Westpac NZ's 55,000 freed hours: how to measure your own AI gains

What Westpac reported

Westpac NZ released its result for the year to 30 September on 2 November 2026. The bank reported net profit of $1,087m, down 9% on 2025, with operating expenses up 4% to $1,532m. Westpac said part of the cost rise came from ongoing investment in technology and tools for its staff. Chief executive Catherine McGrath linked that spending to concrete operating results. She said 55,000 hours had been freed for bankers to help customers. Small business onboarding fell from an average of three weeks to 24 hours, and most home loan applications are now processed the same day or the next.

The Post ran its coverage under the headline "Westpac's big AI year". Westpac's own release talks about "new technology" and doesn't break down how much of the 55,000 hours came from AI. Some could have come from workflow automation, process redesign, or the Westpac One Business platform, which more than 25,000 businesses now use. The bank's best-documented AI deployment is Microsoft Dynamics 365 Contact Centre as a Service. Westpac announced that platform in April, when it said the AI would show a caller's banking profile and relevant product details to staff in real time, and planned to have it in every contact centre by August.

Westpac's workforce numbers stayed steady through all of this. BusinessDesk reported that full-time-equivalent staff were roughly unchanged at about 5,240. NBR reported that the bank committed to its branch network and ruled out a mass restructure of its local workforce while it keeps investing in technology.

Reading the numbers

The 55,000-hour figure is the one that will get quoted, but on its own it's a modest number. Spread across about 5,240 full-time staff, it comes to roughly ten and a half hours per person over a year, or less than two working days each. That's our own calculation from the published figures. Westpac hasn't said how it counted the hours, which teams they came from, or whether the total is concentrated in a few high-volume functions such as contact centres and lending operations. A concentrated saving would mean far more for those teams than the per-head average suggests.

The onboarding figure says more about how the work has changed. Cutting small business onboarding from three weeks to 24 hours mostly removes waiting time, and very little of that comes from faster typing. Elapsed time in an onboarding process usually builds up in queues, handoffs between teams, document checks and back-and-forth with the customer. When a bank goes from three weeks to a day, someone has redesigned the sequence of steps as well as adding tools. The customer gets a working account sooner, and the bank starts earning from that relationship sooner. Neither benefit shows up in an hours-saved total.

Westpac has also said where the freed time went: back into customer conversations. That claim matters because hours saved by AI often vanish into other work, with nobody able to say what happened to them. A firm that can't say where reclaimed time went will struggle to defend the spending at its next budget round.

Every one of these figures is self-reported in a results release. Westpac hasn't published a baseline method or an independent review, so treat the numbers as the bank's account of its own progress and not as a benchmark.

Why NZ businesses should pay attention

Most New Zealand firms deciding on AI aren't short of tools. They're short of evidence that the tools pay off. Prospa's latest SME Sentiment Tracker, a survey of 501 NZ small business decision-makers, found that 34% of firms not using AI cite uncertainty about return on investment and 32% aren't sure which tool fits their needs. Upfront cost was cited by only 12%. The same survey found 59% of SMEs now use AI in some form, up from 50% six months earlier, so the firms still holding back are increasingly the ones waiting for proof.

Westpac's reporting gives those firms a format they can copy at a much smaller scale. The bank picked processes with a clear start and end point and reported elapsed time before and after. It also stated where the freed capacity went. A 20-person distributor or professional services firm can do the same for customer onboarding, quote-to-order, supplier setup or month-end close, provided it records the starting position before anything changes.

The workforce stance is also worth copying. Westpac told the market it isn't planning a mass restructure. Staff who'll be asked to document how their work really gets done, including the workarounds, are far more likely to cooperate when they know where they stand. Firms that leave the question open tend to get cautious, incomplete process maps, and AI built on an incomplete map automates the wrong steps.

Practical next steps

• Pick one workflow with a measurable wait. Customer or supplier onboarding is a good candidate because the start and finish are easy to timestamp. Record the current elapsed time, the hands-on time, and the number of handoffs before you introduce any AI.
• Write down how you'll count saved hours. Decide upfront whether a saved hour means less hands-on effort, fewer rework loops or shorter queues. Keep the method the same every time you report.
• Report cycle time alongside hours saved. A drop in elapsed time often matters more to customers and cash flow than a total of staff hours.
• Say where freed time goes. Assign reclaimed capacity to named work, such as customer follow-up, backlog reduction or quality checks, and track whether it actually gets used that way.
• Separate AI from process redesign in your reporting. When you remove a handoff and add an AI step at the same time, log the two changes separately so you know which one moved the number.
• Tell staff the workforce position early. A clear statement on roles before mapping starts gets you more accurate process detail.

If you're weighing up where AI would earn its keep in your operations, our AI consultancy work starts with this kind of baseline. Workflow redesign that cuts elapsed time sits with our process and automation team.