. That gives you a strong argument for the next debate about how you steer.
Questions and answers
Where do a bank's customer service costs really come from?
Mostly from outside customer service. Service costs are contact volume times cost per contact, plus knock-on costs. The volume is created in product, IT, operations, communications, sales and compliance: the confusing letter, the buggy app release, the ID check that gets stuck. According to Bain, 50 to 70% of calls in retail banks are typically bad or avoidable.9 Customer service simply picks up the bill.
Can't we just cut service costs with fewer staff or with AI?
In the short term, yes. Over time, often not. HMRC, the UK tax authority, cut frontline staff by 9%, and waiting times rose from around 5 minutes (2018-19) to almost 23 minutes by early 2024.1 AI automates what it finds: mistakes get cheaper, not rarer. Gartner expects that by 2027 half of the organisations planning to shrink their service workforce significantly through AI will abandon those plans.10
How do we get rid of avoidable contacts for good?
In five steps, with the board accountable rather than the head of service alone. You understand every contact reason and its cost, hand each driver to the department that causes it, fix root causes at source, only then digitalise genuine customer needs, and steer with new KPIs. According to Bain, banks that work systematically on root causes cut total demand by a lasting 20 to 40% within two years.9
