Zen3 touts conversational AI call centre platform developed with Teletext Holidays

A natural language artificial intelligence analytics system developed for UK OTA Teletext Holidays is being opened out to third party subscribers Continue reading →

A natural language artificial intelligence analytics system developed for UK OTA Teletext Holidays is being opened out to third party subscribers.

Sayint has been developed by Indian technology firm Zen3 Group, which took a majority stake in Brighton-based travel tech specialist CWT Digital in September 2015.

The technology turns call centre conversations into text so that they can be analysed for sentiment and compliance and the data used to improve performance and provide business intelligence.

Zen3 says the Service as a Software (SaaS) system takes advantage of reduced costs of computing to offer firms an affordable version of enterprise systems that can cost up to £200,000 to implement.

CWT founder, Steve Endacott, non-executive chairman of Zen3 and chairman of Teletext, said: “There are loads of big data companies out there but one area that is neglected is voice.

“Go to any call centre and they will all have call recording in place but in practice only one to two percent of calls are actually monitored by a person because they have to listen to the entire call.”

The biggest influence on the reputation of your brand is not just people who book, it’s people who do not book with you.

Endacott said because all calls are data mined managers can be directed to which need to be listened to, either to resolve problems, or to understand what best practice looks like.

He said traditional call centre analytics focus on the people who book, but if a firms achieves a 25% conversion rate, that leaves 75% of interactions it is not learning from.

“If you look at the amount of people who look at review scores before bookings, it’s a large proportion,” Endacott said.

“The biggest influence on the reputation of your brand is not just people who book, it’s people who do not book with you. They dictate your brand perception just as much as bookers.

“Improve the satisfaction levels of those people and you improve your review scores and in turn you improve your costs of acquisition. But usually you have limited visibility of non-bookers.

“In a social media world it’s all about measuring the deliverable of your customer service, it’s not about spending more money on advertising.”

Endacott said Sayint allows firms to identify the best performing staff for conversions, but will shed light on whether this is being achieved at the expense of brand perception on non-booking callers.

Since implementing Sayint 18 months ago, Teletext claims to have seen a 26% improvement in customer satisfaction among non-booking callers.

The online agent, whose sole fulfilment partner is parent Truly Travel, operates a click to call model, acquiring customers online but directing them to its Indian call centre.

Endacott said Sayint had been configured for Teletext to address specific challenges of outsourcing to an overseas call centre and the inherent issues with spelling and language.

Error identifying and fixing has saved the firm £177,000 in potential costs and the number of people employed to manually check for errors has been reduced from 16 to just four.

Sayint has been developed to be adaptable and configurable to the needs of any travel companies handling large volumes of calls.

Users can input bespoke compliance rules and trigger words that indicate sentiment so the performance of call centre operatives or homeworkers can be monitored more easily.

Managers can search calls for these trigger words and quickly find where in each call they were said by the customer or agent.

A dashboard provides a clickable key term word cloud showing which words are most frequently said during a call. It also monitors call duration, hold time and how much silence there is on the call.

Any call back promises can also be monitored to make sure they happen and configurable category filters can also be set so that key tasks, like booking disability assistance on a flight, are done.

Call quality monitoring means compliance expectations can be set which could help firms achieve specific KPIs like the offering and sales of add-ons and ancillaries.

The system also analyses non-voice engagements like social media posts, reviews, and chatbot conversations to shed light on which are being used most successfully to provide customer service.

Endacott said: “This is not an ‘out of the box’ solution. This is a foundation that allows companies to understand the issues they have.

“It’s a SaaS model, so it’s pay as you go. We will work with the client to see what their issues are and they can invest more and more as they see returns.

“If after six months they decide they do not need it, it can be turned off.”

Zen3 is targeting Sayint at large call centres, OTAs, homeworking firms with at least 50 operatives and says costs are dependent on call volume.