Healthcare call center software has been around for a while. Our medical call center services team use AWS Connect Contact Center for all our customers’ centralized call center work. Here’s a guide on how you can build your own healthcare call center software as well (not much coding needed as AWS connect has done the major heavy lifting for you already).
If you are new to building out your medical call center, read this first.
Questions to answer before you use AWS Connect contact center
Here are a list of questions that we recommend you and your team be prepared with, before you move your call center to the cloud using Amazon connect contact center.
Do you need physical phones?
If you do not have a hard set requirement of having physical phones, you can consider AWS connect contact center. If you do have an absolute requirement for physical phones – stop right here, as Amazon connect contact center is not for you.
Do you need voicemail set up for each one of your numbers?
Amazon connect still doesn’t have a voicemail feature out the box (yes, you read that right) as Amazon connect itself is in a nascent state (however, many large customers have moved on to using amazon connect).
If you need voicemail for each phone number, you can still use AWS connect, but you will have to deal with some workarounds. If you are not OK with workarounds, AWS connect is not for you – stop here. As an example, you can have one regular phone that allows voicemails – and you can forward all amazon connect calls that lead to voicemails, to this phone number for the caller to leave a voicemail.
Inbound dialing, outbound dialing or blended dialing using your healthcare call center software?
Are you going to move all your inbound call center agents to use this software or are you going to reserve this for only your outbound call center agents?
In more cases than not, you are probably going to have blended calling – we do the same.
As you decide on this, keep a couple of pointers in mind.
- Adding and removing people/agents from inbound routes in AWS connect is super easy.
- Agents using softphones and a headset can very easily open up multiple browser windows and access various systems as needed for a call from a patient, a referring provider, a pharmacy etc. Productivity gets a very big boost.
- You can add your agents to more queues than one (not just the inbound one)
- You can add agents to website chat and inbound call queues very easily so the same agents can handle both channels (chat and phone calls)
- Outbound dialing is VERY fast using a softphone (compared to manually dialing each number using physical phones).
- Softphone allows you to dial from anywhere instead of being tied to a physical desk phone
- Automated dialing features that AWS connect brings to you elevates the number of dials your agents can make per day.
You do not have to limit agents to only inbound or outbound dialing. In fact, most call centers would do a blend of both. For most healthcare practices, inbound calls are much higher Mondays-Wednesdays each week and slow down a bit on Thursdays/Fridays. However, this also depends on the speciality in question.
E.g. FQHCs and hospitals/health systems are required to be open 24×7 – their call volume is pretty steady.
There’s also seasonal spikes – cold and flu season as an example. What’s your plan to handle those spikes?
How many numbers would you be provisioning into your call center software?
For heavy outbound calling (like we do), you need to have a bank of phone numbers that’s proportionate to the number of agents that dial out and the number of calls they make per day.
A general rule of thumb is that if you have 30 agents dialing 150 outbound calls per day, you need 30*150 = 4,500/100 = 45 direct inward dial numbers for agents to use.
This is because most phone companies frown upon heavy calling and are beginning to mark phone numbers as “SPAM LIKELY” if high call usage is detected. Imagine having to provision a new phone number each time a DID has to be provisioned.
How long would it take you to get that done with your phone company?
AWS connect contact center makes it very easy to purchase numbers. You can do that each day if you want to, or even every hour if you want to.
It’s as simple as buying a number for $1 a month, assigning it to an agent and releasing the previous number that has been marked as SPAM likely by a phone reputation management company.
Avoid being marked as SPAM / SCAM LIKELY
These days, phone companies (the number is not guaranteed) frown upon numbers that make more than 200 outbound calls a day. Like it or not, that’s the reality.
As of Dec 2019, the war on robocalling has increased, helped by the TRACED act. While you might not be making SPAM calls because you have a legitimate reason to call patients, your patients might “inadvertently” mark you as SPAM. It only takes a few patients to mark your number as SPAM for phone companies to start updating your number in their databases as SPAM / SCAM LIKELY.
You need to have a plan to monitor your caller ID reputation these days. There are companies like calleridreputation that help you with this. All you have to do is to register your phone numbers with companies that monitor the reputation scores of your caller IDs (the numbers your agents use to dial out) and monitor all phone numbers for being flagged as SPAM or not (right or wrong).
KPIs for success of your call center
Whether you are handling inbound, outbound or blended calling – what metrics are important to you?
Here are a few examples
- Call abandonment rate (inbound)
- Average time in queue (inbound)
- Service levels (percentage of inbound calls answered within a specific time in seconds)
- Average speed to answer inbound calls
- Handle time (be careful about defining smaller handle time as better – that doesn’t quite work well in good customer service and certainly not in healthcare)
- After call work time (this shows you very good insights into how much manual after call work agents have to do, and thereby increasing the load on inbound queue wait times for callers).
- Customer satisfaction with the call
- Occupancy rate of agents for inbound calls
- First call resolution (although, we have found that sometimes patients call back more than once a day for no fault of the agents)
- Peak hour traffic and longest wait times
- Number of calls made per day per agent
- Number of connects made per day per agent
- Number of conversions per day per agent
- Number of calls per agent per patient
- MOST IMPORTANT OF ALL – CALL QUALITY METRICS
You can define your own set of metrics as well. Just make sure that you do define what success looks like before embarking on this journey.
Shifts and working locations
Our call center team, for example, works on multiple shifts that overlap with each other. You can have a centralized medical call center that observes multiple shifts to support multiple time zones across the country.
Not all your agents need to be co-located and you can have multiple locations across the country (or the world) handling inbound and outbound calls. Some can work on-site, some off-site/remotely.
During this process you are going to need to decide on hours of operations as well (e.g. our inbound team starts at 8:30 AM EST whereas our outbound team starts at 9 AM EST)
Channels to support in your healthcare call center software
Our call center supports voice, email, SMS, chat, facebook, google.
What channels does your call center need to support?
Amazon connect allows you to connect voice, SMS, chat seamlessly through integrations with Amazon Lambda, Amazon pinpoint, Amazon SES etc (more on those below).
You can add multiple channels as your call center matures. At the very minimum, start with the voice channel and progress from there.
Call volume (inbound or outbound) and average call handling time
Before you get started, you are going to need to analyze the past 12 months of call data to figure out the call volume (inbound) that you are going to support.
During this analysis process, please do also account for seasonal call spikes.
It is preferred that you also have a decent idea about the variations (if any) of call volume over days of a week.
All these analyses are going to contribute to how many agents you need to staff your inbound call center team.
You can ask your current IT team or your phone company for call records – that should give you a good handle on call volumes, the trends (weekly vs monthly) and the average call handling times.
Most call centers use an Erlang-C model to figure out the staffing needs. Here’s one example (there are many)
Holiday planning for your healthcare call center
Holiday support is something you need to figure out before you get started. Many practices have emergency numbers and voicemails set up. If you have an onsite and offsite call center, you will need to figure out how you are going to support the holiday schedules of multiple countries. In Aws connect contact center, it is quite easy to set up operating hours – however, you do need to make some changes for routing calls to appropriate places based on the times and days of the calls.
Outbound number that would show up when your call center makes outbound calls
You need to put in a lot of thoughts into this. You would also have to register your CNAM (caller ID Name) for your caller ID.
Let’s say you have 30 agents dialing an average (manually) of 200 each per day. That means you are going to need (as mentioned above) about 45 outbound numbers. You need to provision these 45 numbers, then also register all your phone numbers with CNAM providers.
There are a few service providers that make this process (a bit) easier for you – namely openCNAM by Telo, Neustar etc. Do keep in mind that you can also call your service provider to do this for you but it is hard to do so programmatically with them. With Amazon connect, you would still need to do the same (as of writing, they don’t do it for you).
If you want to learn more about whether CNAM databases actually work or not – read this blog post. While informative, you still should be on the safe side and register your numbers in various CNAM database providers.
What number do you want the customer to see when your agents are dialing out? Let’s say that a called person (patient, provider or whoever else) gets a call from one of your agents, do you want the agent’s phone number to show up or do you want your practice’s phone number to show up on the called person’s phone?
If you are not showing the calling agent’s phone – what’s your current / future plan on callbacks? There are several patients that call back based on a missed call. Should these numbers go to the central inbound number? How would you route those calls to the agent that had made the call?
The inbound agent taking this incoming call would most probably not be aware of the call the outbound agent made – how do they determine who to route the call to (other than shouting across the floor asking who called this person).
Inbound calls into your healthcare call center software
Are you going to have a single inbound number (DID) for all your patients/providers etc? This is based on what your current situation/setup is.
Many times, our customers tend to have a phone number per location – is that what you want to do (vs having a single number for everyone to remember)?
If you are going with a single phone number – are the inbound queries bound by or transferred to specific locations?
What happens when the person your agents are transferring calls to, are not available to pick up the call? Do the callers get an option to leave a voicemail?
Centralize all numbers on Amazon or split with existing phone set up?
Since you are reading this article, you probably do not leverage AWS connect call center or any other cloud based contact center.
If so, and you have a regular phone system setup, would you continue using those numbers as-is? Or would some (or all) of those numbers be ported into Amazon?
Read the note about voicemail above – you could very well decide to keep your existing phone system and reduce the total numbers you have down to 1. Then, all calls that need voicemails could be routed to that one phone number.
You could also have a system design wherein you port all the current numbers from your phone system into amazon connect and purchase a separate number that is set up with extensions to desk phones.
Alternatively, it might make sense for you to leave some DIDs with your regular phone system and transfer the rest to Amazon connect.
Regardless of your decision, Amazon Connect can accommodate all configurations.
Caller Greeting set up in your call center software
In most cases, our healthcare customers have some kind of voice greeting set up. If you have a phone greeting set up – do you want to continue using the same after moving to Amazon connect?
You can easily play pre-recorded audio files as greetings in Amazon connect. Do you have access to the audio files of your current phone greeting?
Amazon Connect gives you the ability to have “Text to speech” greetings – we find this option very powerful. In less than 5 minutes, you will have any greeting you want, spoken in a very professional manner.
To tackle incoming calls during the COVID crisis, most of our customers needed to update their caller greetings – it took us little to no time to make those changes. If we had taken the audio file route, it would have been a lot longer to coordinate, get the recordings done in a professional environment, upload them to Amazon connect etc.
Weigh your pros and cons here. Custom voice recordings might seem great to have, but do you want to give up all the flexibility that Amazon Contact Center’s TTS (text to speech) offers you?
Decide on the call flow for your healthcare call center
Before you get started, make sure you do take the time to document your current call flow. What happens when an inbound call comes in?
- Is there a voice greeting?
- Does the caller make a choice of language?
- Does the caller make a choice of service they need? (e.g. billing, pharmacy, appointments etc)
- Does the caller get an option to leave a voicemail?
- If a voicemail is left – is there a general voice mailbox or what happens?
- Does the caller get an option to be called back based on their position in the queue?
- Does the caller get an option to leave their mobile number to get an SMS back from the agents?
- Does the caller get transferred to another department if they make the wrong choice in their selection of services/language?
- Does the caller get disconnected if all lines are busy?
- Does the caller wait forever if all lines are busy?
- What happens when the caller is connected? Are calls routed by a live person to another person?
- What happens when the caller gets connected and wants to talk to someone specific?
- Is there a dial by extension/company directory option?
- Does the caller have to verify themselves before they are serviced?
- Does your team greet callers by their name if their phone number is already identified?
These are things that you currently already deal with, but not have documented.
Take some time to document all these call flow processes.
Connectivity of various systems needed for agents to do their job
First, you need to understand the various systems that your agents use to be able to service a caller (patient or provider or vendor or pharmacy etc). Typically, this includes multiple systems.
We have found that most inbound, outbound or blended medical call centers are not equipped with tight integration between various systems needed.
Usually an agent needs access to the EMR or EPM, the fax machines/servers, some kind of patient CRM that’s also HIPAA compliant, provider database or CRM, provider directory, internal phone extensions.
While it is always advisable that agents have all access in a single, 360 degree view of the patient kind of portal (patient relationship management portal), it is not always possible to do.
Make sure you have a plan for that. Amazon connect allows you to connect to 3rd party applications via a URL, so it would be highly beneficial if your medical call center software can also connect to these 3rd party software.
Next, what kind of data is gathered on the call? Usually, after call work accounts for a large part of an agent’s day. Agents complain about ”after call work” the most and it is advisable to provide the proper tools and systems to agents so they can reduce the amount of “after call work” they have to do.
E.g. every manager (and quality analyst) would want to know the detailed records of who called, time of call, the conversation length, the nature of the call, how it was disposed etc (see KPIs above). If your medical call center software can solve some of these problems automatically without the need for agents to document each thing manually, you will see huge productivity gains.
We typically advise using a patient CRM for this task. However, we have found that not many healthcare organizations use CRMs and HIPAA compliant CRMs are hard to come by. So, as an alternative, is the data going to be in a worksheet? Think through that.
Keep in mind that while the agents are tending to calls (inbound or outbound), staff at various locations are also using the same systems to tend to patients, carry out billing etc. If your team is gathering data in worksheets, make sure that there is a plan to sync the worksheet (offline) data with the actual systems used to handle daily practice management tasks.
EMRs are traditionally not the fastest software in the world. If you expect your call center agents to be performing at their highest capacity, our recommendation is to not have them bound to your EMR.
EMRs were also not meant for relationship management. These software were meant to be used to store clinical data.
For high performing calling teams, software akin to a CRM is more beneficial.
Campaigns that you will run
Hopefully, you are looking to leverage Amazon connect contact center for leveling up your marketing efforts. What kind of campaigns would you be running?
As an example, we run these kinds of campaigns for our healthcare customers:
- No shows campaigns – our agents call patients that have not showed up for their last appointment. We call and re-appoint these patients. Majority of the patients do not answer their calls, but the majority of the patients that pick up the call also agree to be re-appointed.
- Appointment reminder campaigns – our agents call patients that have not confirmed their appointments via SMS and ensure that they do confirm the appointment via a phone call. In addition to this, we also call patients a day before the appointment to give them a final reminder.
- Patient bump campaigns – We “used to” not call this a campaign because you do not plan to intentionally have patient bumps (i.e. calling patients to reschedule their appointments due to unavoidable circumstances like provider unavailability, inclement weather), but we noticed over time that it truly is a regular campaign every month. We cannot predict when it would be needed, but each month, we have to assign agents to this campaign for at least 1-2 days.
- Reactivation campaigns – We run campaigns for our healthcare customers and reactivate patients that have fallen out of care for more than 6 months. 6 is not a magic number – you can also run campaigns for 3, 9 months (however you want to define your campaigns)
- Well visit campaigns – Most primary care physician groups need patients to come in and have their yearly preventive screenings done. Are you going to be running wellness visit campaigns year long? E.g. For men, Abdominal aortic aneurysm, Blood pressure, Cholesterol, Colon cancer, Depression, Diabetes, Lung cancer preventive care visits are covered. Same goes with women, infants / children/teens and pregnant women (here’s a preventive measure card for reference)
- No encounter campaigns – Most of our medical groups tend to have patients that do not show up for their first appointment itself. These campaigns are slightly different from the other no-show and reactivation campaigns we run. These patients are one step above the community outreach campaigns where we are calling to introduce ourselves to the patients near our customers’ locations (essentially, cold calls). No encounter patients faintly remember our clients’ names and also, sometimes, remember making an appointment. They are not aware of the greatness of the quality of our customers’ brands or doctors but they do recognize the names of our customers’ medical practices when we reach out to them.
- Community outreach campaigns – These are flat out, cold call campaigns. You can purchase people data from various data providers like infoUSA. We separate these patients out in our patient CRM as they are, technically, not patients just yet – so, they do not reside in our customers EMRs. They reside in our customers’ patient CRMs as prospects.. Until they make an appointment with our customer.
- Patient demographic update campaigns – We run these campaigns for our customers as they help with patient retention and also help our billing team maintain the correct patient information at all times. This, in turn, reduces future medical billing errors and also the chances of losing a patient due to change of address or phone number(s). We do not run these campaigns more than twice a year.
- Referring Provider relationship management campaigns – We run this every day. Our firm belief is that you need to keep in close contact with your referral partners. Each week that you are not in touch with your referring partners, you stand a chance of your referring partner being courted by some other specialist. We call our customers’ referring partners on a weekly basis to touch base about the referred patients and discuss issues if any.
- Patient satisfaction survey campaigns – we send patients SMS after their visits for gathering insights into our customers’ practices, getting patient reviews etc. Not all patients respond to these SMS. We have discovered that a simple phone call to nudge patient behavior helps us get better patient reviews (good or bad).
How will those campaigns be running
Are you going to be running campaigns for all outbound? Are you going to be using this in a blended environment? If so, then you need to decide whether agents will receive call backs from patients and if so, where those calls will be routed to – directly to the agent that called or whether it will go to the main number and then be routed to the agent that called.
If you decide on the latter – decide how the inbound agent will determine who had called the patient last.
We do it by identifying the agent “last called by” in our patient CRM – you need to be able to have this information handy as well, else it leads to agent and patient confusion + frustrations.
Are you going to run those campaigns from the CRM or ad-hoc?
This is an important aspect.
If you are running campaigns (which, we hope you are), are you driving those campaigns from your patient CRM? Or are you running them ad-hoc?
You need to be careful about where you drive those campaigns from and consider Amazon connect contact center as a call center tool only. AWS is not going to define campaigns for you, so it is best that you keep all your campaign logic in your CRM and enable click to call capabilities in your CRM through AWS connect contact center.
You can very easily enable pop ups when a know number (patient) calls in – this leverages Amazon connect contact center capabilities and ties into your CRM for patient intelligence.
How many campaigns will be running simultaneously?
Keep in mind that it is not the best practice to have the same patient belong in multiple campaigns at the same time. Amazon connect is only providing you with the call center software in the cloud. It is not maintaining any complex campaign logic for you.
We highly recommend maintaining and controlling your campaigns from your CRM to ensure that the same patient is not in multiple campaigns. We also recommend that you ALWAYS show agents information on last contact date and last person that contacted the patient. That alone solves a lot of embarrassing and awkward situations with patients.
For appointment reminder campaigns, we also recommend that your agents and healthcare call center software handles situations where the patient has two appointment dates within a day or two of each other. Patients don’t look kindly at reminder calls within 2 days of each other. Your healthcare call center software should be intelligent enough to realize that one off those appointment reminder calls do not need to be made.
Agents distribution across campaigns
Sometimes it might make sense for you to have agents dedicated to only one type of campaign. Sometimes it truly is justified as well. As an example, very few agents excel at hard core cold calls. Some are only going to be handling appointment reminder calls.
Most of your agents are going to be somewhere in between. These agents are not the specialists and will be moved around within various campaigns.
For our customers, we have a similar distribution as well.
Keep in mind that changing the routing profile of an agent is really easy in Amazon contact center, however, your call center CRM should be able to handle such changes as well.
We’ve had situations where agents had to be moved from outbound campaigns to inbound ones. Our CRM is able to make that switch and the same is done in Amazon contact center software very easily.
Along with these changes, your call center/CRM analytics /reporting software is also going to need to be able to adapt.
RCM related campaigns
This is a special category in itself.
There are calls made to payers for denials and no responses on claims. These are 1-800 numbers (mostly) and don’t cost you anything but the point to note here is that these calls typically have long wait times, unlike any other outbound calling campaigns.
Meanwhile, balance collection campaigns are to be handled very carefully.
We recommend NOT using your regular outbound number or your main phone number for making balance collection calls. No patient wants to receive balance collections calls, and many times, out of spite, that mark your phone number as SPAM.
This hurts every other campaign INCLUDING your patient reviews SMS campaigns.
We urge you to be careful with the numbers/DIDs you use for these campaigns. We recommend changing these outbound numbers often (every 3-4 weeks).
Automated dialing (not robocall)
Are you going to utilize automated dialing? As you might have noticed, these days most people do not pick up calls from numbers they do not recognize. CNAM databases are not guaranteed to work exactly the same way on the patient’s phone, so even though your caller ID works on some phones, your caller ID might not show up on each patient’s phone.
We’ve noticed that in our medical call center, if the patient picks up the phone, they do agree to being proactive about their health. Our close rates are very high, however, the connect rates are pretty low.
One way to combat this would be to increase the number of dials made per agent per day. If your connect rate is 30% and your conversion rate is 80%, you’re closing 24 out of 100 dials you’re making. So if you’re targeting to close 100 people a day, you need to make 4 times the total number of calls you’re currently making.
This is only possible with automated dialing.
Words of caution. Do not robocall.
It’s annoying, disrespectful and frankly, most laws are shutting them down anyway.
Instead, you can use Amazon APIs to automate the dialing process. This way, you optimize the time it takes agents to copy/paste the patient’s phone number into the Amazon soft phone. The call center software automatically dials patients’ numbers one after the other and if the patient picks up the phone, the agent can speak to them.
If you’re venturing into predictive dialing, that increases your outcomes even further, reduces call wastage, uses complex algorithms to figure out agent availability, call handling times, connect ratios etc to get you higher connects and outcomes.
Do keep in mind that predictive dialing does tire agents out as the number of connects are very high.
At the very minimum, we would recommend using at least autodialer logic.
This is not built into Amazon connect contact center by default, but can easily be created using Amazon’s APIs.
Almost every call center operates on scripts of some kind. I’m sure yours does too.
When an agent logs into our patient CRM, they’re already assigned a campaign. As soon as they login, their AWS connect contact center soft phone launches and they have to login to it.
After that, they can start their daily dialing tasks. At all points, the agents have a 360 degree view of the patient, so when the patient picks up the phone, the agent knows all details of the patient as the details pop up on their computer screen.
In addition to this, based on the campaign they’re assigned to, an appropriate calling script shows up that’s created at run time based on the patient’s data that our CRM already has. This greatly assists the agents in a smoother call flow. Considering that in the call center business agents are always joining/leaving, this also makes the job of training and deploying junior agents a lot easier.
Finally, since agents are switched between campaigns every month, it keeps the agents on track as well.
Do not call / DNC
No matter what you do, you’re going to have patients that ask to be put on do not call lists. What’s your current process of handling DNCs?
Do you mark them as inactive patients? Do you have a patient status in your EMR for the DNC?
As an FYI, we use a status in our customers’ EMRs as DNC. When a patient asks us to not be called, we put them on a DNC list and mark their status in our EMR as DNC.
These patient records are thereafter never in our outbound campaigns and hence are not picked up by our patient CRM for agents to call.
How do you handle DNC numbers that you purchase for community outreach campaigns? You would need to do the same moving forward as well – it’s probably best to scrub these lists before you load the information into your patient CRM.
Dispositions after call
What dispositions are you currently using? We have noticed that our healthcare customers either use too many dispositions (helps with business analytics and reporting) or too few dispositions. Neither approach is correct.
We attempt to tie our dispositions with the EMR’s dispositions. E.g. we tend to use dispositions like below
|SMS/Voice||Call back||Wrong number||Move to RCM team to find phone number from payer portal|
|SMS/Voice||Cancelled||No show||Move to noshows team|
|No Authorization||Move to RCM team|
|Patient rescheduled||Move to reminders team|
|Office rescheduled||Move to reminders team|
|Tardiness||Move to reminders team as front desk will give them another appt for being late|
|Patient left||Move to reactivation team|
|Not needed||Move to reactivation team|
|No reason given||Move to reactivation team|
|Called to cancel||Move to reactivation team|
|Other||Case by case basis|
|Do not call||Mark pt as inactive in CC|
|Deceased||Mark pt as deceased in CC|
|Changed provider||Mark pt as inactive in CC|
|Call back||Wrong number||Move to RCM team to find phone number from payer portal|
|Lost||Do not call||Mark pt as inactive in CC|
|Deceased||Mark pt as deceased in CC|
|Changed provider||Mark pt as inactive in CC|
|Inbound||SMS/Voice/Chat||Appointed||New patient||Fax||Notify acct mgmt team|
|Werq||Notify acct mgmt team|
|Par8o||Notify acct mgmt team|
|Referwell||Notify acct mgmt team|
|Provider office call||Notify acct mgmt team|
|Postcard||Notify marketing team|
|Notify marketing team|
|Patient referral||Notify marketing team|
|Zocdoc||Notify marketing team|
|No reason given||Move to reactivation team|
|Called to cancel||Move to reactivation team|
|Confirmed||Pt called to confirm|
|Transfer||Language barrier||Transfer to Spanish agents|
|Pharmacy||Send email to JAMTechs or BXtechs or JHTechs|
|Medical records||Notify sarai, keesh, saurabh|
|Other||Pt complaint||Notify Debra|
|Inquiry||Case by case basis|
|Collections||SMS/Voice||Call back||SMS sent|
|Wrong number||Move to RCM team to find phone number from payer portal|
|Lost||Do not call||Send to collections company|
|Deceased||Mark pt as deceased in CC|
|Refused to pay||Send to collections company|
|Collected||Settlement in full||Move to RCM team for payment posting|
|Payment in full||Move to RCM team for payment posting|
|Promise to pay||Move to RCM team to keep track of|
|Partial payment arrangement||Move to RCM team to keep track of|
HIPAA compliant call recordings
We hope that you use call recordings in your call center for audit and training purposes. You can continue doing so with Amazon Connect as well. Strictly speaking, it is not a feature of Amazon connect, however, the software connects with Amazon’s S3 solution – which in itself is dirty cheap and you can store all your call recordings in Amazon S3, in a fully HIPAA compliant manner.
Amazon Connect links to each call’s recording in the call record itself so it is quite easy to listen to each patient call as well.
At a bare minimum, you should have your collections call recordings available. This is the most litigious area and even though you might be calling as a first party collector, you do need to keep the call recordings in place.
Do note each state’s laws regarding notifying the called person about the call being recorded vs not.
We also recommend that you do NOT allow agents the ability to control live call recordings (i.e to be able to turn them off on demand).
Called person call backs
How do you handle customer call backs? As mentioned above, this is something that you need to think through. It is very easy to have an IVR set up with Amazon connect. You can assign each agent an “extension”.
However, you also need to think through the process you currently have. Are your agents leaving voicemails? If they are leaving voicemails, are they leaving their name and extension?
If the agents are not leaving voicemails, this becomes harder to do (callbacks) if you have your outbound number show up as the company main number. On the other hand, it is easier to handle if each person calling has their own DID number show up on the patient’s missed call log. This way, when and if the patient does call back, they reach the agent that called them.
Do keep in mind that call center is a high volume game, so it is always better to have the Amazon screen pop show the patient that’s calling back.
Or, at the very minimum, if your CRM has an easy way to search by patient phone number, that makes agents’ lives easier as well.
This part is outside the purview of Amazon Connect contact center – however, the screen pop can be very easily created with Amazon Connect.
When you are deciding on call backs, you also need to determine whether you are going to send the callbacks directly to voicemail or not. This leads into the next discussion of the Amazon queues. You can also choose to have the customer be on hold, waiting in queue.
Call center queues
Amazon call center software, much like any other healthcare call center software has the concept of queues. These can be inbound or outbound.
Think of queues as “tasks to be done” and agents as ones picking tasks from the queue to get them done.
Monitoring queues regularly and defining the maximum number of calls allowed to be enqueued is an important part of running a call center.
For our customers’ call centers, we manage an 80-20 rule, where 80% of the calls are to be answered within 20 seconds or less.
The rest of the callers, get put in a queue.
You can decide that you will only allow 10 people in the queue and the rest of the callers will go straight to voicemail – however, your team does need to monitor the queue daily to see how many calls are going to voicemail and when that’s happening.
This gives you invaluable insights into whether you are understaffed or overstaffed at your call center. If there’s a trend in the number of calls getting queued or going to voicemail during particular times of the day, you could also adjust your inbound calling staffing accordingly.
You can also decide that instead of having an overflow of calls go to VM, they are put in a callback queue in Amazon connect. This allows you to have a system and process in place to handle callback queues.
However, you do need to decide which set of agents handle those callbacks. For our customers’ call centers, we have a small portion of the inbound team call back on our callback queue contacts.
Answering machine detection
Most healthcare call center agents do not leave a voicemail. We have vacillated between leaving a voicemail vs not. There has been no solid proof / data points thus far in our operation that would allow us to make a data based decision.
Whether you leave a voicemail or not is ultimately going to be your team’s decision.
Detecting answering machines on the other hand is a very important part of outbound calling. If you are leveraging power dialer, autodialer, progressive dialer using Amazon APIs, then you do need to undertake this answering machine detection headache.
So far, we haven’t found any solution available in the market that can detect answering machines correctly 100% of the time.
Your custom reporting or AWS reporting?
There are several KPIs that we measure, hence our reporting is custom developed. We leverage Tableau for reports/charts etc. You do not necessarily have to use Tableau and instead can use some of the AWS reports as well.
AWS connect provides quite a few reports out of the box itself. However, do keep in mind that AWS limits the amount of data they provide in each report that can be exported from the Amazon connect dashboard.
Amazon encourages you to tie your reports to Kinesis and store a large volume of data. All you need to do is to get those reports and use SNS as a way to email yourself the reports.
However, if you place a large volume of calls and want the ability to slice and dice data, you are going to need something like Tableau, Looker etc.
We have consolidated reports that we provide management vs more granular reports that are used by the quality analysts.
Here’s a sample of such reports (note, this is from the COVID time period, hence the numbers are horrendously low).
|RCM Data for 3/23/2020|
|RCM Data||Total Count||Total Revenue|
|Prior auths needed||0||$0.00|
|Prior auths obtained||0||$0.00|
|Remittance recd for prior claims||0||$0.00|
|OTC Collected yesterday||1||$0.00|
|Copay forecast for today (Cancelled Appointments)||22||$655.00|
|Appts requiring R/S due to eligibility/auth/referral issues||0||$0.00|
|Pt Balance collected (Not working in this campaign)||0||$0.00|
|RCM Data||Total Count||Reprocessing Claim Amount|
|Denials recd for prior claims||59||$28,594.98|
|Charge re-posting for denials||0||$0.00|
|No responses – claims past due (30 days)||0||$0.00|
|Denials called on for RCM team||65||$12,777.86|
Inbound call center tasks
|Call Center for 3/23/2020|
|Call center Data||Total Count|
|Resolved on first contact||69|
|No shows reappointed||0|
|Recall pts reappointed||0|
|Outreach new pts booked||0|
|Appt requests recd (Inbound)||6|
|Appt requests resolved||6|
|Referral partners connects||25|
|Appts confirmed by agents||0|
|Appts self confirmed by pt||0|
|Activated Patient Portal for today||5|
|Unable to activate due to email ID for today||0|
Inbound appointment request sources
|Appt request sources for 03/23/2020|
|Appt request sources||Received|
|Google my business||0|
|Inbound abandoned call||4|
|Inbound Not Resolved call||1|
|Inbound calls from providers||6|
|Inbound calls from patients||76|
More detailed reports (summary versions look something like this)
|No Shows – Jan to 20th Mar 2020|
|Month||Total Appt||Total Billed Appt||Ratio|
|Appointment Reminder – Sept’19 to Dec’19|
|Month||Total Appt Reminders||Confirmed||Billed||Ratio|
|Community Outreach – Jul’19 to Dec’19|
|Month||Total Appt||Total Billed Appt||Ratio|
Breaking down the referrals received with new appointment requests
|New Referrals – Jan’19 to Dec’19|
|Month||Referrals Received||Referrals Handled||Balance|
The various dispositions of appointment requests
|Outcome / Status||Total|
|Back To Sender||387|
|To Follow Up||355|
|Cancelled By Patient||87|
|Cancelled By Sender||3|
|Reactivation – Jun’19 to Dec’19|
|Month||Total Appt||Total Billed Appt||Ratio|
Appointment Reminder – Mar’19 to Aug’19 (we break it down by month)
|Appointment Reminder – Mar’19 to Aug’19|
|Month||Total Appt Reminders||Confirmed||Billed||Ratio|
No Encounter campaign results – Jul’19 to Aug’19
|No Encounter – Jul’19 to Aug’19|
Or, more detailed reports like
|Inbound Appointment request – Dec-20|
|Source||Week||Request||Scheduled||Billed||Billed Amount||Estimated Revenue||Variation||Billed Vs Scheduled Ratio||AVG FTE|
|Inbound Call||I Wk||230||230||131||56.96%||4|
|Inbound – Website & Email – Jan’20|
|Source||Month||Request||Resolved||Billed||Billed Amount||Estimated Revenue||Variation||Billed Vs Resolved Ratio||AVG FTE|
|Website/ Email||I Wk||45||19||7||36.84%||1|
|Appointment Reminder – Dec-99|
|Week||Total Appt Reminders||Confirmed||Billed||Billed Amount||Estimated Revenue||Variation||Ratio||AVG FTE|
|New Physician Referrals- Jan’20|
|Week||Referrals Received||Referrals Handled||Balance||Ratio||AVG FTE|
Week by week inbound campaign results
|Inbound Campaign – Jan 20|
|Week||Calls Received||Calls Handled||Calls Dropped||Calls handled Ratio||1Day Repeat calls||1 Day Repeat Ratio||1 Day Unique Calls||AVG FTE|
Appointments booked vs billed broken down by the week
|Week||Total Appt||Total Billed Appt||Ratio||AVG FTE|
|Patient Referrals – Jan’20|
|Week||Total Patient Referrals||Total Patient Referrals Billed||Ratio|
Patient reactivation reports
|Reactivation Report – Mar 20|
|Monthwise Reactivation Report – Mar 20|
|Month||Total Reactivation Assigned Call||Total Patients Dialled||DNC||Connected||Total Done||Total Attempted Calls||1st Attempt Calls||1st Attempt (Done)||2nd Attempt Calls||2nd Attempt (Done)||3rd Attempt Calls||3rd Attempt (Done)||Calls Attempted Conversion Ratio||Rescheduled Conversion Ratio|
|Month||Mar-20||<- Change the month to get the monthly report.|
|Weekwise Reactivation Report – Mar-20|
|Week||Total Reactivation Call||Total Patients Dialled||DNC||Connected||Total Done||Total Attempted Calls||1st Attempt Calls||1st Attempt (Done)||2nd Attempt Calls||2nd Attempt (Done)||3rd Attempt Calls||3rd Attempt (Done)||Calls Attempted Conversion Ratio||Rescheduled Conversion Ratio|
|Agentwise Reactivation Report – Mar-20|
|Agent Name||Total Reactivation Call||Total Patients Dialled||DNC||Connected||Total Done||Total Attempted Calls||1st Attempt Calls||1st Attempt (Done)||2nd Attempt Calls||2nd Attempt (Done)||3rd Attempt Calls||3rd Attempt (Done)||Calls Attempted Conversion Ratio||Rescheduled Conversion Ratio|
|Call Datewise Reactivation Report – Mar-20|
|Appt Date||Total Reactivation Call||Total Patients Dialled||DNC||Connected||Total Done||Total Attempted Calls||1st Attempt Calls||1st Attempt (Done)||2nd Attempt Calls||2nd Attempt (Done)||3rd Attempt Calls||3rd Attempt (Done)||Calls Attempted Conversion Ratio||Rescheduled Conversion Ratio|
Week wise Performance for various campaigns
|Week wise Performance|
|Week||Audit Calls||Error||Error %||Perf %||QC Call||Score||Call Score %|
|Campaign Name||Audit Calls||Error||Error %||Perf %||QC Call||Score||Call Score %|
Day wise Performance for various campaigns
|DAY||Audit Calls||Error||Error %||Perf %||QC Call||Call Score %||DAY||Audit Calls||Error||Error %||Perf %||QC Call||Call Score %|
Our Tableau reports look like these
SMS, Email, Website chat, Faxes, Google, Facebook channels on our healthcare call center software
Our inbound team handles multiple channels for our customers. We are sure yours does as well (or a mix of it).
All our phone calls go into our Amazon contact center. Some of our customers use a hybrid/bridge system where some of the calls go to their desk phones / other phone numbers and the remainder come to Amazon connect contact center.
Our customers’ emails come from various sources
- Fax triggered emails – the faxes are not stored on the fax servers, so we typically move our customers’ faxes to a HIPAA compliant fax server provider like mfax or Aculab. Note that Amazon doesn’t have a native/out of the box fax server solution yet, however, you can use one of their partner provided fax servers.
- Referral website emails – our customers receive emails about inbound referrals from referwell, par8o, p2p (jointhenetwork) etc. These inbound emails simply notify our inbound agents to check the websites for referrals.
- Authorization websites like covermymeds
- Collections emails from stripe. We usually set up our customers for patient collections using stripe.com payment processor, so each time a patient pays their balance due or has a recurring plan set up and the payment is posted, our inbound team gets notified via email.
All faxes go to either our customers’ fax server, their physical fax machines or are emailed to their practices from the fax servers. In all these cases, we move the fax servers to a HIPAA compliant fax server (as mentioned above).
We also support SMS channels. Typically, our customers tend to not have an SMS enabled practice but we help them enable their practice to be able to receive SMS from their patients. This improves patient engagement multi-fold and helps reduce the task load of incoming and outgoing phone calls our call center agents have to make. This, in turn, reduces the overhead costs for our customers.
Most of our customers do not have website chat enabled and tend to lose a lot of new patient appointments by not having a website chat. During our transition period with our customers, we enable website chat on their websites (and create their websites, if they do not have one).
Google My Business channel
Once we transition business to our contact center (call center), we ensure that our customers are connected to Google My Business and also enable the Book Now + the messaging capabilities that GoogleMyBusiness allows. Google My Business does not currently have a publicly available API (time of writing), so we handle this channel manually (assigning a person to it).
We handle Facebook Book Now and messaging channels manually as well. We do this because this way, we can assign the social channels to specific agents.
Amazon connect contact center can easily support chat and phone channels out of the box, right from the Amazon connect dashboard itself. Amazon connect allows you to assign agents (inbound) to both web and chat channels. You can decide how many chats are going to be allowed per agent. Amazon connect also takes care of the fact that if an agent is currently answering an incoming phone call, they should not be assigned an incoming chat.
It is not very difficult to integrate Amazon SES (simple email service) to Amazon connect either (although, there’s no real direct connection to it) Amazon’s pinpoint can be used to enable SMS sending capabilities – as of writing Pinpoint does not allow 2 way SMS conversations.
However, we do not use Amazon for these SMS and Email channels. We have our own patient engagement software that we use to have conversations with our customers’ patients. We use Twilio for these purposes – however, you can also use the industry giants like Bandwidth for this.
For the email and faxes channels, we use a custom task manager application developed by us. This was developed using an opensource application (Nodered) since we cannot use our other favorite Zapier due to lack of HIPAA support. This helps us manage all our incoming faxes, emails – NodeRed simply listens for email and fax events, presents these as “new tasks” that our inbound agents have to handle.
Satisfaction surveys at the end of each call
Hopefully one of your metrics includes caller satisfaction. Amazon makes is super easy to call an AWS Lambda function and send a satisfaction survey SMS to the person that called. Of course, many calls will be from landlines and these SMS messages would fail. However, when the call is from a mobile, the caller (patient, vendor, another practice) will get an SMS asking them to rate their satisfaction with the call.
We have found this to be an invaluable tool to keep tabs on our agents’ performance and to identify issue areas in our practice (e.g. patients not being satisfied with the call if their medical records take long to pull up)
There are many more factors to consider – we will keep adding those over time. Hopefully these have helped you create or launch your own cloud based healthcare call center software leveraging Amazon Connect Contact Center.