Fivetran contributes to 99% reduction in scam texts for Spark

“We’re investing in AI assistants for Skinny’s contact centers across chat and now voice, and real-time access to data is critical for success. Fivetran helps us bring together data from operational systems and other sources to create a holistic view in near real time.”
— Peter Langham, Domain Chapter Lead for Data Engineering at Spark

Spark is New Zealand’s leading telecommunications and digital services provider, offering millions of customers a wide range of mobile, fixed, and IT solutions. As a key player in the industry, Spark is dedicated to enhancing customer experiences and navigating the evolving digital landscape through innovative data solutions.

In 2024, Spark completed a multi-year data platform modernisation initiative, moving to a centralised, cloud-native data architecture and decommissioning legacy systems. Spark’s Domain Chapter Lead for Data Engineering, Peter Langham, says the move has enabled the use of data to optimise business operations and make bold AI plays for the future.

Spark is now focused on maximizing the value of its data platforms by exploring innovative solutions in generative and agentic AI to help New Zealand thrive in a digital world.

Fivetran has played a key role in accelerating the onboarding of new data sources while significantly reducing operational overhead, now managed by just one person from the data operations team. Previously, the team faced frequent data and pipeline failures that required round-the-clock triage.

“As a result, our data engineers can now focus on high-value initiatives such as data curation and emerging AI projects, rather than basic ingestion processes,” Langham says.

Real-time data helps power 99% reduction in scam texts

A standout achievement of Spark’s modernisation with Fivetran is its powerful new anti-scam SMS solution, launched in November 2024. Built on Singleview data ingested via Fivetran’s change data capture (CDC), the system uses call and message metadata to detect and block scam activity up to 5x faster than before.

This solution is part of Spark’s broader fraud prevention strategy, which also includes a new SMS firewall. Together, these efforts have driven a 99% reduction in scam texts across the network, dramatically improving customer safety and trust.

The impact has been so strong that scammers are actively avoiding Spark’s network, targeting less-protected providers instead. In February, Spark customers accounted for just 1.8% of all scam reports, according to New Zealand’s Department of Internal Affairs.

“The solution detects anomalous SMS usage patterns and auto-blocks SIMs engaged in scam activity,” says Langham. “It’s built on data in Snowflake, and Fivetran is absolutely critical to getting that data with the low latency we need. This has been a major win for Spark and for our customers.”

Data unlocks 90% operational efficiency gains

Spark’s switch to real-time data has supercharged operational efficiency across the business. In retail, sales dashboards powered by NTSPOS data ingested via Fivetran’s CDC now offer instant visibility into activity at 60 stores, driving faster stock adjustments and better customer service.

For the Call Investigation team, Fivetran and Snowflake have unified CRM, billing, and other systems into a single, near-real-time view. Simple 6-month data requests now take under 1 minute (versus 15 minutes before), and complex historical queries drop from overnight to under 10 minutes — a 90%+ efficiency gain.

“Assembling data used to span multiple systems and hours of manual work,” says Langham. “Now requests are answered almost instantly, freeing our team to focus on higher-value investigations.”

How data is powering Spark’s autonomous AI future

Fivetran has become a key enabler of Spark’s AI-driven transformation, supporting initiatives across the entire business — including its low-cost brand, Skinny.

With a common data layer powered by Fivetran, Spark has launched a number of AI-driven data products, boosting productivity for frontline agents, B2B sales teams, and Consumer & SME Marketing. These AI projects rely on timely, unified data to deliver relevant insights and customer support at speed.

“We’re investing in AI for chat and voice at Skinny, and reliable real-time data is critical to making that work,” says Langham. “Fivetran brings all our systems together to create the holistic view we need.”

By modernizing its data infrastructure, Spark continues to unlock new efficiencies and accelerate innovation, strengthening its ability to help New Zealand thrive in a digital world.

[CTA_MODULE]