Host: Andrew Birmingham, Editor - CX | Martech | Ecom

Banks, telcos, and insurers are rethinking how they engage customers, shifting away from mass marketing campaigns toward real-time decisioning systems designed to respond to individual behaviour, according to Jonathan Tanner, a senior executive at Pegasystems.

Tanner said many organisations still struggle with fragmented customer experiences, where interactions across channels are disconnected and force users to repeat themselves. “They get a very jarring experience,” he said, pointing to structural issues such as product silos and outdated segmentation models that fail to reflect how customers’ needs change over time.

The emerging alternative is a decisioning approach that continuously evaluates customer context, including behaviour, signals and lifetime value, to determine the next best action. Unlike traditional campaigns, which Tanner described as a “blast approach” delivering only marginal returns, these systems aim to personalise interactions in the moment, sometimes choosing not to sell at all.

“What we’re talking about here is a very different approach,” Tanner said. “It may not even be a selling decision at that point in time… but over time what that does is it builds that NPS, it builds that customer connection.”

The shift requires a willingness to invest and the change. Firms are committing to significant investments annually over several years to build the underlying infrastructure. While returns can reach “multiple hundred percent,” Tanner said the gains depend on sustained investment and organisational change, not just technology deployment.

“You’re not going to just wake up, implement this technology, and then suddenly discover that everything’s great,” he said, noting that many firms underestimate the effort required to align people, processes and systems.

Artificial intelligence is central to the transformation, but Tanner warned against treating it as a single solution. Instead, organisations need to combine multiple approaches, including rules-based systems, statistical models and generative AI, each suited to different tasks.

“If I’m making a decision that’s backed up by a set of very well-defined rules, why would I be hammering away at an LLM spending tokens… and getting a probabilistic decision?” he said. Deterministic systems, he added, remain critical for real-time execution, compliance and auditability.

The stakes extend beyond marketing. Financial institutions are also using decisioning platforms to combat fraud, which is rising alongside real-time payments. Faster transactions benefit customers but also give fraudsters less time to be detected.

“One of the best ways of preventing it is to add just a little bit of friction into the process,” Tanner said, citing examples such as delaying payments to new accounts.

More broadly, Tanner said the most effective use cases focus on building trust rather than driving immediate sales. Examples include helping customers access government benefits or providing proactive support during financial hardship or natural disasters.

“The obvious immediate reaction is, well, how can that possibly be a benefit to the bank?” he said. “But of course… it’s building customer loyalty… it’s building connection.”

Looking ahead, Tanner expects the industry to move beyond the current hype cycle around AI and focus instead on practical outcomes. “I’d like to see us moving to it being more of a system-based conversation,” he said, where value is measured not by the technology itself but by the decisions it enables in real time.

Host: Andrew Birmingham, Editor - CX | Martech | Ecom
Banks, telcos, and insurers are rethinking how they engage customers, shifting away from mass marketing campaigns toward real-time decisioning systems designed to respond to individual behaviour, according to Jonathan Tanner, a senior executive at Pegasystems.
Tanner said many organisations still struggle with fragmented customer experiences, where interactions across channels are disconnected and force users to repeat themselves. &ldquo;They get a very jarring experience,&rdquo; he said, pointing to structural issues such as product silos and outdated segmentation models that fail to reflect how customers&rsquo; needs change over time.
The emerging alternative is a decisioning approach that continuously evaluates customer context, including behaviour, signals and lifetime value, to determine the next best action. Unlike traditional campaigns, which Tanner described as a &ldquo;blast approach&rdquo; delivering only marginal returns, these systems aim to personalise interactions in the moment, sometimes choosing not to sell at all.
&ldquo;What we&rsquo;re talking about here is a very different approach,&rdquo; Tanner said. &ldquo;It may not even be a selling decision at that point in time&hellip; but over time what that does is it builds that NPS, it builds that customer connection.&rdquo;
The shift requires a willingness to invest and the change. Firms are committing to significant investments annually over several years to build the underlying infrastructure. While returns can reach &ldquo;multiple hundred percent,&rdquo; Tanner said the gains depend on sustained investment and organisational change, not just technology deployment.
&ldquo;You&rsquo;re not going to just wake up, implement this technology, and then suddenly discover that everything&rsquo;s great,&rdquo; he said, noting that many firms underestimate the effort required to align people, processes and systems.
Artificial intelligence is central to the transformation, but Tanner warned against treating it as a single solution. Instead, organisations need to combine multiple approaches, including rules-based systems, statistical models and generative AI, each suited to different tasks.
&ldquo;If I&rsquo;m making a decision that&rsquo;s backed up by a set of very well-defined rules, why would I be hammering away at an LLM spending tokens&hellip; and getting a probabilistic decision?&rdquo; he said. Deterministic systems, he added, remain critical for real-time execution, compliance and auditability.
The stakes extend beyond marketing. Financial institutions are also using decisioning platforms to combat fraud, which is rising alongside real-time payments. Faster transactions benefit customers but also give fraudsters less time to be detected.
&ldquo;One of the best ways of preventing it is to add just a little bit of friction into the process,&rdquo; Tanner said, citing examples such as delaying payments to new accounts.
More broadly, Tanner said the most effective use cases focus on building trust rather than driving immediate sales. Examples include helping customers access government benefits or providing proactive support during financial hardship or natural disasters.
&ldquo;The obvious immediate reaction is, well, how can that possibly be a benefit to the bank?&rdquo; he said. &ldquo;But of course&hellip; it&rsquo;s building customer loyalty&hellip; it&rsquo;s building connection.&rdquo;
Looking ahead, Tanner expects the industry to move beyond the current hype cycle around AI and focus instead on practical outcomes. &ldquo;I&rsquo;d like to see us moving to it being more of a system-based conversation,&rdquo; he said, where value is measured not by the technology itself but by the decisions it enables in real time.

A weekly wrap of the “must-know” developments in Marketing, Media, Agency and Technology for leaders and emerging leaders in the industry. Veteran industry journalist and Mi3 Executive Editor Paul McIntyre talks each week with guest marketers who are in the know on what matters at the nexus of marketing, agencies, media and technology. Powered mostly by Human Intelligence (HI).

A weekly wrap of the “must-know” developments in Marketing, Media, Agency and Technology for leaders and emerging leaders in the industry. Veteran industry journalist and Mi3 Executive Editor Paul McIntyre talks each week...

Mi3 Audio Edition

‘Why hammer at an LLM spending tokens’: Pega’s Jonathan Tanner warns on AI hype; banks pivot to rules, context and real-time decisioning as CX, fraud prevention, and lifetime value collide

https://api.omny.fm/orgs/fa326977-3de5-4283-9b8b-af3500c58607/clips/f451d394-bd8d-4f71-ad67-b435003af55e/transcript