Agile Squad Transformation Improving Release Frequency and Customer NPS
ING reorganized into 350 agile squads, improving release frequency from 5-6/year to bi-weekly and NPS by 60 points.
ING Bank N.V., a Large Enterprise Banking company, created value through Team Structure and Accountability.
ING Bank, the Netherlands-based financial services group with approximately €1 trillion in assets, faced a structural execution problem by 2014. Like most large banks, ING operated through traditional functional silos: IT, product management, marketing, and compliance each had separate reporting lines, budgets, and priorities. Getting a new digital feature from concept to production required sequential handoffs across departments — from product management to IT architecture to development to testing to compliance review to deployment — a process that took months or years. Release cycles were 5–6 major launches per year. In an era of mounting fintech competition from challengers such as Revolut and N26, ING recognized that its product velocity was its primary competitive vulnerability. Customer satisfaction, measured by Net Promoter Score (NPS), had deteriorated to approximately negative 30 in Netherlands retail banking.
In June 2015, ING launched a wholesale reorganization of its Netherlands headquarters based on the Spotify squad-tribe model:
| Metric | Before (2014–2015) | After |
|---|---|---|
| Digital feature release frequency | 5–6 major launches/year | Bi-weekly (per squad) |
| Net Promoter Score (Digital Platform Tribe) | ~−30 | ~+30 (+60 points, within 1 year) |
| Squads formed | — | ~350 (9-person multidisciplinary) |
| Tribes organized | — | 13 |
| Traditional roles eliminated / rehired | — | ~3,500 eliminated; ~2,500 rehired into squads |
Transformation launched June 2015 at ING Netherlands headquarters; model subsequently exported to Belgium, Germany, and Luxembourg operations.
The central mechanism behind ING's release frequency improvement from 5–6 per year to bi-weekly was not primarily org chart restructuring — it was eliminating sequential handoffs by placing all required skills (product, engineering, design, data, compliance) inside a single squad with a single decision authority. In the traditional functional model, each departmental boundary was a latency point: queues, re-briefings, competing priorities, and sign-off cycles accumulated at every interface. Squads controlled their own deployment pipelines and could release independently, not because the central IT team got faster, but because the functional boundary through which latency was generated no longer existed.
The 60-point NPS improvement in one year is an unusually large single-year gain for a retail bank, attributable to a second-order effect: squad ownership compresses the feedback loop. In the functional model, a product manager specifies a feature and sees the customer response months later, mediated through multiple reporting layers. In the squad model, the product owner, engineers, and designer encounter customer feedback directly and in near-real time, creating the behavioral response cycle that produces genuine product improvement rather than compliance with requirements documents. This explains why employee engagement also improved — the work became legible in terms of customer outcomes rather than task completion.
The limits of the model are worth naming. The Spotify squad architecture is poorly suited to functions with heavy regulatory interdependencies — capital adequacy, AML, Basel compliance — that require coordinated rather than autonomous decision-making. ING validated the model for digital channel features, not for core regulatory processes. The net 1,000-headcount reduction (3,500 eliminated, 2,500 rehired) also represents a significant one-time restructuring cost not captured in the operational metrics — the velocity improvement was partly funded by an organizational transition that most institutions would execute over years, not months.
Revenue Growth Through Serial Insurance Agency Acquisitions
93 to 116 Diamond Clients in Three Years: What Vertical P&L Structure Made Possible