Services We Automate

Purpose

To allow patients and physicians to check on the status of recent orders while remaining HIPPA compliant.

Background

There are two types of order status inquiries available for our Life Science / Med Device clients:

Before automation can help with Order Status, the caller must first go through a HIPAA compliant authentication process.

Most order status calls can be handled through automation which has a significant impact on overall call volume as well as other associated benefits.

Our Approach:

Once the account has been properly authenticated, the next challenge is to find the correct order and the correct items on that order and provide the appropriate status. Very often, the order is “in process” and a lot of order status IVR applications provide just this general response. The problem with that is that it tends to cause the call to go to a live agent anyway. Our objective is to work with our clients to provide the same information that a live agent would, even if that means using a lot of AI to give the most complete answer.

With the use of Natural Language, we can structure the conversation in such a way that only the information the caller is after is provided and simple complications can be addressed without the aid of a CSR.

We tailor the above approach to meet our client’s specific operational requirements. Fundamental to our application design is the question, “How do your agents handle that situation?” Given the sophistication of IVA®, we can almost always determine a way to emulate the same service delivered by live agents.

Insight

The following best practices are recommended for the design of Order Status applications through voice self-service:

• Use generalities when describing an order and only go through a line item-by-line item inventory if the caller asks. “Your order of 5 items which includes x and y is still being processed and is expected to ship from our warehouse on z”.
• Be as specific as possible on the actual status portion as a general response of “it’s still processing” will almost certainly result in the call going to the call center anyway.
• Integrate with third party shipper systems to provide the delivery information as reported directly from shipper.
• If orders were shipped separately, indicate the information for each shipment.
• Offer to provide a text or email update when the item or order has finally shipped.
• Provide all information gathered, to the call center representative via CTI or WebCTI in order to avoid asking the caller the same information the system just did and to improve agent handle time and caller experience.
• Allow callers to opt out easily so that they do not get frustrated. It helps you maintain a good customer experience.
• Be only slightly persistent. Even when callers opt for a live agent, collect the purpose of the call from the patient in order to help route the call and provide better WebCTI data for the agent.

See here for an overview of generic IVR best practices.

Performance Statistics

After deploying the Life Sciences Order Status solution, you should experience performance characteristics similar to those experienced by our existing clients.

• Since IVA® is all natural language, it is rare for a caller to end up in Order Status by accident. Basically, every single caller that asks for it will stick around to actually hear the status – 99% or better.
• Complete Order Status: Of those that have orders, IVA® provides an up-to-date order status to about 88% of them. However, only around 52% of callers that request their order status actually have orders, a unique challenge in this industry.
• Transferred calls: The other half (48%) of those callers often receive a shipment on a regular recurring basis, so they assume the order has shipped when, for some reason, it has not. Maybe it is set to ship sometime in the near future, their prescription has expired and they need a new one, or they need additional paperwork/authorizations before the order is able to ship. This percentage can vary widely depending on the data available to IVA® and the specific business rule preferences of the client.

Relevant Unique Capabilities of IVA®:

Artificial Intelligence allows systems such as IVA® to process information in a very dynamic and complex way, automating the most sophisticated thought processes of a live agent. As it relates to Order Status Calls, IVA® has several special capabilities that might not even be noticed, because they just work.

• The system allows for multiple ways to identify an order and items specifically on that order.
• The system allows callers to say “hold on” or “could you repeat that” or other similar expressions. Given the typical demographic of callers, this allows them to locate the required information.
• Given the number of different products available, the system has special handling to use the wording that the customer is familiar with instead of the technical or part-number / part-description which might be confusing.
• By default, our system is able to integrate with most shipper tracking systems including UPS and FedEx.

Purpose

To properly identify callers for security purposes, prior to allowing any further service.

Background

For healthcare in general, before most services can be rendered, the patient must be properly identified. This is due to HIPAA regulations, of course. This task–more or less the same for each patient or provider that calls—is perfect for automation. But it still proves difficult because, unlike many other industries, multiple pieces of information must be validated and the key data (i.e. Patient ID) are not usually top-of-mind for most callers.
In addition, the healthcare needs of older people are generally greater, meaning a voice interface must be well structured, methodical, and understandable to older patients who may not be used to interacting with technology, have a hard time hearing, or are slower in responding.
Importantly, if you’re able to automate this process, it has the ability to make a significant reduction in agent handle time.

Our Approach:

SmartAction’s Life Sciences team has spent considerable time over the past several years refining and perfecting our approach to patient identification and authentication. We’ve found that the two main challenges to automated authentication are adhering to government regulations (federal and state) and determining the personal information that callers have most readily available.

Government Regulations

Many laws and regulations govern how, when, what, and to whom protected health information is released. While the HIPAA rules govern federal requirements, confusion can occur when state laws go further and vary, such as with HIV or genetic information. So here’s how our clients guarantee that they meet all regulatory standards. Generally speaking, they keep between six and eight pieces of patient-specific data that are available for authentication purposes. This data might include:

• patient’s legal name,
• date of birth,
• sex,
• address,
• telephone number,
• guarantor,
• subscriber,
• next of kin,
• Social Security Number, or
• unique identifier such as patient ID.

• Of this list, most clients require callers to accurately provide three of the data elements in order to positively identify them.
  Which information when?
  The real complication ensues when an automated system tries to figure out which three pieces of information the caller actually has available. To alleviate this, we advise clients to start with the data elements that are most specific to the patient, have similar formats, and/or are most likely to be known. Naturally, we start by asking for either the phone number on the account or the patient ID itself. These numbers come from different sources and are validated in different ways, but we have discovered that they provide the highest probability for narrowing down who we are talking to. If they don’t have this information, we move on to other pieces of data.
  We tailor the above approach to meet our client’s specific operational requirements. Fundamental to our application design is the question, “How do your agents handle that situation?” Given the sophistication of IVA®, we can almost always determine a way to emulate the same service delivered by live agents.
  Of course, in order for this approach to work properly, we work with our clients to design and develop ways to pull these data elements from the right places and push updates when needed.

   

Insight

The following best practices are recommended for the design of authentication processes through voice self-service:

Ask for one of two data elements at the same time; it increases the odds that the data being requested is actually something the patient knows. For example, ask for patient phone number OR Patient ID all in one prompt and make sure the system can tell them apart. Hear an example of this here.

Make it clear to providers and health plans that they can use the system too. Of course, they need the appropriate patient data and must be reasonably authorized to have the patient health information, as well.

If you’re planning to start a contact center transformation project, account authentication is a great call type to start with. You’ll then be able to easily evolve and iterate from there, providing self-service and other routing capabilities next. Use patient information, collected during authentication, to further segment and personalize the caller experience.

Provide all information gathered, whether authenticated or not, to the contact center representative via CTI (computer telephony integration) to improve agent handle time and caller experience.

Keep track of and report on caller behavior, such as repeat numbers (i.e. people that call multiple times in a row) and early hang-ups. They help to tell the true story about how the solution is performing and provide insight for making it better over the long term.

Allow callers to opt out easily so that they do not get frustrated. It helps you maintain a good customer experience.

Be only slightly persistent. Even when callers opt for a live agent, collect the purpose of the call from the patient in order to help route the call and provide better WebCTI data for the agent.

Make sure that your agents authenticate callers the same way the automation does to discourage callers from “shopping” for a non-conforming agent.

See here for an overview of generic IVR best practices.

Performance Statistics

After deploying the Life Sciences Authentication solution, you should experience performance characteristics similar to those experienced by our existing clients.

• High Engagement Rates: On average, callers engage with IVA® 96% of the time. Those that do not engage are often seeking information that is not related to authentication–such as the company’s mailing address, health plan phone numbers, or fax number.
• Full Authentications: Of the 96% of engaged callers, callers fully authenticate about 70% of the time. Those callers that do not fully authenticate often have incorrect information and struggle to authenticate even with a live agent.
• Decrease in AHT: The authentication process through IVA® takes approximately 85 seconds, compared to live agent authentication which is often longer. So every call authenticated by IVA® saves the call center at least 85 seconds, if not more.
• Decrease in Caller Effort: Since the handle time through IVA® is only 85 seconds, the caller also has a better customer experience and ease of use.
• ROI: Simply automating this small process within most calls, the cost savings over a live agent were over 80%.
• By automating authentication, our clients can then automate other processes, for use by callers anytime, most importantly during off-hours or crunch times.

Reduction in call volume and flattening out of spikes allows the contact center to have more predictable workforce staffing and adherence. The point is that you cannot have full self-service automation without first doing authentication.

Relevant Unique Capabilities of IVA®:

Artificial Intelligence allows IVA® to process information in a very dynamic and complex way, automating the most sophisticated thought processes of a live agent. As it relates to patient authentication, IVA® has several special capabilities that might not even be noticed, because they just work.

• The system allows for different inputs at the same prompt. (i.e. “Please tell me the patient number or the phone number.)
• The system allows callers to say, “Hold on,” or, “Could you repeat that?” or other similar expressions. Given the typical demographic of callers, this allows them to locate the required information.
• Voice self-service often struggles with name matching – matching what is said to a name in the database. But IVA® has exceptional natural language processing capabilities, so it is able to confidently match names and appropriately accept formal versus informal versions of names.
• The system allows for spelling phonetically for specific inputs, such as name if necessary. (i.e. “B as in Bravo, O as in Oscar, B as in Bravo”).
• IVA® does not need to confirm each response that a caller gives. Only if there is reasonable doubt that IVA® misheard will the system confirm a response.

Purpose

To allow patients and caregivers to review and update an address, phone number, or doctor’s information on file with Life Sciences companies while remaining HIPAA compliant.

Background

Patients and caregivers may call Life Science companies that provide products or services when they change their billing address, physical address, move into a healthcare facility, change contact phone number, change doctor name, or otherwise want to confirm that the information on record is correct.
Before automation can help with these changes, the caller must first go through a HIPAA compliant authentication process.
Address changes are typically straightforward but also tend to be fairly long, low value calls. The same can be said for phone number and doctor changes. Having said that however, these changes, if not recorded correctly and in a timely manner, can cause all sorts of process issues within your organization. Shipping to the wrong address and getting returned goods, not being able to contact the patient by phone to confirm information, and being unable to get reliable authorizations from doctors and insurance companies to deliver products in a timely manner are just some of the very good reasons to make these changes easy for patients and other customers.
Additionally, because the calls are long, significant Return on Investment (ROI) can be had with only a small number of such calls a day.
However, change of address and phone number can actually be more accurate through automation than with a live agent because the process is highly tuned based on best practices. Callers tend to pay closer attention to the system than a live agent when it confirms changes because the confidence level is inherently different.

Our Approach:

Once the account has been properly authenticated, the next challenge is to determine exactly what the caller would like to change and to make it clear that more than just address can be updated through the system. This clear messaging helps to retain callers who might otherwise think that the system is not sophisticated enough to handle all of their changes.

Address

Changing address seems simple, but from a business requirements perspective, it is not necessarily straight-forward. For ordering and re-ordering supplies, the address change is usually related to shipping address. When the system captures a new shipping address, we also make sure that any changes to billing address are updated, as well.

Capturing the actual address can be difficult because of variability of address information, in general. We use a multi-step process which makes this data collection more effective.

• Ask for the zip code of the new address first.
• Ask for the house number and street address next.
• With this information, we are able to use a combination of tricks and databases to validate the information. If IVA® has low confidence that it captured the address correctly, it makes an educated guess before structuring the validation prompt.
• The system repeats back the first line of the address and provides the city and state (based on the zip code lookup); then IVA® asks the caller to confirm that it is correct.

• If it is incorrect, the system will go through each element and confirm each piece until it is correct.
• Finally, if there is a unit number or a second line of the address, we capture that last and validate it on its own, as it is often not dependent on the other elements of the address to be valid.

Once the new address has been captured, it is also important to run that new address through a validation process. Just because the caller confirms that the address is correct does not mean it is; sometimes callers don’t listen when the system is confirming, and therefore an errant address happens from time to time. If so, it will need to be re-reviewed by the caller and the system, this time with the caller listening more closely, hopefully.

Phone Number:

Capturing a new phone number is very straight-forward. The most important aspects of this demographic change are:

• Caller education: callers know that they can use self-service to make the change and don’t request transfer to the contact center before the system even has a chance.
• Intelligence: the system uses its intelligence to proactively ask if the phone number needs to be updated when performing other demographic changes (like address).

Physician:

The final demographic change typically made is changing the physician assignment for a given patient. There are few aspects to this change:

• The change is not necessarily related to an address or phone number change and therefore the system needs to be flexible enough to recognize exactly what the caller is wanting to change without a lot of interrogation.
• Physician names can be extremely difficult to capture via an IVR, especially if the doctor is part of a large practice, but there are some tricks to help improve performance. In order to make this effective, a comprehensive list of doctors is needed; often, this list can be derived from all the doctors of current customers/patients. Other solutions are available as well.
• Capture the phone number or zip code of the doctor’s office to narrow down the list of potential physicians.
• If the list of physicians at that number or zip code is small (three or less), provide those names to the caller and let the caller choose.
• If the list of physicians at that number or zip code is medium in size, then narrow down what the system should be listening for. This improves the speech recognition rate significantly.
• If the list of physicians at that number or zip code is very large, then the system will ask for other limiting information, such as the first letter of the physician’s last name.
• Finally, just because the system was able to identify the proper physician, it does not mean that that physician is eligible to be updated due to practice type, non-acceptance of insurance, and other limiting factors.

Finding and updating the physician information on a patient record can be very time consuming for the contact center. It is vital to get this information correct however, since many steps in the overall business are dependent on this accuracy.

Benefits

With the use of Natural Language, we can structure the conversation in such a way that only the information the caller is after is provided and simple complications can be addressed without the aid of a CSR.

We tailor the above approach to meet our clients’ specific operational requirements. Fundamental to our application design is the question, “How do your agents handle that situation?” Given the sophistication of IVA®, we can almost always determine a way to emulate the same service delivered by live agents.

See here for more information about how conversation flows are developed and used as part of our design process.

Insight/Best Practices

The following best practices are recommended for the design of Demographic change applications through voice self-service:

• Make it clear that the system can handle a wide array of changes. Generally, we like to use open ended prompts that ask the caller what they want to update, followed by an acknowledgement of what they asked for.
• Whenever possible, emulate the processes and prompts used by the best CSRs.
• Use logic and data where possible to look for discrepancies, just as your best agents would. If a phone number change is occurring, is it changing to the same area code as either the shipping or billing address? How about the physician: is it generally in the same area as the shipping address? If not, simply ask the caller if this discrepancy is actually correct or not. Be proactive.
• Allow callers to opt out easily so that they do not get frustrated. It helps you maintain a good customer experience.
• Pass all data collected on to the agent in cases where the call does not finish in the system. That way, the caller does not have to repeat any information that was successfully collected.
• Be only slightly persistent. Even when callers opt for a live agent, collect the purpose of the call from the patient in order to help route the call and provide better CTI data for the agent.

See here for an overview of generic IVR best practices.

Performance Statistics

After deploying the Life Sciences Demographic Change solution, you should experience performance characteristics similar to those experienced by our existing clients.

Agent Average Handle Time: Generally, at the outset, the number of callers that need to update demographic information is only about 15%. But this process is time-consuming. With IVA®, handle times should decrease by 10-20% and, more importantly, they will be cheaper automated minutes instead of expensive agent minutes.

Call Deflection: Despite the relatively small number of callers that make demographic changes, automating this task is all about deflecting calls from the call center. Typically, IVA® can successfully assist about 2/3 of those that self-serve in demographic changes. Phone number change is usually the easiest to deflect, while address change is the most challenging. Over time, as callers become accustomed to the system, business rules are more clearly defined, and CSRs are trained to educate callers about the self-service option, self-service successes naturally increase.

Confirmations: In many cases, successfully assisting a caller requires no more than verifying that we have the correct information on file. More than half of callers that use the self-service option for demographic information actually do not make any changes; they are simply confirming that what we have is correct.

Relevant Unique Capabilities of IVA®:

Artificial Intelligence allows systems such as IVA® to process information in a very dynamic and complex way, automating the most sophisticated thought processes of a live agent. As it relates to Order Status Calls, IVA® has several special capabilities that might not even be noticed, because they just work.

• The system allows for changing various patient demographic data without marching the caller though a laborious menu tree.
• The system allows callers to say, “Hold on,” or, “Could you repeat that?” or other similar expressions. Given the typical demographic of callers, this allows them to locate the required information.
• The demographic change capabilities are tightly integrated into the overall system of capabilities which means that address changes, for example, related to Assignment of Benefits (AOB) or other documents can be for a one-time purpose or related to actual demographic changes.
• IVA® uses a very sophisticated spelling algorithm that allows for both phonetic spelling (A as in Alpha, B as in Beta) as well as just straight name spelling. Using dynamic data, such as available doctor names within a zip code, helps boost the success of changes.