About Aaron Wellman

Aaron Wellman is a Product Manager at Angel and has been with the organization for over 10 years. Aaron is the product lead for the Site Builder, Virtual Call Center, and Caller First Analytics product lines. With a background in User Interface design and bringing phone application development tools to the market, Aaron will be blogging about anything ranging from new Angel product offerings to unique IVR implementations.  Prior to Angel, Aaron was a Wall Street technologist, designing UIs for investment banking applications in the cloud. Aaron has a Bachelor of Fine Arts (BFA) from James Madison University.

Web Analytics to Phone Analytics Translator

Web and IVR Design

It’s always troubled me when businesses with just as much or more traffic to their IVR than their website clearly put more effort into their web presence than their phone or “Voice” presence. Perhaps the “out of sight, out of mind” adage is a reason why the emphasis on voice isn’t as important. It could be that there are just not that many voice experts out there. It could also be that tools like “Google Analytics” for the phone are not prevalent, economical, easy to use, nor produce easily comprehended data.

Where would web designers and marketers be without Web analytics? They would be completely in the dark – making design decisions based on speculation and the feedback of a handful of users.

Where are voice user interface (VUI) designers without phone analytics? They’re pretty much in the same boat as their web counterparts. Sure, there are known best practices for VUI design that can be taken into consideration, but you may never really know the type of people that will be calling your application and the types of issues they’re having trying to get done what they want done. Until you can take a bunch of calls over a period of time and identify the metrics that tell you where the users are having issues, or where they’re being successful, chances are that your callers will have a poor experience.

Where does one start with phone analytics? Fortunately, there are many parallels to web analytics and should feel familiar. On Angel’s platform, where a functional node of the IVR is referred to as a Voice Page and the collection of those pages is a Voice Site, it makes things a little easier to relate.

For the uninitiated, here’s a few of the metrics I find to be the most useful and examples of how the web and phone metrics relate.

1. Measuring the customers who just give up – Exit and Hang Up Rate: 

Web: The exit rate denotes the percentage of visitors who exited a site from a specific page, rather than, say, continuing to fill out a contact form, however, some exit points may indicate that people had trouble finding what they needed.

Phone: If the section of the application the caller hung up in is not a designated section where a caller should hang up, for example after a confirmation, there’s a relatively high probability that the caller was having an issue in the particular section where they hung up.

2. Hitting the hot spots – Web Page Views and Voice Page Hits:

Web: Page Views tell us the how frequently a web page is visited, which tells marketers that the web page is a popular spot or it could also mean that you’re sending users to a place unintentionally.

Phone: Voice Page Hits tell us the sections of the IVR that are the most popular. It could also reveal issues in the application, especially if it’s a section of the application that’s not supposed to be hit often or only in error.

3. Time after time – Average Time on (Web) page and Average Time on (Voice) page:

Web: Depending on the content, the average time spent on a Web page tells us if the user is interested in what’s on the page, like if there is something to read.  It can also tell us if the page is confusing or malfunctioning if the purpose of that page is to perform a quick operation and continue on to another page.

Phone: If the purpose of the phone call is to get something done or to make a selection to get to a place to get something done, that shouldn’t take a lot of time. If the duration seems longer than it should, there’s likely an issue.

4. Getting things done – Goals and Task Analysis:

Web: In web analytics and specifically, Google Analytics, goals can be set for measuring a process, like completing a sale or a subscription process.

Phone: On the phone, especially if the purpose of the application is to automate a task that would otherwise be handled by a live agent, it’s crucial to identify those tasks and measure the completion rates.  For those tasks that were not successful, it is also important to identify where in the process callers were having the most trouble.

Here are a few more examples:

Web analytics to phone analytics translator.

Hopefully these examples brought a few things to light when it comes to the types of metrics used for analyzing a phone application. Taking a little time to analyze IVR metrics and making the appropriate changes to IVR and continually repeating that process will delight your customers and help you retain them. It can also lead to a lot of cost savings, especially for large volume applications


Responsible Smart Polling

Smart Polling

Running a Google News search on “robocalls” will return a number of blog entries and newspaper stories with varying opinions on IVR “Smart Polling.” The hottest topic on the subject from the last few months was that former Maryland Governor Bob Ehrlich’s campaign manager was indicted for sending 110,000 “misleading” robocalls to voters.

These reports obviously aren’t good for the Smart Polling cause. Just goes to show that you better get your messaging right from the outset. All it takes is one click of a mouse and that campaign is as good as complete, with no humans there to question the poll questions.

Cases like the above have prompted lawmakers to tighten controls on automated political calls. On July 13, Senator Dianne Feinstein of California introduced the Robocall Privacy Act. According to this article: “This legislation would ban automated political calls between 9 p.m. and 8 a.m.; disallow more than two calls per day to the same number from the same campaign or group; ban robocallers from blocking the caller ID number; and require an announcement at the beginning of the call that identifies the individual or organization that is making the calls and states that the message is prerecorded.”

We’ll continue to keep tabs on Smart Polling use and misuse over the coming months. Stay tuned.


Smart Polling for President

With the 2012 election season getting underway, I’m curious if the use of IVR technology or “robocalling” for polling will be under the same scrutiny that it was during the 2008 election.

A few years back, I wrote a blog post, “IVR Polling: Overcoming its C.R.A.P. Reputation,” on robocalling for the 2008 election. One of the issues that opponents of robocalling were concerned about in the 2008 election was that a machine can’t tell if a call is being answered by the intended person. This led me to a web search to learn if any advances in technology have been made along these lines since the last election. I was unable to find any products out there that can detect age or gender using a speech recognizer but did find a few academic papers on the subject that dated back a few years.

Perhaps the market need just isn’t there for these types of products. Yet Rasmussen Reports, who uses outbound IVR technology to collect their data, tied with Pew Research as the most accurate pollster for the 2008 general election. We’ll see if this was anomaly but my guess is that it wasn’t. Also, what does it matter if a kid answers the phone? Chances are that they have the same political views as their parents.

While we’re talking about this technology, doesn’t “Robocall” sound so 1980’s? Are robots picking up the phone and placing calls and speaking in robotic voices? I’m referring to the practice as “IVR Polling,” but perhaps a more appropriate label would be “Smart Polling.” Automating these types of calls is smart because it allows pollsters to gather more data in less time and with less money than they did using humans. And “Smart Polling” is proving to be just as accurate.

We’ll be reporting more on the use of Smart Polling over the coming months. Stay tuned.


Android Takes the Lead

On March 7, comScore announced their most recent smartphone market share results. According to their data, Android now has largest share of the smartphone market in the U.S., replacing RIM.

Android now has a 31% share, up 7.7% from the three month period ended in October 2010. RIM’s (Blackberry) share fell 5.4% with 30.4% of the market share. Apple gained 0.1% and remains in third place, with a 24.7% share.

View the comScore Press Release


Android passes Apple in U.S.

On January 6, comScore announced their most recent smartphone market share results. According to their data, Android now has the second largest share of the smartphone market in the U.S., replacing Apple.

Android now has a 26% share, up 6.4% from the three month period ended in August 2010. RIM’s (Blackberry) share fell 4.1%, but remains in 1st with 33.5% of the market share. Apple fell 0.8% to third place, with a 25% share.

Click here to read more »

« Older Posts