Our journey with Amazon Connect Contact Center

Why Amazon Connect Contact Center? Here’s our journey into the patient contact center line of business, why we chose Amazon Connect and how we became Amazon Connect Contact Center partners. We talk about how we got into the patient contact center business, why we didn’t go with other established contact center software, how we scaled our patient contact center team over time.

Why Amazon Connect Contact Center? Here’s our journey into the patient contact center line of business, why we chose Amazon Connect and how we became Amazon Connect Contact Center partners. After running for months.. We found out that our actual cost per agent per month, along with dialing charges were around $70/month.

How we started our patient contact center

We didn’t start our patient contact center with Amazon Connect right from the start. Heck, our patient contact center didn’t even start as a patient contact center to begin with. While working with one of our healthcare customers, we noticed that their EMR was filled with patients that had not been seen (recalled) in more than a quarter. Considering our experience with regimented outreach, we felt that things could be a lot better.

Starting with reactivation of patients not seen in more than a quarter

We decided to just pick up the phone and start dialing patients that had not confirmed their appointments. We started with Skype 🙂 We didn’t quite call ourselves a patient contact center back then. We were just making sure that our client’s patients showed up. The experiment was a huge success. We added 5,200 new patient appointments in 2 months with 3-6 agents calling each day. We grew our team from an initial size of 6 people.

Adding appointment reminders

We also started noticing that our customer’s no-show rates were consistently high. They were relying entirely on their CareCloud EHR to send patients text messages for confirmations. This customer’s team was not really calling patients to remind them of their appointments. Not because they didn’t care to, but because they were simply overwhelmed. Considering that it was an ophthalmology practice with mainly older patients, we were pretty sure that text messaging alone was not going to cut it. 

Challenges with Skype

While we did operate on Skype for a month, it was becoming a headache. Thankfully we were not handling inbound calls / inbound IVR so we didn’t really have that big of a problem yet. Skype numbers did receive inbound calls, but patients started to call at all times of the evening/morning – beyond business hours. That didn’t go over well with our agents. Plus, each week, we had to ask the agents for their call logs. Agents are NOT technically savvy folks 🙂 Getting them to send us calls and SMS logs was… a bit of a challenge 🙂 Considering that we were a data led business, we really didn’t know what was going on with our call center. We couldn’t figure out average handling times that easily. We were wrangling spreadsheets (CSV files exported from Skype) to get typical call center metrics and KPIs out the door for ourselves and our client. Agents were going through hell trying to get phone numbers from google sheets and calling them from Skype. We had to get off of Skype soon.

The hunt for a call center software

Remember, we were not really thinking of being a call center 🙂 We did our research and of course, we gathered a list of the top call center software in the market – Vicidial, Talkdesk, BrightPattern, NICE Incontact, Five9, FluentStream.

We were dialing manually and were averaging about 150 calls / day / agent X 6 agents = 900-1000 calls per day. In a month, we were averaging 20,000 calls. Our average call handling time was around 5-7 mins per call.

What we needed from our call center back then

Our needs were quite basic – we just needed a way to:

  1. Dial outbound numbers using a browser
  2. Receive inbound calls using a browser
  3. Have defined business hours
  4. Be able to pull call center metrics and KPIs (real time and historic)
  5. Be able to transfer calls to our client’s team as and when needed
  6. Not be bound to a contract
  7. Be able to scale up as needed
  8. Be able to integrate with and leverage the rest of the AWS ecosystem (this was super important to us)

Costs of various call center software

From our research, here’s what we gathered

Vicidial – $400 per server per month hosting fee (after first month) plus cost of minutes. 1.5 cents per minute for outbound calls as well as inbound toll calls. 2.8 cents per minute for inbound toll-free calls

Talkdesk – Talkdesk starts at $65 per seat per month. Talkdesk also provides telephony (using Twilio) and charges a price per minute starting at $0.02 per minute for inbound calls and $0.03 per minute for outbound calls.

Bright Pattern – Bright Pattern pricing starts at $70.00 per month, per user. They do not have a free version.

NICE inContact – NICE inContact pricing starts at $100.00/month/user, 

Five9 – Five9 pricing starts at $100.00 per month, per user. They do not have a free version.

FluentStream – FluentStream’s “Essential” package starts at $20 per user per month. However, for access to more advanced features, such as integration and SMS capabilities, you might have to upgrade to their “Advanced” ($30 per user per month) or “Complete” ($45 per user per month) package. 

All of the above didn’t include the pricing for dial minutes.. Which adds to the costs even more so.

So, doing some basic math.. With the lowest of around $ Our small call center of 6 people would cost a bare minimum of $200/- for seat costs plus the dialing charges. If our team grew to 50, seat costs would be a minimum of $1,500/- per month

Comparing to Amazon Connect

Doing a quick comparison with amazon connect’s pricing, 

  • Direct Inward Dial number costs less than $1 per month. 
  • Inbound dialing costs were $0.0022 /minute (0.22 cents per minute)
  • Inbound dialing costs were $0.0048 /minute (0.48 cents per minute)

So, Amazon connect already beat the per minute pricing for all the other providers.

Next up was to compare how much it would cost to use for 8 hrs / day * 20 days a month 

Amazon prices it as (Voice usage) $0.018 per minute. So, that would cost us around $172/month per month per agent if they use the software constantly for 8 hrs per day (which we hope they would).

Comparing this to the “per seat” cost of the other software, this seemed the most expensive. What we saved on dialing costs, were partially lost to this “per agent” cost.

NOTE: After running for months.. We found out that our actual cost per agent per month, along with dialing charges were around $70/month

Customizing our contact center

Almost every call center software integrates with some CRM or the other. We got a little stuck here. We needed to be 100% HIPAA compliant. So, we couldn’t really use any traditional CRMs. We were already using a no-frills healthcare CRM that helped our agents daily anyway. It was HIPAA compliant and we needed our call center software to be able to integrate with it easily. Early on, our management team had already decided that we are not going to be a traditional patient contact center that is in the business of selling “agents”. We wanted to be in a different line of business. We wanted to lead with technology and provide agents for the final mile.

Forecasting future needs

Being in the healthcare business for a while, we knew that software alone will never solve all the problems. Humans are needed for the final mile. Healthcare is so fragmented that we knew we could not rest on our laurels by just connecting to a CRM. We had to connect to a CRM that somehow connects to an EMR and probably an EPM. We lived in the world of faxes – unlike any other industry. So, we knew we would have to connect voice, chat, faxes and more channels of communication. 

Our patients would write to us via SMS, whatsapp, iMessage. Our partners would send us faxes. Some of our calls to payers would last more than an hour. In other words, healthcare was (and still is).. A bit of a messy business. So, we knew we were going to have to customize things a lot.

We were already heavy users of AWS’ ecosystem. We leveraged a LOT of their HIPAA eligible services and had no intentions of moving away from them. Being able to customize our contact center and begin to leverage the AWS ecosystem was a REALLY big deal for us. We decided to try out Amazon Connect Contact Center.

Getting started in less than 30 mins

We decided to give Amazon Connect a shot. What’s to lose? They already had a free tier for us to play around with. It took us less than 30 mins to get started. We were floored – we didn’t need any training and we were up and running. Almost as easy as Skype! Next up, we tried some ACD and set up the IVR. Easy peasy.

We took the IVR a step further and decided to integrate with Lambda for some of the other custom things we were doing. Easy.

We tried to do automated dialing to leave voicemails and to allow patients to talk to an agent by pressing 1. Easy.

We went through various experiments one by one – they all seemed pretty easy.

We ended up choosing Amazon Connect Contact Center.

Adding no-shows re-appointments

Meanwhile, once we started calling for appointment reminders, a natural progression was to start calling for no-shows as well. Our client (like most others) didn’t really have a plan for recalling no-shows. They would, on a sporadic basis, create a list of no-show patients from CareCloud and hand them over to another offshore team to call. 

We didn’t think that approach was regimented enough to improve patient visit volumes. Drawing from our experiences from other industries combined with our own marketing strategies, we figured no-shows should be part of a regimented program as well.

We added calling, dropping voicemails and texting no-shows to our team’s assignments. The team size also grew along with. Taking the route of voicemail drops was not a big deal at all – thanks to Amazon Pinpoint. 

Taking on inbound call handling

At one point we were asked to produce KPIs for our customer’s front desk call handling. As it turned out, each patient had to wait for at least 15 mins for the front desk to get to the phone with them. That’s when we started handling inbound calls as well. Again, Amazon connect contact center came to the rescue. We were ready to do so within a day. All we had to do was to modify the sample contact flow that Amazon connect provides and we were good to go. The only thing that took a few days was porting the existing phone number of our customer into Amazon connect contact center. From day 1, we started handling approximately 250 inbound calls. Our patient CRM came to the rescue as well. For each inbound call, the screen pop would open up the patient record fetched by the patient CRM from the Carecloud EMR and our agents were ready to service the patient.

Amazon’s Spanish speech to text was fantastic (they’ve recently upgraded those features a lot as well, making it even easier).

Ability to do skills based routing helped a lot. We created Spanish, Bengali, Hindi and English queues. We also created quick connects to be able to forward billing and prescription requests to the technicians working in specific locations.

If you would like help from our call center team (patient contact center) services, you can read more here or reach out to us.


Learn more about contact centers