Debunking Amazon Connect's Competitive Threat
At Enterprise Connect last month, Amazon Web Services got all kinds of "oohs" and "ahhs" when it announced the Amazon Connect contact center service. But were they warranted?
The Amazon Connect announcement served as the catalyst for a number of news stories speculating about troubled times ahead for some of the most established contact center players. Below is a sampling of headlines and quotes I've seen in mainstream publications regarding Amazon Connect.
- Amazon Connect Could Mean Big Trouble For Twilio
- Amazon Strikes Contact Center Market
- Companies including Five9 Inc, Broadsoft Inc and inContact are most at risk from Amazon's new offering
I get it. AWS is cool, disruptive, and anything it touches will turn to gold at the expense of everyone else, correct? Well, to quote President Donald J. Trump, "Wrong." Let me explain why.
Given the massive size and scale of AWS, Connect certainly has some potential. However, its self-service nature makes deployment a challenge. Connect ties into other AWS services, such as those for storage, compute, and analytics, as well as with Lex, its bot framework. Someone needs to put all of this together, and so it seems that AWS is targeting current AWS developers. But these are not the main buyers or even influencers of call center software.
Businesses can't really do anything with Connect until they go through the seven-step AWS sign-up process, during which they need to specify which S3 bucket they'll be using for storage, and so on. To be successful, AWS needs to get to contact center managers or the heads of customer support -- but they are totally new buyers for the company, and they don't care about the infrastructure. AWS makes the self-service model seem like a good thing, but many organizations simply don't have the skills to set up a call center from basic building blocks. For those organizations, a full turnkey solution is a better approach.
Also, while AWS has been the poster child for cloud computing, there's a big difference between selling IaaS/PaaS and SaaS. While we toss everything under the big umbrella of cloud, cloud contact center is SaaS, and that has an entirely different sales motion and buyers. AWS has dabbled in SaaS, via its WorkMail email and calendaring service (dubbed the Exchange killer at one time) and WorkDocs document sharing service, but has not had much success to date. In another indication that AWS is getting more serious about SaaS, it also recently introduced a video team meeting service called Chime... but still, success requires more than just a handful of products.
Another issue I see is the usage-based pricing. Conceptually the pay-for-what-you-use model makes sense, but budgeting for it is difficult. Also, I'm not convinced it's cheaper. The advertised rates for Connect are 1.8 cents/minute for the software services and another 1.2 cents/minute for inbound toll-free calling. Add these up, and you get 3 cents/minute. Take a call center agent that talks four hours a day, or 240 minutes, for 22 days per month, and that's 5,280 minutes/month. Let's get conservative and knock that down to 4,500 minutes; at 3 cents/minute, that's $135/month per agent. This doesn't include storage, encryption, speech analytics, or any other AWS services. I can easily see how Connect usage could approach $200 or more per month per agent. So where are the savings?
Another inhibitor to adoption could be Connect's limited functionality, as the service is voice only. Almost all modern call center solutions today offer a rich omnichannel experience. Connect is designed for use from a CRM application or desk phone. If a contact center wants to leverage softphones, the only option is a WebRTC client, which today limits browser support to Chrome and FireFox. I haven't done a statistical study of this, but from what I know, few contact centers are using WebRTC given the number of old computers out there running Internet Explorer.
Additionally, call routing seems very basic. Connect does not offer workforce management, outbound predictive dialing and has very limited skills based routing. For example, it does not support the ability to assign proficiency levels on skills, so the routing is binary. This means the call center can't be set up to send high-value calls to highly skilled agents and other ones to lower levels.
While the product is telephony only, the phone connectivity is limited. The service is locked into a single PSTN carrier so the contact center must use AMCS LLC, which appears to be an Amazon entity, from what I can tell. I'm not sure which network operator sits behind this service, but Connect is available today in the U.S. and 18 European countries. That's not bad, but it doesn't meet the needs of a global organization. AWS appears to offer a minute set of phone numbers, with no ability to search for specific numbers.
Calling Connect a Version 1.0 service is being generous, as it's missing so many critical capabilities that all businesses need. You could make the argument that this might appeal to a small business, but such a company wouldn't have the skills to build a contact center from a bunch of Lego blocks and then do the integration via APIs and SDKs. AWS may be successful with Connect, sometime in the future, but this isn't an easy industry to service. The contact center is the lifeblood of many businesses, and I don't believe a provider can back its way into it with a half-baked solution. If GE Appliances, featured as part of the keynote, is using Connect to the extent it says it is in a proof of concept, it mustn't have an omnichannel or multichannel contact center, and is doing voice only. That may be sufficient for some organizations, but it isn't for most.
This stuff can be very difficult to do well. When the voice industry transitioned from TDM to IP, vendors like Cisco and Microsoft spent years getting their products to the point where they were business grade; the call center space is no different. Customers demand a lot, and the vendors need to deliver a lot, as only excellence will do. The bar is set extraordinarily high, and I can't imagine that level of expectation can be met with the swipe of a credit card and the assemblage of a handful of building blocks.
AWS made a lot of noise at Enterprise Connect, but we'll see if it has the desire to do what it takes to stay in the contact center business for the long haul.