Siri is a house­hold name despite not being a real per­son. It goes with­out say­ing that voice tech­nol­o­gy is here to stay, but what does that mean for data analy­sis? This blog post will help your orga­ni­za­tion under­stand and imple­ment voice analytics.

Voice technology: a short history

Voice inter­ac­tion with com­put­ers has been around for a while, pop­u­lar­ized ear­ly on by Drag­on Dic­ta­tion. How­ev­er, it didn’t go main­stream until Apple Siri hit the mar­ket. And now-a-days, it’s stan­dard for smart phones to respond to com­mands in the blink of an eye.

Siri and its alter­na­tives cer­tain­ly make intel­li­gent per­son­al assis­tants. That being said, they are still dis­con­nect­ed from a direct rev­enue stream. Enter Alexa Voice Ser­vices (Ama­zon). With Alexa, you can pur­chase any­thing avail­able on Ama­zon using just your voice and an Echo. “Alexa, add cook­ies to my shop­ping cart.” “Alexa, play the sound­track to Stranger Things.” ‘Alexa, do this and that.’ Your house can con­stant­ly be wait­ing for your next request!

Luck­i­ly, Alexa and oth­er voice plat­forms are open­ing up to 3rd par­ty devel­op­ers. As a result, your busi­ness could start engag­ing with your users/customers with­out them ever need­ing a phone or com­put­er. The future is great!

But what does the future mean for analytics?

As dig­i­tal mar­keters and ana­lysts, data is our sword and shield. When mobile apps first came out, we didn’t know what to do, leav­ing us rather exposed. We had to adapt the solu­tions that we already knew (web ana­lyt­ics) to this new world.

Now, we are in the same posi­tion again with voice ana­lyt­ics. We are unsure how to adapt, and end up mak­ing best guess­es on solu­tions to new prob­lems. If users don’t use phones or com­put­ers to engage with your busi­ness, then tra­di­tion­al mea­sure­ments become use­less. Track­ing site refer­rals, down­loads, and video plays? All irrel­e­vant. Per­haps the future is not so great.

The silver lining

This all may sound bleak, but voice ana­lyt­ics aren’t as com­pli­cat­ed as they sound. It helps to com­pare it to an inter­ac­tive voice response sys­tem (IVR). We are all famil­iar with ring­ing up a call cen­ter. You end up trapped in the night­mare of an IVR sys­tem, mash­ing the “0” but­ton in hopes of get­ting to a human. But as annoy­ing as talk­ing to a com­put­er may be, the solu­tions imple­ment­ed in an IVR are some­times use­ful. At min­i­mum, they can read a caller’s account his­to­ry and inter­pret the tone/temperament of the the caller.

The prob­lem? These solu­tions just route you to the right per­son or area of the IVR. In oth­er words, they can’t col­lect data in a mean­ing­ful way.

Now imag­ine that you could col­lect voice ana­lyt­ics on each indi­vid­ual user (or in aggre­gate, for you pri­va­cy cau­tious folks). Think about the ques­tions you would need answered. What was the user try­ing to get help with? How did they ask their ques­tion? What response was most effec­tive in resolv­ing their interaction?

Due to a matur­ing ana­lyt­ics mar­ket, there are ana­lyt­ics prod­ucts that can answer these ques­tions. Thanks to flex­i­ble offer­ings like Google Uni­ver­sal Ana­lyt­ics, Mix­pan­el, and oth­er event dri­ven ana­lyt­ics solu­tions, you can ana­lyze users’ voice queries with­in your ana­lyt­ics software.
Speech Analytics

Voice analytics example implementation

Your devel­op­ers can include voice ana­lyt­ics code like they nor­mal­ly would oth­er event dri­ven ana­lyt­ics solu­tions. Even bet­ter, they can do it in a lan­guage they already use, such as Node.js.

We’ll use Alexa Skills as an example.

In the devel­op­ment envi­ron­ment, run the following:
Google Uni­ver­sal Ana­lyt­ics: (g)
npm install universal-analytics --save

Mix­pan­el: (m)
npm install mixpanel --save

Include and adjust accord­ing­ly in your appli­ca­tion index.js file:
(g) var ua = require('universal-analytics');
(g) var gUA = ua('UA-XXXX-XX'); // your Tracking-ID

(m) var Mixpanel = require('mixpanel');
(m) var mixpanel = Mixpanel.init('aaa111bbb222); // your Token

Track­ing Events
If the user was not understood:
(g) gUA.event("user error","misunderstood statement").send();
(m) mixpanel.track("user error",{result: "misunderstood statement"});

A suc­cess­ful query:
var utteranceData = ("intent: " + utteranceValue).toString();
(g) gUA.event("user query","successful query", {query: utteranceData});
(m) mixpanel.track("successful query", {query: utteranceData});

A failed query:
var utteranceData = ("intent: " + utteranceValue).toString();
(g) gUA.event("user query","failed query", utteranceData).send();
(m) mixpanel.track("failed query", {query: utteranceData});

Cap­ture bugs in your Intent:
var veryBad = false;
(g) gUA.exception("out of memory", veryBad).send();
(g) gUA.event("intent error","out of memory", "fatal: " + veryBad).send()
(m) mixpanel.track("intent error", {error: "out of memory", fatal: veryBad});

User ends session/engagement:
(g) gUA.event("exist","session ended).send();
(m) mixpanel.track("session ended");

Of course, there are some pret­ty sub­stan­tial dif­fer­ences among voice plat­forms. Pro­gram­ming lan­guages and user iden­ti­fi­ca­tion in par­tic­u­lar tend to vary. But it’s noth­ing some app-based mod­i­fi­ca­tions can’t fix.

Now that your devel­op­ment team is build­ing appli­ca­tions with voice inter­ac­tion, each of those inter­ac­tions can be sent to your ana­lyt­ics solu­tion. You’ll be well on your way to under­stand­ing the user expe­ri­ence in this new technology!

If you’d like to learn more about ana­lyt­ics for non-web cen­tric tech­nolo­gies like apps, mobile apps, voice tech­nolo­gies and more, get your free tick­et to ObservePoint’s Ana­lyt­ics Sum­mit. I’ll be host­ing a ses­sion called “Hey Siri, What are App Ana­lyt­ics?”, where you’ll get a crash course in track­ing data for these emerg­ing technologies.