How far is your organisation from DBS?
DBS spent thirteen years building nine capabilities before its AI returned about S$1 billion a year. Nine questions show which of them you have, which DBS era you resemble, and which gap to close first.
The questions follow the order DBS built things in, from taking engineering back in-house in 2009 to measuring AI value against control groups in 2022. The order matters: each capability was the base for the next. Answer for your organisation as it runs today, not as the strategy says it will.
- Engineering ownership
- Infrastructure
- Experiment culture
- Customer journeys
- APIs and reuse
- Operating model
- Data governance
- AI platform
- Value measurement
No sign-up. Your answers stay in your browser.
Question 1 of 9
01 · Engineering ownership
Who builds and changes your core systems?
02 · Infrastructure
How is your computing infrastructure run?
03 · Experiment culture
How often do your teams run controlled experiments, such as A/B or test-and-control?
04 · Customer journeys
How is work organised around customer outcomes?
05 · APIs and reuse
How reusable are the functions inside your systems?
06 · Operating model
Who owns a technology-enabled service in your organisation?
07 · Data governance
Can teams find and use trustworthy data under clear rules?
08 · AI platform
How do models get built and put into production?
09 · Value measurement
How do you measure the value of your AI?
–/ 27
Capability by capability
DBS: About 85% outsourced when Gledhill arrived in 2008; 85% insourced by end-2017 and 90% by 2019.
DBS: Data centre revamp from 2009; a cloud-based data centre in 2017; 99% of applications on the virtual private cloud by 2019.
DBS: In 2015 running an experiment was in every KPI and the bank ran about 1,000.
DBS: About 250 senior managers each sponsored a journey from 2016; more than 60 cross-functional journeys covered over 60% of revenue by 2024.
DBS: 155 APIs at launch in November 2017, over 350 a year later.
DBS: 33 platforms from 2018, each led two-in-a-box by a business and a technology lead on shared KPIs, funded as a platform.
DBS: PURE data-use principles in 2018 (Purposeful, Unsurprising, Respectful, Explainable); the ADA data platform live by 2019.
DBS: ALAN, a single AI/ML platform with 100% of models checked against its governance framework, fully deployed in 2021; time to value cut from 18 months to 2 to 3.
DBS: S$180m in 2022, S$370m in 2023, S$750m in 2024 and about S$1 billion in 2025, measured by test and control and reported as economic value.
One check the score leaves out. DBS's pace of change outran its resilience: outages in 2021 and 2023 brought an extra capital requirement of about S$1.6 billion and a six-month freeze on non-essential IT changes from the Monetary Authority of Singapore. Put board oversight of technology risk in place before the pace picks up: the technology risk board pack has the twenty questions.
Read how DBS built each capability.
The case study sets out all nine in order, with the platform operating model, the outages and the regulator's response, and what the S$1 billion measures, from 87 numbered sources.
This diagnostic is a self-assessment against a sequence drawn from DBS Bank's public filings, press releases and published analyses. It does not constitute an audit, a benchmark study or advice, and DBS Bank has no connection with it. The era comparison is illustrative: DBS's path depended on its size, market and regulator. Your answers stay in your browser and are neither sent to Terence Kok nor reviewed by anyone.
The eras the score compares you with.
Nine capabilities in order, scored onto five DBS eras
Each answer scores 0 to 3. The top option is where DBS got to.
Total, 0 to 27
Capabilities, points and band edges from this diagnostic's scoring. The era comparison is illustrative: DBS's path depended on its size, market and regulator. Sources for every step are in the DBS case study.
| Score | Result | DBS at the same point |
|---|---|---|
| 0–8 | Before the basics | DBS before 2009 |
| 9–15 | Fixing the basics | DBS, 2009 to 2014 |
| 16–21 | Digital to the core | DBS, 2014 to 2017 |
| 22–25 | Platform organisation | DBS, 2018 to 2021 |
| 26–27 | Measured AI value | DBS, 2022 onward |
DBS published its first AI value figure in 2022, thirteen years after it began. An organisation that starts with AI use cases and skips the ownership, data and measurement work in the rows above should not expect the same curve. The DBS case study has the sources for every row.
Boards keep putting the DBS figure in front of me and asking why their own AI has not paid off the same way. The fair answer is usually that they are comparing their year one with DBS's year thirteen. I built this so a leadership team can see in four minutes which of the nine things DBS built they already have, and which one to fix before the next AI budget is approved. Be honest with the answers; nobody sees them but you.
