Bringing Social Heat to Enterprise Voice Communications

Posted on April 20, 2009 in Blog, Communications | 0 comments

By Dwight Irving

This post is a big change from the topic of the previous two posts, to one of adapting common enterprise communications services and usage models to incorporate Social Networking and Web 2.0 methods. As my introductory post on this topic I’ll be resurrecting some ideas from old presentations.  I’ll also try to not dig too deep into technical details, so as to avoid audience whiplash.

What are the areas where social networking methods can beneficially impact enterprise voice and video communications?  At a high level, they’re the common motherhood and apple pie issues familiar to most organizations.

  • Improved customer experience
  • Increase employee productivity
  • Improve the product
  • Grow organizational heath

If I were to take off from that list in the normal direction, this blog would be as bad as those bland articles that keep the ads from running together in many industry fishwraps (not you Internet Telephony mag!).  So where does the master of the obvious take this blog now?  How about I pick one feature or product representing each of the above four points and describe what’s cool about it.  To keep this blog to a manageable size, I’ll do the first two points in this entry, and the next two in a future entry.

1. Improved customer experience

Have you, as a jedi-level geek, ever called a support line and gotten someone who  started by asking you to reboot your system and see if that helped?


Have you, as a tech-noob, ever been asked to verify that your subnet mask is appropriate for your LAN segment?  Or has someone called you a tech-noob and you didn’t know what that meant, but you did feel somewhat insulted?

In either case you’ve run into the dreaded support skill level mismatch.  It’s a common issue that I’ve seen repeated on many service quality surveys, yet it has a relatively simple fix.

After many support interactions, the support caller often has the opportunity to rate the support agent.  Why not allow the support agent to rate the caller?  Agents rating callers  should provide feedback on customer temperament, willingness to follow directions and technical skill level.  The next time that caller needs support, that information can be incorporated into Automated Call Distribution (ACD) system logic and an agent with complementary skills and temperament can be selected.

I’ve heard from others that this type of logic is already being incorporated into call center products, but I can’t claim first-hand knowledge of it.

2. Increase employee productivity

For this example, I’ll focus on employees who work remotely.  When you finish installing your mobility solution that provides employees with all the features of the enterprise PBX while they’re on the road, you might want to start thinking about how you can replicate all of the back-channel communications paths you find in an office environment.  How do you replicate a team going into a conference room together, where all can hear and contribute to the discussion equally?  How do you replicate the conversational snippet heard over the cubicle wall that drives you to jump onto another agent’s call and provide the solution you figured out last week?

Appropriate social networking and Web 2.0 tools that can be applied to these situations are Instant Messaging (IM), Twitter, tag clouds, and social groups.  Tag clouds?  Review this and meet me below.

Let’s say that you’re a call center agent working from home. How do you hear something happening over the metaphorical cubicle wall? One method would be to define a common set of tags (keywords) appropriate for the types of issues the call center handles, and then incorporate as tags the companies and contact names from the corporate CRM system. Live conversations can be monitored for matching tags using Voice Recognition for voice calls, and a keyword text search that does the same thing for emails and text chat sessions.

A tag cloud display can be moved to the side of an agent’s screen and is observable without  too much distraction from the agent’s focus of attention.  In a tag cloud, the tag size, position, typeface and/or color can be made dependent on the number of times the tag is found in the monitored content.  The “value” of the tag is its count.  When using tag clouds to monitor call center conversations, the tag value should be calculated based on additional factors.  Potential value factors may be:

  • Customer weight – high/med/low revenue, high/med/low influence, etc
  • Emotion – using contextual analysis and voice stress, higher emotional levels should raise the tag value
  • Agent experience – an inexperienced agent may need more assistance so tags associated with that agent’s conversations should be weighted more.

Different tag attributes, like color, position, size and font can also be mapped to represent any of the individual value factors.  For example, color can be related to emotional level and the tag can be bolded for passing a threshold of the number of times the tag appears in a context involving important customers.

Tag value should change quickly to reflect current conversations, so a decay factor should be added.  I like to use a standard half-life decay factor of 2 ^(t/thl) (2 with the exponent of t/thl where t is the current elapsed time and thl is the half life time).  I’d suggest a thl of 400 to 600 seconds to start.

That’s enough for one sitting.  Feel free to pick your own example of social networking or web 2.0 techniques that can be used in communications and add it to the comments.