Published on the 24/09/2014 | Written by Newsdesk

There has been a step change in enterprises’ ability to economically capture, store, tag and index every conversation that their call centre ever has with a customer…

In the past the cost of computer power and storage meant that companies which wanted to analyse voice calls from customers would typically sample 10-20 percent of those conversations. But according to Daniel Ziv, vice president of customer analytics and enterprise intelligence solutions for Verint, the default position today is that organisations analyse all their calls.

Ziv, who last week spoke at a conference in New Zealand, said that it was now possible for organisations to store 20 million calls, containing 20 billion individually tagged words, on a single server.

Local users of Verint’s speech analytics systems include HCF, the Commonwealth Bank and Australian Direct Insurance. Ziv said that after large volumes of conversations were collected and the words individually indexed, it was possible to search for patterns in the data.

Ziv added that voices could be analysed for sentiment and emotion. “A five minute phone call typically has 1z000 words. Speech is super revealing,” he said, adding that the system could detect changes in pitch, volume and tone to determine the emotion of the caller.

“When you combine words and emotion…you can identify statistically valid trends,” he said.

One of the benefits he said was to provide companies with insights about which of their call centre scripts worked well. So for example a crude attempt to upsell a customer on a call might regularly fail, while a more sophisticated approach which took in personal data about that customer and then used that to offer a benefit might get better results.

He said that the return on investment could be significant. Ziv claimed that a large Canadian telecommunications company had identified $180 million worth of savings in the first year of using the system, improved customer satisfaction ratings and reduced churn.

Over the coming two to five years Ziv predicted that the technology would grow increasingly accurate, and by linking voice analysis to social media and webchat analysis, companies would be able to develop “very robust enterprise feedback systems”. This would allow real-time analysis of a conversation, with the system feeding insights into the call centre operator’s scripts on the fly.