What Are You “Seeking” with Your Interaction Analytics Solution?

hideandseek

Common Misconceptions

Lately, as I’ve been out at conferences or just around in daily life, I’ve made an interesting observation.  When I meet someone new, they will inevitably ask what my company, Nexidia, does, and when I tell them that we do customer interaction analytics, or speech analytics if you will, I almost always get a response that goes something like this: “Oh, so you have a technology that does word spotting.  It lets me find words and phrases in recorded audio.”  I get this same answer from nearly all people, whether they work in the contact center industry or not, and regardless of whether they had any prior knowledge of speech analytics.  Previously, I haven’t thought much of it, but lately I’ve come to realize how this response is representative of a misconception that we need to do a better job at dispelling.

Playing Hide and Seek

Many people think of interaction analytics as a sort of “hide and seek” technology.  They perceive that there are key words and phrases “hiding” in their recorded audio that offer a wealth of business information, and the solution goes out to “seek” it.  And while technically, the solution is searching for words or phrases – or in the case of a solution that is based on phonetic indexing and searching, phonemes that comprise words or phrases – in order to be successful with interaction analytics what you really want to find with the technology are events.  An event can be anything that happens such as an upsell, an agent enrolling a new member, an authentication or a debt collection attempt.  And to find it, you need to do more than just search for a word or phrase.

For example, a wireless company has found that they can reduce their churn rate if agents offer customers who complain about rate increases an upgrade to a better mobile device.  If the wireless company wants to determine how many agents are using this save technique, they’d need to search for the event of making this specific offer.  Simply searching for the words “mobile device” within their recorded audio wouldn’t necessarily uncover what they were looking for, because that could involve calls where customers called to request the mobile device on their own, agents were attempting to upsell this device, etc.

Crafting a Search

The magic behind interaction analytics is that it allows users to target and quantify specific events, which can then be trended, reported on and used to generate meaningful change.  This is done by creating complex queries – oftentimes referred to as advanced searches – through something called “structured logic.”  The term “structured logic” refers to the ability to weave words and phrases together with time operators and Boolean logic to express complex events.  Boolean and time-based operator-enabled searches allow for terms such as AND, OR, NOT, and WITHIN x SECONDS, refining results and delivering the specific interactions that best represent the scenario you’re trying to find.  In the example above, a complex query would have been created that searched for phrases like “offer” AND “mobile device” AND “upgrade” WITHIN 10 seconds of “cancel my contract” OR “want to quit” AND “rates too high” OR “costs too much” in order to find agents who made an offer for an upgraded mobile device to a customer who just threatened to quit over rate increases.  Additional queries could be written to determine how many times customers accepted the offer to calculate the success rate.

OctChart

Measuring Impact

By using an interaction analytics solution to search for events, rather than a simple word spot, you’re able identify and disambiguate business level concepts.  This gives companies the ability to statistically quantify the number of times something is happening.  By measuring the frequency of an event, the impact to the company can be determined so that the appropriate actions can be taken.  Studying events also allows for correlations to be drawn.  Understanding why something happened, versus just when or if a word was said, enables relationships to other business metrics like CSAT scores, retention rates or net promoter scores to be determined.  Over the long-term, these types of measurements and relationships are far more valuable than simply being able to find a few examples where someone said a particular word or phrase.

Concluding Thoughts

So as an industry, I think it’s time we change the vernacular.  The next time someone describes interaction analytics as a solution that finds words or phrases within captured interactions, we need to expand the definition.  We need to go from word spotting to event spotting.  Words and phrases are static but finding an event is finding an action that is being taken by either the customer or the agent that’s having an impact on the business.  So using interaction analytics to quantify and understand the events taking place in your contact center is the key to creating meaningful change throughout the rest of your organization.

{Photo from Flickr}

Categories: Intro to Interaction Analytics