From “build it” to adoption: Alan Carnaby on making a smart data economy real
What struck me in this conversation with Alan Carnaby is how often governments stop at the moment the ink dries.
A regulation is passed. A standard is published. A launch is announced.
And then, quietly, the real work begins: turning “framework” into an ecosystem that people actually use.
Alan sits in that messy middle. As Director of Smart Data Economy at New Zealand’s Ministry of Business, Innovation and Employment (MBIE), he is building the operating layer for consumer data rights, open banking, and e-invoicing. The work is technical. The politics are real. The adoption challenge is relentless.
But the through-line is simple: delivery is not just building the thing. Delivery is getting the thing used.
The citizen outcomes are plain: less friction, more time back
When reform programmes are described as “open banking” or “e-invoicing”, it is easy to lose the plot.
Alan keeps it anchored in everyday outcomes.
For small businesses and households, the promise is not abstract interoperability. It is:
Time saved on manual work like invoicing and reconciliation
Less back-and-forth to obtain information that already exists elsewhere
Lower friction in routine financial tasks that currently waste attention and trust
The best test, Alan argues, is whether the infrastructure becomes invisible over time.
Not invisible because it is unimportant. Invisible in the way a road network or good plumbing is invisible: you do not think about it, because it works.
That is the standard a smart data economy should be held to.
“Build it and they will come” is not a strategy
Alan is explicit about a common government failure mode.
Government can be excellent at creating policy and regulation, then treating that as the finish line.
Even delivery teams can slip into a softer version of the same assumption: build a platform, publish an API, and wait for the market to arrive.
But in ecosystems—especially ones that require trust—adoption does not happen by default.
Alan frames adoption as a legitimate part of the job for a public agency, at least in the early years:
Driving awareness
Building confidence
Creating momentum
Supporting uptake until the system has enough mass to sustain itself
The goal is not to “market forever”. It is to get the flywheel turning.
Consent is the Achilles heel of data sharing at scale
If adoption is the job, trust is the constraint.
Alan describes consent not as a compliance box, but as the design hinge of the whole regime.
New Zealand’s consumer data work includes an accreditation regime for fintechs operating in open banking. Part of the assessment is straightforward: check the processes, dispute handling, and operational controls.
But the deeper point is about systemic fragility.
If consent is done poorly, it can damage confidence in the entire regime.
One visible failure can undo years of adoption work.
That is why Alan keeps returning to two requirements that are often in tension:
Sound process that can withstand scrutiny and real-world disputes
User-friendly experience so consent is legible, amendable, and not punitive
In practice, that also pulls digital identity into the frame.
Alan notes New Zealand is still early in its digital identity journey, with a trust framework in place and credentials emerging. He points to practical building blocks like evolving the New Zealand Business Number into a digital business identity.
Identity becomes a trust spine, not a separate policy lane.
Delivery looks like engagement, not just governance
When Charlie asked what strong delivery looks like—across banks, fintechs, agencies, and vendors—Alan’s answer was not a neat framework.
It was a description of temperament.
He emphasised:
Determination and consistency (because it is hard, and it stays hard)
Strong engagement with the market (because assumptions break on contact)
The engagement point is specific.
In open banking, you will have parties with different incentives, different risk appetites, and different views of what “good” looks like. The temptation is to retreat into process.
Alan argues the opposite: lean in.
Engagement is how you:
Find the nuance policy cannot capture
Surface grey areas early
Adjust the approach when initial assumptions prove wrong
Provide the clarity industry needs to implement consistently
This is not consultation theatre.
It is practical systems work.
The operating model: treat it like a mini business unit
One of the most useful parts of the conversation was Alan’s description of what it actually takes to run these ecosystems.
He describes the need for a small, multi-skilled unit—almost like a mini business—rather than a purely shared-services approach.
The skill mix spans:
Technical architecture and standards
Regulatory enforcement and compliance
Communications, engagement, and adoption
Project and product management
The reason is speed and coherence.
A small unit can “move like a speedboat”, adjusting course as adoption realities emerge, rather than turning like a frigate.
This is not a critique of large organisations.
It is an acknowledgement that ecosystem delivery needs tight loops.
Collaboration needs a beacon, not consensus as an end
Cross-sector collaboration is where good intentions go to die.
Banks and fintechs are not aligned by default. Agencies have their own constraints. International counterparts have different models.
Alan’s practical advice is to be explicit about the North Star—a “beacon on the hill”—and to keep returning to it.
Without that anchor, collaboration drifts.
With it, collaboration can still be genuine without becoming paralysis.
He is careful here: listen, engage, and seek consensus where possible.
But do not make consensus the goal.
The goal is public value delivered through a system that is safe enough to trust.
Designing for scams and misuse without killing innovation
As volumes rise and payment initiation becomes more common, the risk profile shifts.
Alan describes this as a sweet-spot problem.
You need robust accreditation and controls, because bad actors will probe new initiatives for weaknesses. But you cannot set the bar so high that participation collapses.
His real point is about posture over time:
Set a bar you can defend at the start
Be prepared to mature the settings as volumes and risks increase
Keep adjusting as new risks emerge, including agentic AI patterns that will change how fraud and automation interact
Trust is not a one-time design phase.
It is continuous governance.
The next decade: foundations first, flex at the delivery layer
Looking ahead 5 to 10 years, Alan does not reach for predictions.
He talks about capabilities.
The capability that defines a high-performing smart data economy is not a flashy feature.
It is the ability to build foundational infrastructure—consumer data, digital identity, and real-time payments—that becomes invisible while enabling delivery teams to move faster and with more consistency.
Alan points to New Zealand’s “big three” as an emerging foundation layer:
Consumer data and open data initiatives
Digital identity (now stood up under the Department of Internal Affairs)
Real-time payments clearing (under the Reserve Bank)
They are distinct, but they “talk to each other”. If they are built well, they lift the ceiling for every agency above them.
This is what Government 3.0 looks like in the data economy: connected infrastructure that lets services improve continuously.
And it is what prepares the ground for Government 4.0: a world where agentic systems will accelerate both benefit and risk.
The hard part is not writing the standard.
The hard part is building an ecosystem people trust enough to use—and then getting it used.
That is where the work becomes real.




Comments