ANZ Banking Group is exploring ways to securely host snapshots of de-identified transactional and time series data that can be used for trial runs of new software or algorithms.
The thinking represents a potential expansion of the way the bank wants to use software by Sydney-headquartered Data Republic, which it struck a partnership with earlier this year and also made a strategic investment in.
Data Republic is mostly known for its secure data exchange platform and data matching technology, both of which fall under the Senate brand.
The most obvious use case for Data Republic at ANZ is in understanding how to securely share consumer data with third parties, as banks will be required to do when the open banking regime comes into effect from July 2019.
Indeed, that has been the main use case for Data Republic discussed by ANZ to date, but the group has now revealed it is not the only use case under consideration.
“A discussion we’re having more broadly about innovation is that often when we want to test pieces of software, we need to provide some form of data for the software to operate on,” head of data innovation and partnerships Jeffrey Mentiplay said.
“There’s a whole class of AI startups, risk evaluation models and even apps where you think, ‘If only I could give them a snapshot of transaction and time series data, they could simulate [some early results and value].
“We are thinking about Data Republic in that context because it’s a cloud platform with a sandbox that no one can get into, I’m happy with their security schema, and I’ve got a pipeline and a governance process to put de-identified datasets in there.
“So we’re thinking about Data Republic not just as a sharing and analytics service, but a safe place where data can be and compute can run.”
Hyper Anna wants to play
Mentiplay suggested that using Data Republic to store chunks of real data for test purposes could already be in-train, as ANZ evaluates the AI capabilities of the Australian startup Hyper Anna.
“There’s a partnership with a company called Hyper Anna which is an AI and exploratory tool. We’re like, ‘we can just dump a chunk of our institutional dataset into a sandbox and they can install Hyper Anna’ and we’re comfortable with that because it’s safe.”
If it is put into full production, the capability could allow ANZ to very quickly work out if what was being offered to them had any potential.
“We think this radically reduces the innovation window to understand the real world impact [of a new tool or technology],” Mentiplay said.
Open banking exploration
ANZ revealed back in May that Data Republic would be used to kickstart the group’s infrastructure preparations for open banking, though it was noted at the time that it was unlikely to be the only tool required to comply.
Mentiplay indicated that thinking is still current.
“We’re likely to have an independent infrastructure for open banking, but what’s interesting is we’re highly likely to use Data Republic to get ready for that," Mentiplay said.
"Because we can create pools of data between us and maybe a smaller bank, have a shared analytics environment and actually [simulate] what we are we going to see when suddenly this data can flow in an [new product or service] application process.
“Data Republic is part of our organisation’s response to [these new data sharing requirements],” Mentiplay continued.
ANZ sees a few key roles for Data Republic in its open banking preparations.
First, it is drastically cutting the time it takes to initiate and implement a data sharing arrangement; until now, the timeframe for creating a one-off shared data asset at ANZ was between six and 12 months.
“Whilst there might have been one or two examples where it was less, that timeframe was very common and directly led to us missing opportunities to competitors or in the market,” Mentiplay said.
“Missed opportunities were a critical experience of ours in data sharing.”
One of ANZ’s problems was lack of a consistent process around data sharing. Put simply, everything was a one-off.
“As an enterprise we didn’t really have a cohesive mindset around how to get approval to share data,” Mentiplay said.
“Every single time we did it was a point solution - [involving] a different lawyer, a different risk person, a different governance person and ultimately a different process - and even when we got agreement that we wanted to just experiment or understand value."
Mentiplay added that this "almost created an entirely new journey to understand how we were going to do it safely, protect customer information if it’s involved, and make sure no one’s going to run away with the data and do things that we didn’t expect.”
Because of that, two things happened.
First, “damningly, the response to all of those concerns was all too often [to run the data sharing off] a USB stick, because at the end of the day at least we knew what was on it, we could encrypt it and hand it to someone,” Mentiplay said.
The other internal response was simply to run data sharing under the radar.
“We went from unknown data sharing, with lots of people everywhere going ‘I want to do this but I won’t tell anyone because it’ll effectively enter the valley of death’,” he said.
Data Republic has brought structure, security and governance to ANZ’s data sharing. “It’s a very consistent, logged and audited experience for a business person to govern and decide and determine that if there is a data share, what will be done with it,” Mentiplay said.
“We’ve used Data Republic as a technical and process platform to push back through our organisation an enterprise approach to data sharing.”
On the clock
On ANZ’s second data share implemented through Data Republic, the bank got the total timeframe down to five weeks.
“We basically agreed in about four weeks [internally to do the share] and then one week later we had analysts in the data,” Mentiplay said.
“It was just a total transformation and that was still while we were learning.”
Mentiplay said Data Republic was treated internally by ANZ as a “lower friction place” to perform secure data sharing.
When governance and risk communities found out a particular project would use Data Republic, there was an attached level of trust that reduced the need to spend huge amounts of time planning and asking reams of questions.
Trust on tap
That trust has also translated into a steady stream of requests internally for new data sharing and matching arrangements.
“That has drawn out of the business like water this demand to understand the value of data by sharing data or by having sandboxed environments where external experts can access data,” Mentiplay said.
“So our list of data sharing projects is going from a known list of 20 to a prospect list of 80, and then as word gets out my small team effectively gets flooded as the business owner of this capability.”
Recognising value faster
Once open banking arrives and consumers have greater control over their banking data - and who it can be shared with - ANZ is expecting to be flooded with data sharing requests.
These requests will mean the bank is sharing data that it previously would have kept tightly under wraps and used to try to create its own competitive advantage.
“As an organisation, we’re going to have a bunch of data assets that previously we felt were proprietary and gave us competitive advantage moving out into the open world,” Mentiplay said.
“In that context, the data sharing is going to massively increase.”
Bare and share
Banks will have to be able to tell consumers about their data, how it is used, where it goes (or went) and why, and what consents had been collected around the use of that data.
But they can still find ways to derive value from aggregated forms of that data - they will simply have to move more quickly to do it.
“I think it’s about having a carefully thought-through structure to say, ‘I can understand the power of a piece of data when it gets to me, when I see it I can interpret it, but I have that capability though very carefully managed data sharing,” Mentiplay said of the balance to be struck under open banking.
“Data Republic allows us to create environments that mean we are ready to act in milliseconds when we see a pattern in data because we have tested it and we know that we have to take a particular action.”