Seven healthcare lines of business, moved to Amazon Connect
A U.S. healthcare company handles patient and member calls for care management, pharmacy, enrollment, telesales and executive support on its contact center. On Zoom Contact Center it had no robust multi-region failover, routing and caller ID changes were manual, patients got calls from unfamiliar numbers, and Spanish callers missed SLA when Spanish-speaking agents were busy. We moved all seven lines of business to Amazon Connect, added Amazon Connect Global Resiliency across two AWS Regions, and moved routing, announcements, holidays and caller ID into configuration the business can change without a release.
- 92–95%
- inbound calls answered within SLA, up from 75–80%
- 60–65%
- outbound answer rate, up from 40–45%
- 7
- lines of business moved to Amazon Connect
- 2
- AWS Regions with ACGR
- Seven healthcare lines of business moved from Zoom Contact Center to Amazon Connect.
- Inbound calls answered within SLA rose from 75–80% to 92–95%; outbound answer rates from 40–45% to 60–65%.
- Resilient across two AWS Regions with ACGR.
- More patients answered within SLA, and more patients answering when the company calls.
- Single-platform risk removed: two AWS Regions with automatic failover.
- Business teams change holidays, announcements and routing without waiting on a release.
- Spanish callers overflow to an alternate queue instead of waiting past SLA.
- Patients see a familiar number on every outreach call.
- Inbound, outbound, callbacks, voicemail and quick connects for each line of business.
- Lambda called from contact flows decides routing and caller ID from DynamoDB configuration in real time.
- ACGR with us-east-1 primary and us-west-2 secondary.
- Configuration schemas, validation rules and naming conventions; Terraform through a pipeline; CloudWatch.
One call, from both sides
On the left, what the member hears. On the right, what Amazon Connect does: it reads the configuration, makes the routing decision in Lambda, and connects. Pick a call to see Spanish overflow, holiday hours, a familiar caller ID or a Region problem.
- Call comes in
- Check settings
- Choose a queue
- Connect
- Line of business
- —
- Hours
- —
- Holiday
- —
- AWS Region
- —
- Caller ID
- —
- Waiting for a call
- Settings checked
- 0
- Queue
- —
- Wait
- —
- Answered on time
- …
Illustrative calls. Members, numbers and wait times are fictional; the routing follows the production design: configuration in DynamoDB, decisions in Lambda, and two AWS Regions with Amazon Connect Global Resiliency.
The challenge
In healthcare a missed call can mean a missed appointment, a late prescription or a patient who drops out of a care program. The legacy platform had no robust multi-region failover, so an outage was a real risk to patient care.
Everyday changes were slow. Holidays, out-of-hours handling, emergency announcements and caller ID updates needed manual work or ran into platform limits. Patients often saw a different number each time the company called, so they stopped answering, and when Spanish-speaking agents were busy, Spanish callers waited past the service level.
The goals: no single-platform risk, consistent caller ID, protected SLAs in both languages, business changes without waiting on a release, and a foundation for automation and AI later.
What made this migration different
Moving seven lines of business is the job. Making the business able to change them safely afterwards is what lasts.
- 60–65%
of outbound calls answered, up from 40–45%.
Dynamic caller ID shows each patient a consistent, familiar number. We treat caller ID as a first-class design requirement in healthcare now.
Holidays, hours and routing changed in settings, without new code.
Holidays, out-of-hours, emergency announcements, caller ID and routing rules live in DynamoDB. No flow redeploys.
Spanish callers moved to a backup queue when Spanish-speaking agents are busy.
Language-aware routing sends calls to an alternate queue when Spanish-speaking agents are busy. The pattern is now part of our reference architecture.
- 2
AWS Regions, on a platform that had no backup before.
ACGR added as part of the move, without building or running a custom disaster-recovery setup.
What changed
- Inbound SLABefore: 75–80% of calls answered within SLAAfter: 92–95%, measured after go-live
- Outbound callsBefore: 40–45% of patients answeredAfter: 60–65%, with a familiar caller ID
- Caller IDBefore: Patients saw a different number each timeAfter: Dynamic caller ID, chosen per patient and region before dialing
- Spanish callersBefore: SLA missed when Spanish-speaking agents were busyAfter: Language-aware routing with overflow to alternate queues
- ResilienceBefore: No robust multi-region failoverAfter: ACGR: us-east-1 primary, us-west-2 secondary, automatic failover
- Holidays and announcementsBefore: Manual changes for holidays, hours and emergenciesAfter: Configuration in DynamoDB, live without redeploying flows
- RoutingBefore: Static rules, hard to change per line of businessAfter: Lambda decides by line of business, number dialed, language and time of day
- Call typesBefore: Platform limits on what each team could doAfter: Inbound, outbound, callbacks, voicemail and quick connects per line of business
- OperationsBefore: Limited visibilityAfter: CloudWatch logs, metrics, SLA monitoring and alerts
- ReleasesBefore: Changes made by handAfter: Terraform and a deployment pipeline, with a release checklist
Hover or tap a tile to see what it replaced.
More patients answered, and more patients answering.
We agreed on two KPIs with the business and tracked them in Amazon Connect reports before and after the move. The first was how many inbound calls were answered within SLA. The second was how many patients picked up when the company called them.
- Inbound calls answered within SLABefore75–80%After92–95%
- Outbound calls answered by patientsBefore40–45%After60–65%
- Region protectionBeforeOne platformAfterTwo AWS Regions
From Amazon Connect historical reports (queue and routing-profile metrics) and outbound attempted-versus-connected calls, compared before and after go-live by line of business.
How a call is handled now
Every live call checks the configuration first, so the business can change behavior without a release.
- 01 · CallA patient calls, or is calledOutbound calls show a familiar number, chosen for that patient and region.
- 02 · CheckLambda reads the configurationHolidays, hours, emergency announcements and routing rules come from DynamoDB in milliseconds.
- 03 · LanguageEnglish or SpanishIf no Spanish-speaking agent is free, the call overflows to an alternate queue before SLA is at risk.
- 04 · RouteTo the right teamBy line of business, number dialed, language and time of day.
- Care management
- Pharmacy
- Enrollment
- Telesales
- Callback
- 05 · ResolveAn agent picks upOr the patient gets a callback or leaves a voicemail.
If one Region has a problem, patients still get through.
The old platform had no robust multi-region failover. Now every line of business runs on Amazon Connect with Global Resiliency: us-east-1 is primary and us-west-2 is ready, and traffic moves automatically if the primary is impaired.
Calls go to the primary Region. The secondary is ready and waiting.
- Same phone numbers
- Same agents and flows
- Automatic regional failover
The steps in every wave
The same 5 steps for each wave.
- 1PlanEntry criteriaTesting and readiness agreed before the wave starts.
- 2BuildBuild from templatesProven IVR templates, standard Lambda logic and existing DynamoDB schemas.
- 3ProveTest and acceptUAT with the line of business, then sign-off.
- 4TrainPrepare the teamAgents, supervisors and support teams trained.
- 5Go liveGo/No-Go, cut over, stabilizeCutover checklist, rollback ready, then a post-wave stabilization review.
The outcome
- All seven lines of business moved from Zoom Contact Center to Amazon Connect
- Inbound calls answered within SLA up from 75–80% to 92–95%
- Outbound answer rate up from 40–45% to 60–65% with dynamic caller ID
- Multi-region resiliency with ACGR across us-east-1 and us-west-2
- Spanish callers protected by language-aware overflow routing
- Holidays, out-of-hours, emergency announcements and caller ID managed as configuration
- CloudWatch monitoring, a runbook, hypercare and trained support teams
- Contact center
- Amazon Connect
- Contact flows
- Callbacks, voicemail, quick connects
- Resiliency
- Amazon Connect Global Resiliency (ACGR)
- us-east-1 and us-west-2
- Logic and configuration
- AWS Lambda
- Amazon DynamoDB control plane
- Monitoring
- Amazon CloudWatch
- Delivery
- Terraform
- Deployment pipeline
- 2–3 week sprints
Four lessons now shape every healthcare migration we run. Keep business configuration apart from flow logic, or seven lines of business start tripping over each other. Design multilingual overflow before peak hours expose the gap. Treat caller ID as a patient-trust problem, not a setting. And make phased rollout, rollback and enhanced monitoring mandatory at cutover.
A partner AWS has validated, on exactly this kind of work.
- AWS Advanced Tier Services PartnerRecognized by AWS for proven delivery, certified people and customer outcomes.
- AWS Migration Services CompetencyValidated by AWS for moving and modernizing workloads on AWS at scale.
- Amazon Connect Delivery PartnerValidated by AWS to deliver Amazon Connect, including migrations from Avaya, Genesys, Cisco and other platforms.
- AWS AI Services CompetencyValidated by AWS for building AI into real products, including the contact center.
- AWS Experience-Based Acceleration (EBA) partnerWe run EBA engagements with AWS to speed up migrations with hands-on, time-boxed sprints.
- ISO/IEC 27001 certifiedOur information security management system is independently certified to the international standard.
- SOC 2 Type IIIndependent auditors have tested our security controls over time, not just at a single point.
More work
All case studies →- Amazon Connect
- Amazon Lex
- AWS Lambda
- Amazon EKS
- Amazon Kinesis
- Amazon DynamoDB
- Amazon Bedrock
- Claude models
- Tool calling








