Numbers game: the clever Kiwi companies using data to get an edge
Authored by Kevin Jenkins, originally published by the NZ Herald on 7 November 2021
Don’t you love a good mailbox held aloft by a welded chain? They used to be common in New Zealand and are certainly just as authentically Kiwiana as the usual cliches like the Four Square man and old tourism posters.
They were also code for: ‘I can do this myself’.
Nowadays, Popular Mechanics is a website primarily featuring reviews of consumer products. But it used to be a fat monthly magazine, chocker full of visions of a bright new technology-driven future, and, more importantly, detailed plans for lots of cool ways to improve your home. It may have been American, but it underpinned a pre-Internet community of DIYers across the globe (or at least the Anglosphere), including all the way down to my hometown Dunedin. My father, who could and did make anything he required, kept a chest of the magazines in our cellar.
In this article, I’ll look at some stories of New Zealand companies and public-sector organisations doing the modern equivalent of adapting ideas from Popular Mechanics to our local needs by taking the advances made in data science in the US and elsewhere and deploying them in innovative ways.
In July I wrote about how the new field of data science, or “data analytics”, is already enhancing our lives in all sorts of ways, but I argued that it’s being done a disservice by some data experts — “the new data priesthood” — over-promising and shrouding the discipline in jargon. The organisations I look at in this article impressed me in how they are finding their own way and just getting on with using data to improve their services.
A major investment fund
Fisher Funds is one of New Zealand’s highest-profile investment funds. They aim to make investing “understandable, enjoyable and profitable for NZ investors”. They’ve always kept a sharp eye on their customers, including tracking some key data points. But that was focused on individuals and historical data — more business intelligence (backwards-looking) than data analytics (forward-looking). It wasn’t that long ago that they would ‘mystery shop’ their clients (pretend to be an ordinary customer) to test the prospects for the firm.
The chief client officer at Fisher Funds, Cath Lomax, told me they got used to that as the way things were done: “We were always looking to the past to change the present and how we did things moving forward. That’s not ideal.”
Particularly in the face of a growing number of innovative competitors, Fisher has now moved on from relying on intuition and old-school survey techniques.
“At Fisher we’re now generating and acting on new data-driven insights to give us a real-time understanding of current and future client needs. We can see what matters and when to our clients, and what makes a difference to business performance,” Lomax says.
Lomax herself took the Popular Mechanics route. She took the time to personally understand the power of data analytics, and then think about how to deploy it in her business in Aotearoa. Lomax is also acutely aware of the risk of bias, and during her time as BNZ’s Head of Diversity and Inclusion, she initiated an internal three-week sprint to understand gender balance within the company. This identified some priority areas of change.
Data analytics has changed her professional approach and how Fisher Funds does business.
“This was both qualitative and quantitative and for me,” she says.
“It opened up a whole new world of language and insight that changed my approach to solving problems and using insights in a meaningful way.”
The health sector
Precision Driven Health (PDH), a research partnership established by Orion Health, Waitematā DHB and the University of Auckland, is at the cutting edge of using data, machine learning, and algorithms to improve the way health care is provided to New Zealanders.
Chief executive Dr Kevin Ross said that research being done through the partnership has led to tools for calculating surgical risk, as well as platforms and governance processes that host and support scenario modelling, risk prediction, forecasting, and planning. These have been used throughout New Zealand’s COVID-19 response (see the New Zealand Algorithm Hub) and have much wider potential to support New Zealand’s reformed health system.
Data and data science can also be used to counter implicit biases in the health system, which have evolved as a result of historical imbalances in access, services, and data. PDH has been working with Te Whānau O Waipareira to make better use of data collected as part of providing Whānau Ora (wrap-around health and social services support for whānau). Through an exploratory research project, Waipareira and PDH were able to identify, for one part of Waitakere in Auckland, what types of services tended to lead to whānau succeeding, given patterns in age, gender, health priorities, and service use. The work also helped identify potential risks for whānau, as well as best practice.
Chief product officer Daymon Nin (Ngāti Raukawa, Ngāti Toa) said that as part of engaging with Whānau Ora, whānau develop a moemoeā (a dream or vision) of what they want to achieve. And the important principle here is that this is whānau-driven. Predictive modelling can be used to help identify what whānau could be working on, and this removes the variability that you might get from working with Navigators (brokers who help with planning and co-ordinating access to support and services) who aren’t as experienced or don’t know all the different options available.
The tool still needs to be piloted with whānau and more research done, but in the future it will mean that whānau can engage with the tool themselves and be empowered to reach their aspirations.
A digital agency meeting rising client expectations
“Client expectations around data are changing,” says Brody Nelson, co-founder of digital agency Translate Digital.
Translate Digital is often starting a new “green fields” project every few months, and this allows them to keep refining their underlying architecture, approach, and technology stack with the best emergent technologies.
“What we’ve noticed is that in the last few years client expectations around data and insights have changed so much that our standard build for almost any customer now includes a data warehouse or data lake and business intelligence tools to surface insights to clients,” Nelson says.
He added that the long-term benefits of having this architecture almost always outweigh the relatively small costs of setting it up.
Clients now expect much more than site analytics and conversion rates. They want product insights and, increasingly, predictive models — for example, the SafeSwim website, which Translate Digital built in partnership with experts now at environmental services firm Puhoi Stour, and Auckland Council. Developing SafeSwim was a huge challenge — it involved harnessing all of the collected water quality data and models from the various stakeholders, and communicating the science in such a way that the health risks of sea swimming were clear to the average Auckland swimmer.
The insights from the various data streams became the consumer-facing SafeSwim product. The transparency this created helped to start a discussion with Aucklanders that eventually led to the political mandate to address the issue of beach water quality head-on. Two summers later, about $450 million was allocated in the Long-Term Plan to build the necessary infrastructure — a great result for Auckland delivered by pooling data, generating predictive models, and clear science communication.
Another noticeable change is the skill set of technology teams and the relative importance of data science and machine learning skills. At Parkable.com, another Translate client, there is a growing data science team — largely because the business has prioritised the skills, but also because modern graduates are coming out of tertiary education armed with machine learning and data science skills.
Parkable is a SaaS (Software as a Service) platform for managing car parking infrastructure. One of its best features is that you can share your parking space when you’re not using it, which is great for businesses who are offering flexi or hybrid working and are looking to optimise. Parkable has built the sharing and occupancy data and insights into the product so that businesses can see how their assets are being used and optimise accordingly.
This is another example of the data and insights becoming the product in a user-facing business.
A third Translate client, Stickybeak.co, is taking a very clever approach to online research. Their product allows anyone to target niche audiences through social media quickly and cost-effectively. The results are anonymised and aggregated and returned to the researcher in real time. This has allowed a global public-sector health organisation to survey COVID-19 messaging in multiple markets very quickly. Through a series of nationally representative surveys it has generated insights into how our feelings about COVID have changed over the last year and a half.
Again, data is no longer analytics and conversion rates — the insights drawn from the data are the product.
A big government agency
As part of its transformation programme, ACC introduced a system that uses an algorithm to fast-track the approving of straightforward claims. This allows a rapid turnaround of the vast majority of the roughly 5,500 claims ACC receives every day. Now you might receive a text telling you your claim has been accepted as you hobble away from your GP clinic.
The previous approach was for every claim to be looked at by a claims assessor, which could delay an injured client receiving the support they need. The new system — which uses data built up over many years — removes much of this manual effort by automatically approving claims that fall within pre-defined criteria, allowing assessors to focus on complex claims.
“Core to this system is that it can’t decline a claim,” Lawrence says.
“Any complex or specialist claim, or any other claim that doesn’t meet the criteria for automated acceptance, is immediately diverted to a member of our specialist assessment team.”
As well as reducing the turnaround for decisions on claim eligibility, the system also helped ACC manage its workforce during the first Covid lockdown.
Lawrence again: “Claim volumes drop during lockdown, but they don’t stop. During the first lockdown our assessment workforce wasn’t yet set up to work remotely, so we made a call to change some thresholds in the model so that more claims would be approved automatically. This allowed us to adapt to a new way of working without leaving lots of injured people waiting to hear what’s happening with their claim.”
This model has led to growing use of analytical techniques to improve the experience of ACC clients. Or as Lawrence said: “Data helps us quickly direct clients to the right level of support, identify services or equipment that will help with recovery, and spot individuals that may be at higher risk of re-injury.”
Can you weld a digital mailbox?
Data is playing an ever-increasing role in New Zealand businesses, and not just because of what the global behemoths are doing or through local roll-outs of global solutions.
Kiwis are taking good ideas and adapting them to our needs, just like we did with Popular Mechanics back in the day. We’re receiving better, faster and more targeted service across lots of domains as a result.
But the welded chain mailboxes have probably had their day (they’re particularly awkward if you live in an apartment).
