Vero Insurance adds AI sentiment checks in New Zealand

Vero Insurance has introduced AI-based sentiment analysis across its sales and service teams in New Zealand. It says it is the first Salesforce customer in the country's financial services sector to adopt the tool.

The system reviews customer phone calls and emails in real time and classifies sentiment as positive, neutral, or negative. When it detects negative sentiment, it prompts consultants to review the interaction and decide whether action is needed.

It also flags possible complaints and signs of customer vulnerability. Vero says this is intended to help staff identify problems earlier, rather than relying on customer surveys, formal complaints, or manual quality checks after the event.

The rollout followed a pilot covering about 45,000 customer interactions over several months. Since the wider launch, the system has produced more than 65,000 sentiment outcomes and now analyses about 10% of customer email and voice interactions in real time.

Managers previously reviewed only a small sample of calls and messages each month as part of quality assurance. Vero says the new setup gives leaders visibility across about 58 customer interactions per consultant each month, nearly 30 times more than under its earlier approach.

Operational shift

The move reflects a broader effort by insurers to use artificial intelligence in customer service, particularly in areas where claims and service discussions can involve financial stress, disputes, or support requests. In that context, detecting dissatisfaction during an active case can affect how quickly a company responds.

Nic Dorward, Executive Manager - Consumer Operations at Vero, said the change was designed to enable earlier intervention and make it more consistent.

"Insurance often comes at a stressful time in someone's life. If we can better understand how a customer is feeling while we're still working with them, we have a much greater opportunity to step in, resolve concerns, and provide the right support before those issues escalate."

Under the process described by the insurer, sentiment analysis is applied as cases close, and the system flags signals of dissatisfaction, complaints, and vulnerability. If negative sentiment is identified, a follow-up task is created for consultants to validate the result and address any unresolved issues where appropriate.

The setup broadens the volume of customer contact that can be reviewed without requiring managers to listen to or read interactions manually at the same rate as before. It also shifts supervisors' roles from sampling a narrow set of conversations to overseeing patterns across teams and channels.

According to Vero, that broader view is intended to support staff coaching as well as customer oversight.

"Historically, managers could only manually review a small number of customer interactions each month. Sentiment Analysis gives us a much broader view of what's happening across our customer conversations, allowing us to identify trends, coach our people more effectively, and continuously improve the experience we deliver."

Customer support

Beyond individual cases, Vero says the tool gives it greater visibility into themes emerging across customer communications, including differences between teams, channels, and partners. That information can then be used to change processes based on direct interaction data instead of relying mainly on post-interaction feedback.

The insurer also linked the technology to its handling of vulnerable customers. By identifying potential warning signs earlier, consultants can direct customers to additional support services, including Vero Support, which offers practical assistance and specialist support.

For brokers, earlier detection could help resolve issues affecting shared customers before they develop into formal complaints. That may be particularly relevant in claims and servicing journeys where insurers and brokers both have contact with the same policyholder.

New Zealand insurers have been under continuing pressure to improve customer treatment, especially in sensitive interactions involving loss, damage, or financial strain. Tools that scan communications in real time offer one route to earlier escalation, although final judgment still rests with staff reviewing the alerts.

Dorward said the system was intended to support, rather than replace, human decision-making.

"Customer advocacy starts with listening and acting with care. Sentiment Analysis helps our teams recognise when customers need us most, strengthening how we show up with empathy and intent."