Compared with on-premise implementations, some important architecture questions and implementation decisions may get different responses or views when implementing Amazon Connect.  Here are 5 topics that you handle differently:

Hardware and Software Requirements

This is definitely a question that you ask when implementing an on-premise Contact Center solution.  But the question becomes irrelevant when using Amazon Connect

On-premise Amazon Connect
In preparation of your on-premise Contact Center implementation, you want answers to following questions.  These enable you to correctly architect your solution:

  • How many users should we support?
  • Do we need a high-available solution?
  • What channels should we support?
  • What will be the peak load on our platform?
  • How many calls should we record? And how long will we store them?

All these questions will be asked for current and future usage, as you don’t want to design a system that can’t support needs

Irrelevant, as AWS ensures that all Amazon Connect (and other AWS) services run adequately

As a customer, you don’t worry about hardware or software requirements when implementing Amazon Connect.  Like other AWS Services, Amazon Web Services ensure appropriate sizing and maintenance  that is needed for running Amazon Connect.

Neither should you think about user licenses. You pay Amazon Connect as you use it, not by the number of agents/logins that are defined on the platform.

Solution Availability

Also linked with the 1st question, availability requirements are a big thing when implementing an on-premise Contact Center solution.  Again, with Amazon Connect, this is handled by AWS.

On-premise Amazon Connect
Some questions that are typically asked when thinking about availability for an on-premise implementation:

  • What are the availability requirements?
  • Do we require a high-available implementation?
  • How do we organise automatic fail-over?
Irrelevant! As described in this article, AWS provides a 99.99% SLA for Amazon Connect

AWS always operates multiple datacenters in one Region/Availability Zone.  In the scenario that one datacenter would fail, this has no effect on the service that is provided.

Solution Sizing

On-premise Amazon Connect
Some questions that are typically asked when thinking about correctly sizing for an on-premise implementation:

  • How many calls do we get today, and what number of calls do we expect in the next year or two years?
  • What is our expected peak load of calls?
  • Should we plan support for other channels, like Whatsapp or Chat?
  • What is the optimal mix to not spend too much money, but still ensure that we may handle peak volumes?
Sizing is not a question as Amazon Connect automatically handles volume changes.

Including more (support) channels is not a question of sizing. It is a question of adding channels to Amazon Connect or include them via other AWS Services

Growing or shrinking your Contact Center happens automatically with Amazon Connect. Regions where Amazon Connect is availableAWS manages all that for you. So, don’t worry if you expect a peak number of calls around Black Friday. You may easily add users in Amazon Connect without having to upgrade, scale-up or scale-out any hardware.


Location is probably not a question that you ask when implementing on-premise.  But it is an important question when going for Amazon Connect.

On-premise Amazon Connect
Clear, that’s in your datacenter(s) What AWS Region is closest to your Contact Center users?

Amazon Connect is available in different AWS Regions.  Recently, AWS announced two additional regions where Amazon Connect is available.

As an example, it would be strange to implement your Cloud Contact Center solution in Virginia (US) if 90% of your agents are located in Europe. In this scenario, you would chose for Frankfurt or London.  Or if your agents are split 50/50 over US and Europe, you might implement two Amazon Connect instances, one in a US region and one in Europe.

It all depends where your agents are located!


Implementing a new solution always requires new skills, or skills slightly change because the selected solution works differently than the legacy solution.  As such, that is not different with Amazon Connect. Although, as Amazon Connect is an AWS Service, some infrastructure skills may not be needed anymore.

On-premise Amazon Connect
Typically, you own your datacenter, and so you need to cater for supporting it.  That means that you need skills in the areas of:

  • Datacenter management
  • Hardware management
  • Software management

To streamline your support processes, you will want to integrate your new Contact Center solution in your ITSM processes too.

AWS manages Datacenter, Hardware and Software for you.

However, you will want to configure triggers and alarms on AWS (and potentially integrate those with your ITSM solution) for streamlined support


When migrating from a legacy Contact Center solution (or implementing a new greenfield project), and you think about selecting Amazon Connect as replacement, these 5 topics will probably change the way you look at your Cloud Contact Center project.  Compared with on-premise implementations, these 5 topics get answered differently.  This has impact on your project, but also on the people and the skills that you need to support your Amazon Connect implementation.

Similar Posts