Chatbot Appointment Scheduling is Great for Business!
RestcommONE Marketplace Partner Blog Series
In today’s blog post, let’s look at appointment scheduling and appointment rescheduling.
Ben Franklin famously observed in his 1748 piece, Advice to a Young Tradesman, that “time is money.” If you’ve been idle, he explained, you’ve “spent, or rather thrown away,” the money you could have made. I like to think of it as having failed to create the value for others that you might have created.
Appointment-driven businesses such as dental clinics and salons are acutely aware of this equation. When a patient or client doesn’t show up, staff and building costs dissipate uselessly at $100 to $200 an hour. No-show rates range from 10% to 30% for most verticals, and that’s counting only un-canceled appointments! That’s a tragic waste of time and talent. One widely quoted vendor report pegs the aggregate cost of this problem nationwide at $150B.
Other non-economic costs are at least as important as the money. For a medical or dental practice, the patient may be missing an essential opportunity for required care. Depending on a range of factors including prior history and any in-progress diagnosis, a no- show may produce malpractice risk and a legal obligation to follow up.
In the age of AI, this just shouldn’t happen. Of course, many obstacles such as lack of cash, lack of time, lack of transportation, and emotional barriers are hard to control. But in the majority of cases, cancellation or rescheduling is a viable option for the client. And for the service provider, even a last-minute cancellation is better than waiting in limbo.
A team from Massachusetts General Hospital published a peer-reviewed study of 927 no-show healthcare appointments. They quantified a spectrum of causes for not showing up. Most would not preclude communicating to cancel or reschedule.
Source: Kaplan-Lewis, Emma, and Sanja Percac-Lima. “No-Show to Primary Care Appointments.” Journal of Primary Care & Community Health 4, no. 4 (July 26, 2013): 251-55. doi:10.1177/2150131913498513.
So why do we still live with this problem? The biggest barrier is convenient two-way communication. Just this week, my friend paid $50 for a missed dental appointment. The issue was neither lack of incentive nor a lack of reminder. Rather, when the conflict became apparent, it was awkward to tie up several minutes going back and forth in a phone call. Voice calls are plummeting, especially for those under 40, and especially for clerical interactions such as calendaring.
Outbound text reminders solve part of the problem and are quickly becoming de rigueur. The reasons are clear: SMS is about as close as you can get to “anytime, anywhere”, thanks to a 98% open rate within the first two minutes of receipt, and sheer ubiquity. SMS is by far the world’s largest social network, crossing geography and economic strata.
But pure outbound isn’t enough. Nor is being able to hit C to confirm; that’s a dead-end interaction except for the happiest of cases. It’s instructive to look at the data from the Sheffield meta-study.
Full resolution requires full interactivity. The client typically wants to resolve the issue right away, whether it’s at 11:00 PM, or while hustling between meetings at work. So does the service provider, so that the timeslot can be filled by someone from a standby list, or otherwise used productively. Everyone wins when everyone knows what will happen next.
|Contact rate due to SMS reminder||97-99%|
|Attendance: effect of voice reminders||5–34% increase|
|Attendance: effect of SMS reminders||same as voice|
|Cancellation or rescheduling: effect of personal voice call||17–26% increase|
|Cancellation or rescheduling: effect of SMS reminder (pure outbound)||insignificant|
A team from universities in Sheffield, UK performed a systematic review of 31 randomized controlled studies worldwide. Among other things, they found in the combined data that SMS reminders did increase attendance about as much as personal voice calls, but did not significantly facilitate cancellation or rescheduling.
Source: Mclean, S., A. Booth, M. Gee, S. Salway, M. Cobb, S. Bhanbhro, and S. Nancarrow. “Appointment reminder systems are effective but not optimal: results of a systematic review and evidence synthesis employing realist principles.” Patient Preference and Adherence 10 (April 4, 2016): 479-99. doi:10.2147/PPA.S93046.
That’s why we at Lumin.ai have built a bot capable of driving appointment-related conversations all the way to an actual calendar change, whether that means confirming, canceling, scheduling, or rescheduling. We’re making it available via RestcommONE for access to both the API and the Marketplace. The RESTful API has been very simple to consume, shielding us from the complexities of traditional telecom integrations.
Lumin.ai is a platform for creating two-way “chatbots” and three-way conversational bots. Its design parameters skew toward some of the most challenging environments in the market: hundreds of thousands of small to medium businesses (SMBs) in a given vertical, in interactions with millions of untrained clients. It provides:
- Natural language processing (NLP) robust enough to handle uninitiated users
- A contextual engine to assemble a picture of end-user needs from fragments of information
- Heterogeneous customization – think one bot with tens of thousands of instances, each executing to a different vendor spec
We’ve made its strengths manifest in the user experience.
Rich communication requires rich NLP. Unlike voice assistants like Alexa or Siri, which accept only rigid commands like “turn on the kitchen lights” or “launch photos”, Lumin’s platform understands a broad range of English constructs so that a client doesn’t have to learn a special syntax. Here are some examples:
- Sorry, I won’t be free on Friday
- How about next week
- I can come in between 10:30 and 3:30 tomorrow
- I’m just not available before noon
- We could totally schedule something for Monday after 3
- Tuesday sometime in the morning doesn’t look too bad
- Friday’s cool
The ability to understand real English isn’t just a concession to technophobes who can’t work a computer, or a curiosity for techno-geeks who always want to be on the bleeding edge. Natural languages have variety because people have a range of needs to express.
Real help requires building context. There’s a big difference between a bot that merely reacts to or drills for details about your latest question, and one that gets to know your needs as you interact. The Lumin.ai appointment bot watches what you say and imply. As you converse with it, it actively builds a model of your preferences and constraints.
Because of this model-building capability, in some situations the bot can actually outperform a human, meeting a client’s needs where a receptionist would run out of patience or ability to follow details. At the end of this post are five sample sessions on video, built around a scenario that is more or less sequential. The fourth one shows the model building pretty clearly. The fifth one shows how a user with a simple need can still do it all in one simple step.
Each vendor has different personality and policies. In addition to different naming, hours, and holidays, each typically has different policies such as cancellation deadlines, identification requirements, and blocks of time to hold open preferentially.
The Lumin.ai appointment bot modifies itself based on a range of vendor configurations controlled via secure API, allowing it to operate successfully in a heterogeneous environment.
Your mileage will vary, but the evidence from multiple verticals and across the world suggests that letting your clients confirm, cancel, schedule, and reschedule appointments anytime and anywhere – not just receive reminders – will dramatically improve your ability to keep your clinic, office, or spa full and producing the value it’s capable of.
What can this mean financially? Here’s a sample calculation. Suppose you’re a dental office with the income and expenses shown in the callout. Suppose your no-show rate drops from 25% to 18% as a result of timely cancellation and rescheduling.
This causes your net collections to increase to more than $1.6M, after accounting for the increase in direct expenses. The extra $109k increases your net income from $500k to $609k. That’s a 22% increase in net income.
|$1.5M in net collections|
|$330k in total direct expenses|
|$140k in administrative expenses|
|$530k in owner/doctor expenses|
|$500k in net income|
Source (round numbers): Hanks, Brian. “5 Simple Tools Dentists Use to Better Understand Their Financial Statements.” November 20, 2015. Accessed January 10, 2018.
View the videos below to see AI-based Appointment Rescheduling in action.
We'd love to discuss your organization’s needs. Together with our partner Telestax, we can deliver an appointment rescheduling solution that’s tailored to your needs and gives you the technology components you need while letting you keep what’s already in use.Talk to us today