Enter the avatar

Deploying intelligent self-service needs a less creaky contact centre.

Read time 11min 20sec

It's10am on a Tuesday, 30 years ago. You own the corner cafe and you are the customer contact centre for everyone walking through the door.

Fast forward to 2012. Your business has thousands or millions of customers with demands and questions. You sell new products and services every year, change the old ones and enter new markets. Making things even more interesting, your customers and the customers you'd like to have prefer new ways of talking to you, often mobile ones. And many of them prefer controlling the interaction themselves.

What doesn't change much is the customer's desire for good service like it was at the corner cafe decades ago: a desire that someone interacts personally and politely with you, remembers what you tell them, remembers your past transactions and preferences, acts efficiently on complaints, and, above all, resolves the problem that brought you there. Especially if it turns out to be complex.

Your avatar for today

But when speaking to the owner isn't possible, how do you still experience personal service and feel understood? Wouldn't it be great to talk to someone almost human about your problem, someone bright-eyed and bushy-tailed, even at 3am, who can find answers so you don't have to, who never gets angry even if you do?

Almost-humans exist already, working hard for companies in Europe and the US, on Web sites, mobile devices and social media. A few have had their duties extended from support to marketing and sales. Some even sport fetching avatars complete with believable facial expressions on your computer screen. They mostly hear and speak text and they're called Intelligent Virtual Agents (IVAs).

In critical moments of truth, what are you doing to continue the conversation?

Simon Cranswick, Dimension Data

As an example, avatar Lena at Kaspersky Lab Europe has had her duties extended from support to helping customers select products online. Companies such as Virtuoz, Intelliwise, Creative Virtual and VPI provide IVAs for enterprise self-service. The Web site www.chatbots.org lists implemented IVAs. Meanwhile, analysts at Gartner, Frost & Sullivan and Cisco have predicted a growing market for IVAs.

On one side, high customer expectations. On another, really cool self-service tech. And another: contact centres generally not great at making people happy.

There are many reasons for this. Sometimes 'large organisation culture syndrome' sets in, says Kevin Staples, director at Sepia Unified Communications, part of the BBD Group. Then it's a battle just to get basics like customer service and first call resolution (FCR) rates sorted out. High turnover can keep staff mostly underskilled, he adds.

Creaky guts, broken loops

Were all human problems solved in a typical contact centre tomorrow, the technical nightmare would remain: painfully expensive legacy technology from a bunch of vendors pieced together, much of it incapable of Internet protocol (IP) and VOIP (voice over IP). These generally create functional silos with badly structured data incapable of giving a single view of the customer.

Invariably, the non-VOIP telephony system has several other key non-IP-capable technologies hanging off it, such as interactive voice recording (IVR), speech recognition and analytics (and video conferencing for those with enough bandwidth), says Staples.

The net result is suffering when trying to integrate a patchwork of vendor technologies in the contact centre and separate self-service channels, he says. Meanwhile, extensive integration of IP-SIP-VOIP-generation technology back into a company's legacy systems is simple.

A legacy non-IP backbone in the contact centre has at least two other effects. Firstly, getting analytics out of such an environment is incredibly tough, though badly structured data plays as big a role.

Further, a critical technology widely used in unified communications to tie together VOIP, instant messaging, presence information and multimedia for a seamless multi-channel customer experience, and to provide context about customers, is probably not available. That is because SIP (Session Initiation Protocol) relies on IP end-points.

Human problems. Technical problems. But an unseen iceberg in the contact centre industry may wreck even more customer interactions via contact centre and self-service. Some customers intuitively know about this, without being able to put a name to it: 'How come your marketing department knows all about me, but I'm just another faceless number to your contact centre and self-service?'

The reason is a loop that should be closed, but generally isn't: the feedback loop from marketing to the contact centre, and back to marketing. Marketing runs deep analytics on customers and gives the details to sales. But the same details don't yet reach customer service, says Staples.

Customer experience

So how to meet customer expectations, provide self-service including cute avatars, and mend that broken loop?

Frontline staff usually know the gist of a problem, but analytics puts a price on it.

Nicola Millard, BT

Some things won't change. Customers like control. They also like real-time help when they cannot resolve something complex themselves, so there is a human need to be able to pick up a phone after exhausting self-service channels, says Martin Dove, director of strategy and innovation at Merchants. This suits companies, which don't want to lose the interaction and the opportunity to cross-sell and up-sell their customers, he says.

Currently, the vast majority of interaction (estimates vary between 70% and 80%) with customers is still phone calls. So the voice contact centre is not going away anytime soon.

What does need to go away is the executive attitude that is happy to pay for a blonde receptionist, leather couches and fancy Sandton offices, while not worrying about customers getting stuck in phone queues with nasty music on IVR technology, says Hannes van der Merwe, Mitel product manager at Itec.

Rather, the CEO has to infuse the organisation with focus on customer experience right through to every touch point, says Simon Cranswick, GM of customer interactive solutions: Western Cape at Dimension Data.

Closing the loop

Getting analytics from marketing to the contact centre will need another attitude change, this time towards legacy tech.

"The reason that loop is not closed," says Staples, "is that very few South African companies have switched to end-to-end VOIP data-based systems. Even if they have analytics on customers available in their marketing department, it is very difficult at a system level to pull it into their contact centre. They attempt to fix this by training agents."

So, closing this part of the loop can mean letting go of that legacy backbone.

Then the other half remains - closing the loop back from the contact centre to the rest of the business. The huge amount one learns about customers who interact with the company can then fuel better marketing.

What's needed here is the same approach as picking up gripes on social media, contacting the customer and resolving the problem, says Cranswick.

"If an organisation turns around and changes its call centre into a contact centre, and realises it's the heart of the business, the customer's pulse, if it acts on the intelligence coming in, the contact centre becomes an asset," he says.

However, one needs actionable data to do this, says Dr Nicola Millard, customer experience futurologist at BT.

"If you can get the data coming through the contact centre, analyse it and figure out who's the best person to action it, that becomes really valuable operational data on how to improve the customer experience. That often cuts costs as well, because you're improving customer experience and cutting the cost of failure at the front line," she says.

"Frontline staff usually know the gist of a problem, but analytics puts a price on it: number of calls and extent of a problem, and tells you how much it costs you to deal with that problem," adds Millard.

With IP technology and a closed loop underway, one needs to prepare for self-service in general before even thinking of IVA avatars.

The company that offers self-service to customers really must get it right, says Cranswick.

"We'll see a proliferation of apps on mobile devices. That's wonderful, but don't leave it in a channel on its own, because then you're adding to the sins of the past."

Messy processes

Across all self-service and the contact centre, a golden thread should come through, he adds. When a customer has tried self-service but cannot solve the problem and phones in, there is a key question to ask: "In those critical moments of truth, what are you doing to continue the conversation?

“Smart companies use technology to tie all the bits together - Web chat, phone calls, call-backs - to get the context of exactly what you were doing, what your issues were, your previous interaction, your history.

"That is the trick in pulling cross-channel service together. Now the technology to do this is finally available, technology like SIP."

Another key area is cleaning up messy processes both in the contact centre and self-service, says Staples.

"If processes to provide service at a human level have not been figured out, it is virtually impossible to automate those."

When an agent only has to 'press a button' for a process to happen, that process is mostly ready to be converted to self-service so the customer can click the button instead, he continues.

That means the organisation needs to be mature enough so that business services that are ideal candidates for self-service should already be exposed on their Enterprise Service Bus (ESB). A lot of the large local enterprises have done this already, adds Staples.

Once processes are self-service ready, the next thing is creating knowledge for the self-service technology.

If the tablet user is greeted by a friendly person who speaks plain English, that won't scare him away.

Bevin Sandras, Inter-Active Technologies

This means gathering all the available information about processes, products and services into well-structured knowledge management. That becomes knowledge that self-service technology like Web chats and avatars as well as human agents can use when interacting with customers. It can take a large organisation a few years to complete this process.

For example, information in the form of really good training manuals for human agents is not yet structured well enough for self-service IVA purposes. In addition, many contact centre voice records may need to be analysed and structured properly to build out the data set, says Staples.

Other unstructured knowledge lurks in places like product information, troubleshooting guides and Frequently Asked Questions (FAQ).

Once structured knowledge is available, a side benefit can be a dramatic reduction in training costs for human agents, says Staples.

But it needs constant tending and updating, adds Bevin Sandras, GM of operations at Inter-Active Technologies.

"You have to make sure that the knowledge has a content owner who looks after it. Previously you could make information available online and let customers do what they like, but know you have to look at how customers use it, and understand where the hotspots are.

"Information that's not working for your customers creates bottlenecks in your process, and prevents them from moving forward."

Avatar at your service

If you're going to deploy an avatar in a sales-driven organisation, know that customers will expect to be understood by that non-human.

IVAs (as well as any other natural language solutions) have to get roughly 86% to 93% of their 'calls' resolved, for customers to like dealing with them, even if they offer a great experience, says Dove. (These stats for refer to IVAs designed to complete an interaction without handing over to a human agent.)

A number of key things need to be in place to migrate other self-service technology to an IVA, says Staples.

"It comes down to your data-mining and consolidation technologies, as well as willpower within the organisation to put data into a well-structured format that facilitates Web bots and avatars for self-service.

"It's a journey they need to start. There always has to be an iterative feedback loop to improve the quality of that structured data."

By next year, Sandras expects IVAs to be deployed in SA, since a number of businesses are considering these and deciding how to do it, he says. “A likely use of an IVA could be someone non-technical purchasing a tablet in a retail store, getting home and wanting to understand the device.”

"If the user is greeted by a friendly person on the device who speaks plain English, and he can just type in what he wants to understand, and get a plain English response, that won't scare him away," adds Sandras.

Customers will always have complex questions that only well-equipped humans will be able to answer. But for many people who need issues resolved efficiently and confidentially in spaces shared with others, using self-service technology beats phoning a contact centre any day. An attractive, effective IVA could be just the thing to endear a company to real people.

First published in the August 2012 issue of ITWeb Brainstorm magazine.

Have your say
Facebook icon
Youtube play icon