Unleash Speech Analytics and Dig the Gold from the Contact Centre

Sean Murphy takes us through the benefits of speech analytics in the contact centre and explains how the technology works.

Human beings are flawed – we’re prone to subjective evaluations and in the world of customer service this is a problem. It’s really a problem when traditional monitoring methods within our contact centres are reliant on these judgements, costing firm’s time and money. Luckily, speech analytics has evolved to unearth the revenue previously lost in uncharted conversations, but are you using the right vendor?

Whether you’re a debt collector or an internet service provider, monitoring and analysing calls is vital when it comes to assessing trends and reacting effectively. Traditional methods are often labour intensive, requiring coaches and managers, while only monitoring around two percent of the calls. Meanwhile, advisors assessing contact centre conversations risk stereotyping generic content. On the flip-side, vaguely defined, multidimensional customer evaluation forms are not definitive, causing more difficulties than improvements.

To really carve out those revenue generating nuggets, we need to go beyond personal decisions on whether an advisor remained ‘focused’, or ‘attentive to the customer’. Automating the quality management process using speech analytics is one way forward-facing businesses are able to truly tighten up their customer experience, with minimal labour and cost savings.

Making the switch

By switching to speech analytics, businesses can:

• Analyse all conversations across all channels of content in exactly the same way using a single application.
• Reliably identify predefined topics of categorisation of contact reasons and automated Quality Management.
• Automatically discover emerging trends and unexpected events that aren’t pre-defined by users.

Gone are the days when agents are frantically trying to identifying trends, and text and speech customer behaviour is analysed separately. Once conversations are categorised by topic they can now be directly correlated to KPIs using speech and text analytics, exposing the root of performance issues. The business value driven from the ability to search for words and phrases within customer interactions, and predict future trends then, is obvious.

Great, but how does it work?

Well, there are three types of technology available:

• Phonetics – Phonetic search’s can quickly forage through vast amounts of media, allowing search for words, phrases, jargon, slang and other words not readily found in a speech-to-text dictionary.
• Speech to text – Transcribing text to speech is accurate and hardware intensive. As the output is text, analytics can then be used to identify trends.
• Speech to phrase – Whole chunks of speech are converted without first being transcribed to text.

Currently, Phonetics and Speech to Text are the most commonly used form of speech analytics in contact centres. However, Speech to Phrase capabilities are developing quickly, as evodemt in the example below.

Case study: iiNet

Internet provider iiNet took on Genesys’ Speech to Phrase analytics, to conquer high Average Handle Times (AHT) and understand the intricate nuances and purposes of conversations taking place.

The solution included 100 percent monitoring of all calls and targeted advisor training, due to the identified issues. This resulted in the reduction of AHTs of seven percent across all support call centres, saving ££2,271,325.

Michael Toohey, Chief Information Officer at iiNet, explained that: “Every extra second we spend on the phone costs us money, so Speech Analytics starts exponentially saving us money”

Many businesses, like iiNet, now aim 100 percent of their conversations transcribed for analysis, but the majority of analytics products only transcribe around 25 percent. So, businesses that are serious about improving their customer experience are advised to consider the quality of the technology and frequency of conversation monitoring, provided by their speech and text analytics vendor. Who knows what insights they could be missing?