A recent post on Speech Tech Magazine's website mentions that the biggest cost of rolling out a speech project is the professional services and not the software licenses. The author quotes Gartner’s Steve Cramoysan as saying the average per-port price for speech recognition licenses is $770 in 2004, down 10% from 2003. She calculated (without showing the math) that speech software licenses are no more than 15 percent of the cost of a project for speech.

Stepping back, it's fair to say that a well known fact is: labor is the top cost of any endeavor in almost any situation. People cost money and an investment in technology allows people to be redeployed to higher-value (and more interesting and self-fulfilling) work. What's interesting here is that she makes the claim that "much more work is required before users will be able to purchase packaged applications off the shelf and run them on any platform." I don't think that's really the issue at hand.

Personally, I haven't seen a packaged IVR application that works truly out of the box that a company does not want to customize (okay, potentially voicemail and email). At the end of the day, most companies want to differentiate their IVR or have business rules that make things somewhat (or very) unique to their environment or their customers. It's about business advantage and creating a unique and special relationship with a customer.

The key here is not really the 1 time upfront investment and how do I pay for it that drives the rate of speech adoption (that's financing). It's not about implementing a package (or not) or using professional services (or not). The key is: How do you make customers fall in love with doing business with you? I think it's how you "wow" them and nuture them as a business relationship (e.g., with truly impressive customer service). If you understand your customers and provide great service to them, they'll almost never leave you. To do that, you need to think about customer interactions and then make them great. If those are with IVR applications, then consider how strategic they are, how you maintain them, and how you nurture them over time. Just like you nuture business relationships...

Posted by Cory at 07:00 PM | Permalink

George Ou over at zdnet.com recently posted an article which exposes a possible new security flaw in Skype 2.0. Apparently, it attempts to write data into executable memory space, which is a potentially dangerous situation called a buffer overflow.

While it is most likely innocuous in this case (as there is no real reason for Skype to purposefully include buffer overflows in their own software), under some circumstances these are dangerous backdoors for hackers. In short, a buffer overflow is a mechanism by which a programmer can force executable code into memory in such a way that a program is tricked into running it. The fact that Skype exhibits an inability to detect its own buffer overflows indicates that it may be susceptible to intentional overflows.

However, I almost wonder if Mr. Ou's computer has contracted a virus unbeknownst to him. On the other hand, it's possible that the Skype developers simply overestimated one of their buffers, and accidentally spilt out of it in the course of normal (non-malicious) code execution.

That said, George does offer some excellent guidelines on how to enable buffer overflow detection in Windows (called Data Execution Prevention). His article is most definitely worth a read, and his suggestions are a great starting point for the security-conscious.

Posted by Vance at 11:52 AM | Permalink

I think either someone has far too much time on their hands or they really, really don't like the IVR. It's probably the latter. Paul English showed up on ABC 7 News in New York City last week touting how he's out to force companies to abandon their IVR's and make real people answer the phone. And, his blog is helping. [IVR Cheat Sheet]

An example from his blog: Want to get through to Paypal Buyer Credit? Dial 866-571-3012 and then, according to Paul's site, "be silent and at the end of the options press 6." Not bad, considering online businesses are often hard to get in touch with and tend to be fairly web-centric (not surprisingly).

But maybe there is a point here. Businesses of the world -- improve your IVR, or find yourself with more live agent calls. Case in point: Citibank's new Simplicity credit card touts "There should be an easy way to skip automated voice menus." That's a conscious business decision, but anyone else who isn't prepared -- watch out.

Posted by Cory at 11:52 PM | Permalink

New York Java SIG has an upcoming meeting on Wed. Jan. 25, 2006, 6:00 - 8:00pm at the Sun Microsystems office (101 Park Ave - 40th St., NYC).

The speaker is Jeff Barr and the topic is on builting you own Swing Interface to Amazon.com as well as web-application for cellphones using the Amazon Web Services Initiative. "Amazon Web Services is a set of API calls that provide direct, programmatic access (via Java of course) to the underlying Amazon.com web site. This may give us some insights to Audium API for our customers.

The meeting is free, but you must register before the day of the meeting if you wish to attend.

Posted by Susan at 05:31 PM | Permalink