NZ Post: AI for commercial gains and customer trust

Published on the 26/09/2023 | Written by Heather Wright

NZ Post: AI for commercial gains and customer trust

Reinforcing customer trust through AI…

Chris O’Brien is a believer in the power of generative AI for commercial and customer gains – and to build trust. 

As New Zealand Post’s digital transformation lead for data and AI, and leader of two NZ Post squads – the digital communications centre of excellence and the future fit data squad – O’Brien has set himself the lofty mission of revolutionising digital marketing and customer experiences with AI and data-driven communications that deliver commercial results. 

NZ Post, which as New Zealand’s largest delivery and courier company has around 5,000 employees and 850 retail stores, has a number of AI experiments underway.

“The learnings are as valuable as the productivity gains this early in the game.”

Speaking at Gartner’s IT Symposium and Xpo recently, he said much of the focus over the past 18 months has been on the use of predictive AI models. In particular, the company is having success with using propensity and predictive modelling to slow the flow of calls to the contact centre, reinforce trust and save money.

“We’ve been looking at customers’ propensity to take particular actions or use, or not use, particular products, but a big one for us is around our call centre and predicting when a customer is going to call in,” O’Brien says.

The company has ‘hundreds of thousands’ of calls a month into its call centre. But when it analysed the data on those calls it discovered that 22 percent of calls were about deliveries that we still within SLA – parcels that weren’t actually due to be delivered yet, but which customers had high anxiety around. 

“The approach we have taken is using some of that propensity and predictive modelling and different data sources including real time parcel data, to predict that call and what it might be about.”

O’Brien and his team have taken that one step further, looking at how they can proactively communicate with the customer before they pick up the phone, to put their mind at ease that the parcel is on its way. 

It’s a project which has deflected a ‘huge’ number of calls from the call centre, saving NZ Post money while also enhancing trust at a moment when trust is at risk with the customer. 

“We have that vision and mission around driving customer and commercial needs in real time and this as a use case ticks the boxes. I’m really proud about the work that is happening there.”

While many companies are keen to experiment with AI and generative AI, the issue of not having their data in order is a big stumbling block for many. 

O’Brien is open that just like other companies, NZ Post has data scattered across disparate systems.

“We have implemented [Salesforce] Data Cloud and we have been going pretty hard at trying to get to that holy grail of a 360 degree view of our customers.“But it’s not easy. We are getting closer, but it is still challenging,” he admits.

NZ Post has also stood up a proof of concept for an email copywriting bot, leveraging generative AI.

The company produces more than 300 corporate communications emails from scratch a year, ranging form B2B and B2C marketing to crisis comms around natural disaster impacts and compliance related comms. Even with that volume, the on-team copywriter is still turning away work from other divisions within NZ Post.

“With that lens we are looking at some of our historic data and going back five years of really rich email content that is well tagged – so we have this incredible backlog of content available,” he says.

“Leveraging that tool and with [the copywriter] being involved, we are able to support amore teams across the business that we normally say no to, and we can support them at pace as well.

“It’s a way to bring the [copywriter’s] magic across the entire organisation, rather than just the three or four teams she can support at the moment.”

O’Brien admits there were some ‘robust’ early conversations when the proof of concept was stood up.

“There was a lot of excitement but also a lot of nervousness and concern and it broke down some of those silos within the organisation and forced what was initially perhaps uncomfortable conversations but what ended up as acknowledgement that this opportunity and the potential for productivity gain across the business was just too good to let silos get in the way of.”

One key factor in getting NZ Post on the AI and generative AI experimentation path has been a very task-based approach, O’Brien says. It’s an approach which requires identifying where you want to use AI and where it can be responsibly used, clarifying the tasks that AI will actually do and accessing the risks in context.

“It’s that task based approach and going after one use case – for us it was the email content. It was low risk with no sensitive customer information or personally identifiable information, it was really publicly available data or copy that was on our website or had previously been emailed to customers.

So what’s his advice to other companies looking to start experimenting?

“Ring-fence a use case that is low risk and just go at it as fast as you can!,” he says.

“We can see the productivity gains – they absolutely are already there. But the learnings are as valuable as the productivity gains this early in the game when we are all trying to keep up and even get ahead.”

Post a comment or question...

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.


Thank you! Your subscription has been confirmed. You'll hear from us soon.
Follow iStart to keep up to date with the latest news and views...