4Front Credit Union Designed the Experience. Then They Built the Stack to Run It.

Diagnosing the real cost

4Front's call volume came in peaks and valleys, and the leadership team could put a dollar figure on the peaks.

To cover them, the credit union used a third-party rollover phone service that was limited in what it could do once it picked up. The vendor wasn't authorized to handle some requests, so roughly 30% of those calls came back to 4Front's own team. 4Front tracked that number, and it made the math clear: they were paying a premium for calls their own people ended up handling anyway.

"We employ a third-party rollover phone service and it was expensive," said Kent Nordin, Vice President of the Communication Center. "That's what got us started down this road. We wanted to smooth out those bumps."

Before spending anything, they tested the cheapest option available: routing overflow to the branches. They measured the effect on the in-branch experience and ended it quickly.

"It was an experiment that didn't provide the results we wanted," Kent said. "Branches are busy. Too many phone calls, and the people walking in noticed it."

Two weeks of data beat a year of debate. Overflow capacity had to come from somewhere other than people, and it had to be capacity 4Front controlled.

Setting the terms of implementation

4Front went into vendor selection with a specific requirement, and it wasn't a feature.

"Every vendor says there's no big IT lift, and 80% of them are not accurate. We got a very accurate explanation of the IT lift from Posh. We planned for the worst and hoped for the best."

They resourced the project against that estimate. The technology stood up fast. The content work took longer, and 4Front had scoped for it: the team built 833 intents for their Voice Assistant, who members know as Penny, and it was the most demanding part of the project. Posh has since retired intents entirely, so no institution builds them today.

It also produced something durable. Those 833 intents are the most detailed map anyone has of what 4Front members actually call about, and the team has reused that data.

Since Posh has retired intent-based configuration, 4Front is applying this new shift to everything that comes next.

"We built hundreds of intents for Penny. Now for digital, we're letting the AI scrub all the info and we won't build any intents."

Same credit union, same content, same member questions, none of the build work.

How 4Front decided it was working

Kent set his own bar for what success would look like, and it wasn't a dashboard metric.

"I came from the branch world. Three things I looked at in one day tied it all together," he said. "The labor we were spending and where that spend was coming down. The topics Penny was directing, and how accurate she was with very few unknowns. And a member survey comment that said, 'I love that at any time I can call in and check, but when I need someone, you answer quickly.'"

Cost, accuracy, and member sentiment, all moving in the same direction, and a service level the team is now known for.

The third-party contract is still in place, but it barely does anything now. "We've reduced our cost with them by 78% over the last few years. They're handling 78% fewer calls."

4Front's phones close at night. Members self-serve around the clock.

"They no longer have to answer the basic questions. High repetition, low skill," Kent said of his team. "That's helped our team focus on taking care of the members who actually need them. That's the whole point."

The principle that drives every channel decision

4Front has a clear rule for what goes to software and what goes to people, and it governs where they deploy.

"Our self-service tools are there for the low emotion needs," said Luis Pardo, Chief Experience Officer. "And our amazing humans are there to help with the high complexity, high emotional needs."

Applied to the website, that pointed at Posh Answers. The top searches proved the thesis: loan information, branch locations and contact details, and rates. Three categories that drive enormous inbound volume at any credit union, and three where a member wants a fact, not a conversation. They're also buying signals, and 22% of searches end with the member taking the next step.

Digital Assistant is in implementation now, built with no hand-configured intents. 4Front's own traffic data set the priority: the credit union sees far more calls and chats between 7pm and 11pm than between 7am and 11am.

"High volume, low complexity chat requests can be offloaded so my team can spend more time on the complex situations and wow them with that experience," Kent said.

Holding a new product to their own standard

The Posh Simulator had to clear a rule Luis applies to every purchase.

"I always want to make sure we have a clear problem to solve before looking for a solution," he said. "But once we tested it, it actually uncovered a problem we didn't know we had."

In a sales and service environment, there had only ever been one way to practice a hard conversation: with another human. A manager. A peer. A role play nobody wants to do. 4Front's pilot surfaced why that ceiling existed everywhere, not just at 4Front.

"What was unexpected to me is that people don't want to practice with other humans," Luis said. "They prefer being able to sit down privately and practice a scenario over and over again until they feel they got it right."

The scenarios 4Front built aren't transactional. They chose fraud calls. Identity theft. The moments that decide whether a member stays.

"That is a very vulnerable time for a member. They often come to us scared, sometimes embarrassed. With the  Simulator, we can really ratchet up the requirements on empathy and understanding. That way folks can practice with the simulator instead of practicing on our members."

Voice, Answers, and Digital Assistant all point at members. Simulator was the first product 4Front pointed at the people serving them, and it set the direction for two more.

Extending the same logic to the employee experience

Knowledge Assistant answers a problem every growing institution has: policy, product, and procedure change faster than any intranet or internal channel can keep current. An intranet search is only as good as its tagging on the day someone needs it, and a busy Teams channel moves faster than anyone can read. Kent knows exactly where that cost lands.

"When you're new you need a huge glossary of acronyms and you need to know how to do things. And as soon as you learn how to do something, it's dated."

CoachQA kicks off in October, and Kent's reasoning starts with fairness rather than coverage. Manual QA hits the same ceiling at every contact center in the industry. Two analysts can score roughly five calls per rep per month, and those scores feed KPIs and performance reviews.

"Last month our highest volume phone rep had 1,388 calls. He had five calls scored. On election night, they don't call five voters and call the election."

Kent isn't willing to build a performance conversation on a sample that small. When reps question it, he thinks they have a point. 4Front is solving that before anyone asks them to.

Why 4Front consolidated on purpose

This is the part Kent is most deliberate about, and it's the reason 4Front kept building on the same platform.

"If we had a different vendor for every system, the systems wouldn't communicate or leverage each other. When we get ready to launch CoachQA, we can marry that with Simulator. We have a scoring template and we can leverage it across both."

One scoring template, both products, both directions. Coaching finds the weakness and training fixes it, without anyone rebuilding a rubric in a second system. What 4Front learned about why members call didn't stay with Voice either. The team used it to decide what Answers should publish and which scenarios new hires actually need to rehearse.

Every product 4Front adds makes the next decision faster to execute and better on day one.

Want to see what this looks like at your institution? Let's talk.