Share on linkedin
Share on facebook
Share on twitter

Launch is Day One. How High Availability, Low Latency and Continuous Monitoring are Critical for Voice Apps on Alexa and Google Assistant.

Alexa skills and Google Assistant Actions are growing like weeds. Now that more than 50 million consumers in the U.S. have access to smart speakers, 100 million worldwide are expected to use the devices by year end and about 1.5 billion devices have access to a voice assistant through smartphones and PCs, everyone wants access to the large and growing audience. That includes some of the world’s largest brands and media companies. While it is relatively easy to launch a voice app, there is no clear path to setting up enterprise-grade operations, management and maintenance optimized for this channel. XAPPmedia tackled this challenge to offer our customers:

  • High Availability
  • Low Latency
  • Continuous Monitoring

High Availability is Step One in User Experience

Driving voice app discovery may be the most daunting challenge today for anyone launching an Alexa skill or Google action, but once you get a consumer to your property will it be available? Almost no one launching voice apps can answer this question or have a clear understanding of what is involved and how it differs from mobile or web. Voice is different. However, the needs are the same. If you are going to invest in publishing and maintaining a voice app, then it is best to make sure it is operational when users want access.

XAPPmedia has set up redundant instances of our customers’ voice apps so they will automatically fail over when performance issues are detected that are localized to a region. The reason this isn’t more common is that the space is immature. Many companies launch voice apps as an experiment and haven’t thought about what is involved in a long-term commitment to a technology channel. More often than not an agency builds a voice app and then when it launches, throws the code over the wall to company’s IT team that has little understanding about how it all works. XAPP happens to have many customers that treat voice apps as strategic and long-term assets to sustain. As a result, availability was an obvious requirement from the start.

Low Latency is Step Two in User Experience

Everyone frets over their voice app concept and design. Will it be interesting to consumers? Will it fulfill the job to be done? Will the user experience be delightful? These are important questions. However, what many people fail to grasp is that the first two moments of truth for any voice app experience are whether it is available and how quickly it responds. As long as you don’t fail on these two steps, your concept and conversation design will have a chance to shine. But, if you don’t meet an acceptable bar of performance, users abandon before they even get started.

The impact of latency on consumer engagement has become well understood over the past several years. Data from Akamai found in 2017 that 64% of consumers expect a mobile site to load in less than four seconds and 47% expect a webpage to load in less than two seconds. The same study found that a one-second delay in page load can reduce voice SEO by 11% and a 7% reduction in conversions. We don’t know what the corresponding impact will be for voice apps as they mature, but it is sure to make a difference.

Keep in mind that latency is even more critical in voice conversations which are expected to synchronous experiences. Long pauses kill the rhythm of a conversation and the expectation of immediate or contemporaneous response. With this in mind, XAPP has designed our voice application hosting architecture to minimize latency. One element of this is geography, but there are other factors we have discovered in operating more than 1,000 voice apps in the North America, Europe, South Asia and Oceania over about 12 million user sessions. How you are going to address latency is a factor worth considering before you launch your first Alexa skill or Google Action.

Continuous Monitoring Reveals UX Opportunities

There is also the post-launch question of monitoring and measuring voice app availability, latency and user interaction errors. How do you know when these issues occur? Some voice app performance data is available through the Alexa dashboard and Actions on Google console. Other data is only accessible through your hosting provider and streaming media service. The question for voice app publishers is how they are going to track all of this information from disparate sources and what processes and staffing will be in place to identify, evaluate and resolve issues.

XAPP has built out a solution that consolidates the infrastructure monitoring into a single solution that is complemented by point solutions for log analysis and voice user experience evaluation. This is all provided to our customers as a managed service so they don’t need to worry about plumbing.

Launch is Not the End. It’s the Beginning.

Most voice app projects are a mad dash to build the solution and push it through either Alexa or Google Assistant certification. Once it is available, everyone breathes a sigh of relief. However, that is when the real pressure starts. Your Alexa skill or Google Action in are in the wild and being accessed by consumers and your colleagues who are just checking it out. Both user groups elevate the importance of knowing how your voice app is performing so issues can be identified and resolved quickly–ideally before anyone notices. Three key aspects that are often overlooked are availability, latency and continuous monitoring. Be sure to make them part of your pre-launch planning and a cornerstone of your post-launch capabilities.

If you would like to learn more about XAPPmedia’s managed services for voice apps, drop us a note by clicking the button below.



Related Stories