Society’s expectations for more personalized products and services have risen. As a result, vendors need to better understand their customers and prospects in order to remain competitive in the marketplace.
One of the best ways for a company to do this is to spend time examining the interactions between its employees and customers. Speech analytics delivers the insights organizations require to do just that. Speech analytics provides opportunities to improve the customer experience, enforce compliance and protocol as required by law, and mitigate risk.
The marketplace for speech analytics is a nascent environment, and not all organizations are aware the technology exists. With the arrival of 2009, a big question remains in the minds of speech analytics vendors: “Is this the year that speech analytics adoption will cross the chasm from early adopters to the early majority?”
According to an Aberdeen survey conducted in November and December of 2008, 31 percent of Best-in-Class organizations have implemented speech analytics. This compares to only 18 percent of all other companies. Figure 1 shows graphically current and future adoption of this technology. What is evident is the low adoption level and continued growth in the near future.
What Is It, Exactly?
One of the reasons that adoption is low is a lack of understanding of what speech analytics is. Forty-three percent of survey respondents indicate that that they don’t know what speech analytics is.
Aberdeen defines speech analytics as the automated mining of unstructured audio from dialogue faster than real time to gain insight into not only what is being said, but how it is being said (e.g. emotion and tempo). This audio comes from a variety of sources, such as the conversation that occurs between a customer service representative in a contact center and a client who has called up to complain about his widget not working.
How audio is mined is also important to understanding speech analytics. There are two major types of speech analytics: Phoneme-based and speech-to-text (a.k.a. LVCSR — large vocabulary continuous speech recognition). Phoneme-based speech analytics searches audio files for phonemes, which are the most basic sounds we make in a language. It stores the phonetic content of the speech for reference when searching for information in audio files. LVCSR converts phonemes into searchable text, which is then found at a later time by the LVCSR engine, which matches the sounds in audio with specific text in the dictionary.
Why Adopt Speech Analytics?
For speech analytics adoption to continue to grow and thrive, it is important to understand why organizations implement it. Aberdeen probed survey respondents to find out what drives them to introduce speech analytics into their contact centers. Figure 2 shows a snapshot of the overall results.
An overwhelming percentage (66 percent) of companies indicate that they implement speech analytics to improve the customer experience. That being said, organizations also indicate that improvement in customer retention, cost reduction, and improvement in compliance monitoring efficiency are other important reasons for utilizing speech analytics.
Uses and Benefits of Speech Analytics
With the increase in video usage by both consumers (think YouTube) and enterprises (think of that demo video you watched on speech analytics) it is evident that a lot of speech is locked in video. Yet it is difficult to find this information. It is often found using keywords, phrases and metadata (e.g. title, subject, or person) that identify the video. Speech analytics can provide much-needed access to the information locked within the video itself. This can result in companies saving time and money that is spent creating, for example, the right metadata.
The benefits reaped by contact center agents can also be realized by non-contact center users. For example, litigators now have the ability to search hundreds if not thousands of documents that, in the past, would have taken many man-hours to do. In addition, companies in financial services can now closely monitor their business across a much larger sample of conversations for regulatory compliance.
Departments such as marketing, public relations, sales and engineering, as well as industries such as financial services, insurance, healthcare, and pharmaceuticals can also benefit from implementing speech analytics. Furthermore, speech analytics should not be considered just a standalone application. It is also highly useful as a module that can bring value to users of CRM applications, e-learning applications, or business intelligence applications.
Although the adoption level of speech analytics in the contact center is low, there is promise for growth.
Moreover, while the initial growth and acceptance of speech analytics is currently in contact centers, speech analytics is already being adopted in other areas of companies as well as a variety of industries.
Educating the marketplace about speech analytics solutions and its benefits will help to further expand the market for speech analytics.
Steve Lawrence is a research associate in customer management technologies at the Aberdeen Group.