Data quality: the AI barrier Indonesian banks named most
Article
What stops AI spreading further inside Indonesian banks and lenders? Data quality, say 27% of the people we surveyed inside them. It led among respondents at an early stage (31%) and was still in the top three further along (25%). This article covers why data quality leads, how it pushes up the cost of every AI project and what changes when you fix data once and share it.
Only got five minutes? Here are the key takeaways of this article:
27% of Indonesian respondents named data quality as the biggest challenge stopping wider use of AI. Budget came second (21%) and skills and people third (18%).
Only 6% named old systems. Integration challenges and old systems together account for 18% of answers.
A global survey of chief data officers also put data first, so on data the view from inside Indonesian institutions matches the global one.
When each use case cleans its own copy of the data, the same problems get fixed again and again, which pushes up the budget every AI project needs.
Get the full findings in From AI adoption to AI at scale. You can also book a call with vanda.jirasek@oradian.com to talk it through with our Jakarta team.
Want to find out more first? Read the rest of this.
Data quality leads the list
Asked to name the single biggest challenge stopping wider use of AI, 27% of Indonesian respondents chose data quality. Budget came second (21%), followed by skills and people (18%), integration challenges (12%) and regulations (12%). Old systems were named by 6% and not being sure where to start by 3%.
Meanwhile, data quality led among respondents at an early stage (31%). Further along, skills and people, data quality and budget were each named by a quarter (25%).
A process that runs every day needs data that’s right the first time, with no manual clean-up. That’s why data quality turns into a technology question once AI runs in daily processes: models need complete, current operational data, defined the same way across systems, with a record of where it came from.
Why data quality pushes up the budget
When each use case extracts and cleans its own copy of the data, the same quality problems get fixed again and again, with each fix costing money and specialist time. Budget is the second barrier our respondents named. As AI moves from tests into several processes, it needs money for data work, integration, licences and people.
Global research agrees on data
Informatica’s CDO Insights 2025 surveyed 600 chief data officers and chief data and analytics officers across the US, Europe and Asia. They named data (43%) and technology (43%) as the top obstacles to moving generative AI from pilot to production, followed by people (35%) and process (35%).
Our Indonesian respondents also put data first (27%), while integration challenges and old systems together account for only 18% of their answers and budget comes second (21%). Informatica asked data leaders to name their obstacles while we asked the people working with AI inside Indonesian institutions for their single biggest challenge, so the comparison is directional. The same Informatica survey found that 97% of those using or planning to use generative AI face difficulty demonstrating the business value of their initiatives, which makes budget harder to win.
Apply these findings to your own firm
Give data an owner: someone in the business who is accountable for the data AI depends on.
Fix data at the source: fix quality where data is created, so every use case downstream gets it right.
Share controlled access: give models controlled access to that data, so each new use case starts from data that’s already right.
Budget for data: put data quality work on the same plan and budget as the AI use cases that depend on it. Track what data fixes cost on each AI project, so data quality has a budget line of its own.
Oradian: Database Access x Data = AI excellence
Database Access gives your bank a governed, read-only copy of its production database, continuously synced and held on infrastructure you control. Finance, risk, data and AI teams query complete, current data without touching the live system, so every use case starts from the same place.
All product, ledger and customer data sits in one place, so reporting and decisions work from one source of truth. 750+ native APIs and real-time events through Kafka and HTTP webhooks connect that data to the systems where decisions are made. Oradian also builds real-time reporting dashboards with an AI agent your teams can prompt.
Oradian’s own team in Jakarta implements the platform and works directly with yours.
Book a call
Data quality came first for our Indonesian respondents and it shapes the cost of everything after it. Fix it once and share it so every new AI use case gets cheaper than the last.
Read the full findings in From AI adoption to AI at scale and book a call by emailing vanda.jirasek@oradian.com to talk through data with our Jakarta team before the end of the year.