Intelligent guidance: designing for technological humility
Enterprise IVR-to-digital redirection pattern for a top-5 US bank
role | Conversation designer & strategist (via Slalom) |
duration | 3 months |
methods | Cross-functional workshops, flow design, scorecard development |
tools | Lucidspark, VUI design frameworks, Redirection Readiness Scorecards |
deliverables | Adaptive IVR flow system, SMS handoff design, stay-with-me framework, evaluation scorecards, replicable playbooks |
impact | Shipped to production as reusable IVR-to-digital pattern; adopted by additional lines of business as a standard |
the short version: I was designing IVR-to-digital redirection flows for a bank and the original design assumed the NLU could perfectly classify customer intent. It couldn't. Instead of fighting that limitation, I redesigned around it — creating a single adaptive flow that bundles both general knowledge and personalized information, then lets the customer choose how to proceed. The multi-modal handoff respects user agency at every transition. What started as one use case became repeatable infrastructure that other teams could stand up without starting from scratch.
the setup
Customers calling a bank's IVR system often need information they could get faster on the website. Checking a rewards balance, updating an address, finding a policy answer. The opportunity was pretty clear: redirect them to digital self-service when it would genuinely serve them better.
The key word there is genuinely. We weren't trying to dump people off the phone to save the bank money. We were trying to get people to the right channel for the right task — and respect their choice either way. The hard part was doing that without turning "digital redirection" into a cost-cutting euphemism.
the assumption
The original design made a big bet: that the NLU engine could reliably distinguish between “how do rewards work” (a general knowledge question) and “check my rewards balance” (a personalized account action). These sound different to a human. To a natural language understanding model, they're pretty dang close.
The plan was to use that classification to route customers down two completely different paths. General knowledge gets an article. Personalized action gets an account link. Clean, elegant, logical.
One problem: it didn't work. The NLU couldn’t make the distinction reliably.
Why this was risky: forcing customers to articulate their needs with the precision required for accurate intent classification meant brittle intents and fragile branching logic that would only grow more complex over time. Every new use case would mean more branches, more edge cases, more failure modes.
the breakthrough
So we stopped trying to be perfect and started being helpful.
Instead of complex branching based on intent detection, we designed a single adaptive flow that bundles both general knowledge AND personalized information. If someone calls about rewards, they don't need to specify whether they want to learn about the program or check their balance. We give them both.
This sounds obvious in retrospect. The best design solutions usually do. But getting there required the team to let go of a technically elegant architecture in favor of one that actually worked for humans. That's harder than it sounds, especially when the engineering is already underway.
Design principle: Meet users where they are. Don't assume they’ll perfectly articulate their needs. Don't make AI perfectly classify intent. Design for graceful uncertainty.
the flow
Here's what it looks like in practice. A customer calls and says something about rewards:
The system gives a quick general answer — "rewards are calculated based on your card type" — while simultaneously checking their account eligibility. Then it offers a choice: "I can send you an article about how rewards work, plus a link to your personal rewards balance. Would you like that?"
The customer says yes, and they get an SMS with two separate links: one to a knowledge article, one to their personal rewards hub.
Design decision: two links, not one. Cramming everything into a single destination would have forced the same classification problem we just escaped. Education and account action serve different needs and deserve different paths.
Simple. No branching logic nightmare. No NLU gambling. One flow that adapts.
the “stay with me” pattern
The multi-modal transition — voice to SMS to web — is where most systems fumble. You're asking someone to leave the channel they chose (the phone) for a channel you're suggesting (their phone's browser). That only works if every transition respects their agency.
So we designed what we called the “Stay With Me” experience. And it turns out, the right language depends on what the customer is doing.
For task completion — like updating an address — the IVR says: “I’ll stay with you while you update your address.” Active monitoring. The system is your buddy.
For knowledge consumption — like reading a policy article — the IVR says: “The line will stay open in case you have questions.” Passive availability. The system is a safety net, not a supervisor.
That distinction matters. Reading an article while someone watches you feels weird. Reading an article knowing help is a sentence away feels supportive. Same technology, completely different emotional experience.
And crucially: “If you have everything you need, you’re welcome to hang up.” Always an exit. Never a trap.
the scorecards
To figure out which use cases were ready for digital redirection, I developed Redirection Readiness Scorecards — a quantifiable framework that any team could use to evaluate whether a particular task was a good candidate.
We scored each potential use case across multiple dimensions: task complexity, digital experience quality, authentication requirements, user digital engagement history. Address change scored 41/50. Rewards scored 45/50. The scores gave us a shared language for prioritization and made the decision-making process transparent instead of opinion-driven.
More importantly, they made the framework replicable. Any line of business could evaluate candidates for redirection without waiting on a centralized conversational AI team.
the outcome
The pattern shipped and is now the default way this bank handles IVR-to-digital handoff for similar use cases.
One adaptive flow instead of complex NLU-dependent branches. SMS with two links — knowledge article plus personal account action — instead of forcing a single path. “Stay With Me” language that offers support without forcing it, calibrated to the type of task. Playbooks and evaluation frameworks so other lines of business can stand up their own Intelligent Guidance systems without starting from scratch.
As a consultant, I don't own the long-term metrics. but i do know this: the framework shipped, it's live, and it's scaling. No mean feat for an enterprise like this one.
the real lesson
Sometimes the most elegant solution is to stop trying to be perfect and start being helpful. We turned a technological limitation into a design improvement. By accepting what the NLU couldn't do, we created an experience that was simpler to build, easier to maintain, and more respectful of the humans on the other end of the line.
Every design choice is a values statement. Routing a customer to the right channel isn't a cost optimization play — it's an act of respecting their time and their agency. The technology should serve that, not the other way around.