How KOHO's QA Overhaul Scaled Support, Not Headcount

When KOHO started scoring every customer conversation instead of a small sample, quality stopped being a compliance check and became the reason support could scale without adding headcount.

For years, most contact center QA programs have asked one question: did the agent follow the script. A supervisor pulls a small sample of calls, maybe one or two out of every hundred, checks them against a rubric, and moves on. Most of what customers actually say never gets reviewed at all.

KOHO, the Canadian fintech and digital banking app, thought there was a better approach and started scoring every single interaction instead. In financial services, a wrong answer can cost more than a bad review. The wrong answer could lead to escalated complaints that have real operational costs, or worse, potential regulatory impact.

"We're a high-stakes product," said Monika Aufdermauer, VP of User Success at KOHO, during the recent webinar, How KOHO Connects Quality to Customer Experience, Performance & Retention. "We're dealing with people's money, so people getting the right answer is really important."

From sampling to seeing everything

Using Birdie.ai’s Frontline Intelligence, KOHO moved to 100% coverage, evaluating every interaction instead of the small, manually reviewed sample most contact centers rely on. 

That removed something QA teams rarely talk about openly: sampling bias, the risk of skipping a call, or letting a familiar name slide. "AI is looking at 100% of everything, the same exact scoring across all conversations," Monika said. It also meant KOHO no longer had to trust an outsourced team to grade its own work. "I'm not a fan of having the BPO say, here's what we're doing, and giving their own scores. I think it's much better if it's an independent party [reviewing performance]."

That shift freed her team from manual review entirely. The people who used to spend their days checking boxes now spend it coaching and training, working from what full coverage actually surfaces. Feedback that used to take two weeks now reaches an agent's Slack the moment a call ends. "A lot easier to make changes on what you're doing when feedback is in the moment," she said.

What full coverage found

Once KOHO could see everything instead of a sample, they found something they didn't expect. Grammar mistakes, long treated as a serious QA issue, turned out to barely matter. Empathy did. Her team's coaching shifted accordingly, less on script perfection, more on the human connection that actually moves outcomes.

That same shift changed how KOHO decides what to fix at all. A simple model now shows the team where to actually spend their effort: deflect traffic with a better bot, eliminate it entirely by fixing the product, or improve process and training for what's left. Eliminating traffic is Monika's favorite of the three, working with product to remove the need for a contact in the first place. 

"The best support experience," she said, "is one that doesn't have to happen." 

The results back that up. When Monika joined KOHO, about 26% of customers contacted support every month. Today, including AI-handled conversations, that's down to about 8%, driven mostly by product and process fixes, plus better customer education on what the product can and can't do.

Scale through technology, not headcount

None of this replaced KOHO's team. It changed what the team spends its time on. "I want to use AI to make my humans smarter," Monika said, "to free up their time to do things worth doing. I don't want them clicking a button to process a refund. Those things don't require human thought, empathy, or human skills."

That philosophy shows up directly in the numbers. Since Monika joined, KOHO's support team has stayed roughly the same size, while the user base has doubled. "It's scale through technology instead of scale through humans," she said.

The scorecard itself stayed disciplined about what it rewards. "The biggest weighted part of our scorecard is, did the customer get the right answer," Monika told me. "If they're not nailing that piece, they're never going to be able to game a good score."

Where quality is headed next

Monika thinks about the work in two connected layers, what she calls the inner loop and the outer loop. The inner loop is individual learning, changing a specific behavior, sharpening how one agent connects with a customer. The outer loop is structural improvement, making sure an issue gets routed to whoever actually owns the fix, and prioritized correctly once it gets there. 

Keeping those two loops connected is what turns a single finding into two things at once: a coaching moment for one agent, and a signal for what the whole team, or the product itself, needs to change.

Growing that connection is what's next for KOHO. They're not only reviewing what already happened anymore. They're trying to see it coming. "I want to understand what comes next for users," Monika said, "so we don't have to wait for them to contact us." If the data shows early signs someone might churn, her team can reach out before it becomes a support ticket at all, connecting customer context across the account, not just the conversation.

"It really wasn't about the score," Monika said, looking back on the whole journey. "The score is a means to an end. Getting to 100% coverage, automating our flows, impacting the business, partnering with product, understanding what comes next, that's where we're at today."

KOHO didn’t just automate their QA. It changed what quality means as a function, from a score you defend to a signal you act on before the customer ever has to ask.

Watch the webinar, How KOHO Connects Quality to Customer Experience, Performance & Retention, to see more on KOHO’s transformation.