1.1k
Case study
Virtual Chat Assistant Study
Care Advocacy was considering a virtual chat assistant. The question wasn't whether automation could handle some tasks — it clearly could. The harder question was whether patients would accept it without feeling like their access to a human had been quietly replaced. The answer depended almost entirely on how it was introduced, and where it sat in the flow.
tl;dr
- Patients preferred clear communication before the change, with email emerging as the strongest introduction path.
- The ability to bypass the assistant and reach a human Care Advocate ranked as the most valuable capability.
- Immediate responses and 24/7 availability were appealing, but only when paired with easy access to human help.
- Starting with a virtual assistant was mostly tolerated or disliked, while human-first interactions with chat as an option were more acceptable.
- Being transferred from a Care Advocate to automation was more strongly resisted and reinforced the importance of human continuity.
What We Learned
Patients Wanted Clear Communication Before the Change
The strongest preference in the study was not about a feature. It was about how the change should be introduced. Most patients wanted a direct explanation by email before encountering a virtual assistant inside the experience.
Preferred introduction channel
Email was the clearest way to introduce the change
That mattered because a meaningful portion of respondents also said they did not want to be introduced to a virtual assistant at all. The finding suggests that communication needed to do more than announce a new tool. It needed to frame the assistant as useful, explain how it would affect care, and reduce the fear that human support was being taken away.
Convenience Was Valuable, but Human Bypass Was Essential
When patients ranked potential capabilities, the most valued feature was the ability to bypass the virtual assistant and reach a human Care Advocate. Immediate answers and 24/7 availability followed closely behind.
This ranking showed that patients did see real value in automation. They wanted speed, accessibility, and help with routine tasks. But that value only held if the system preserved a clear escape hatch back to a person.
Attribute ranking
Human bypass outranked every other capability
| Item | Rank | Score | Rankings |
|---|---|---|---|
| Ability to bypass the virtual assistant to reach a human Care Advocate | 1 | 950 | |
| Ability to respond immediately to my questions | 2 | 940 | |
| 24/7 availability | 3 | 941 | |
| Ability to refill medications for me | 4 | 945 | |
| Ability to leave a message for my provider/care advocate | 5 | 951 | |
| Ability to schedule an appointment for me | 6 | 935 | |
| Always available wherever you are in your account | 7 | 918 | |
| Ability to help with technical support | 8 | 917 |
In other words, convenience was attractive, but not on its own. The assistant became credible when it felt like a support layer rather than a gatekeeper.
Starting with Chat Was More Tolerated than Desired
Starting Care Advocacy interactions with a virtual chat assistant produced a mixed but cautious response. Many patients said they would dislike starting this way, while a smaller group said they could live with it or might even like it.
Start with chat
The more favorable reaction came from a different model: beginning without a virtual assistant but having the option available later. That shift matters because it shows patients were not rejecting automation in principle. They were rejecting a loss of control at the beginning of the interaction.
Start without chat
Kano framing
Starting with chat landed in the low-value, low-tolerance zone
Expectation match / performance
Attractiveness / tolerance
Attractive
Performance
Indifferent
Must-be
Transfer to Automation Triggered Stronger Resistance
The sharpest warning sign in the study came when patients were asked how they would feel about being transferred from a human Care Advocate to an automated assistant for certain tasks. That move produced stronger resistance than simply offering chat as a possible starting point.
Transfer to chat
The contrast with “never transfer me” was especially revealing. Patients were clear that human interaction still felt like a baseline expectation of Care Advocacy. Automation could be accepted when it supported the experience, but it became risky when it interrupted continuity or looked like a replacement for the person already helping them.
The implication for rollout is clear: automation works as a layer, not a substitution. How it's introduced determines whether patients trust it or resent it.
Never transfer
Outcome of the Research
The study gave the team a clearer set of guardrails for introducing automation into Care Advocacy. It showed that virtual chat assistants could add value, but only if they were framed as supportive, optional, and clearly subordinate to human care.
That shifted the conversation away from “Should we have a chatbot?” and toward a better product question: “Under what conditions does a chat assistant feel like a benefit instead of a loss of human support?”
Implications
- Introduce automation through clear expectation-setting before patients encounter it in care flows.
- Preserve immediate access to a human Care Advocate as a trust-preserving fallback.
- Position chat assistants as support for routine tasks, not as a replacement for human continuity.
It also reinforced that attitudes toward automation can be misleading if you only ask whether people “like” the concept. Patients were far more nuanced than that. They could appreciate convenience in theory while still rejecting the exact moments where automation interrupted human continuity.
It also showed the value of pairing feature ranking with a model like Kano. That combination made it easier to separate what felt attractive from what simply felt tolerable, and to identify where product decisions could unintentionally introduce dissatisfaction even when the feature set looked strong on paper.
Recommendations
The Assistant Needed to Be Introduced as Support, Not Substitution
Clear Communication
Introduce the virtual chat assistant through multiple channels, including email, in-app notifications, and explanations during interactions with Care Advocates.
Human Interaction
Ensure that patients have the option to bypass the virtual assistant and reach a human Care Advocate when needed.
Training and Support
Provide training and support so patients understand how to use the virtual chat assistant effectively when it can save them time.
Feedback Mechanism
Implement an ongoing feedback loop to gather patient input and improve the experience as expectations evolve.
Continue Exploring
Contact
Want to talk through the chat assistant study?
The Kano framing, the transfer-to-automation finding, or how this shaped the rollout recommendation — happy to get into any of it.
A good conversation is usually the best start.