The other day, I was on my bank’s mobile app to view my recent credit card activity when I noticed an icon at the bottom of the screen that I hadn’t seen before. It was a microphone with “Voice” below it. When I hit the button, a “Siri-like” voice asked me to speak or type my request. Either choice brought up my recent activity in just two steps. My new “assistant” made my service much better!
Self-service in contact centers has been around just about as long as contact centers themselves. IVR and web-based applications support simple, routine transactions and answers to frequently asked questions, thus deflecting calls from (expensive) live agents. A new breed of virtual and live assistants has entered the self-service domain to make these interactions more robust and successful, increasing usefulness to customers and the center. This article shows variations in “assisted service” and the impact they are having (or could have) on the customer experience.
Virtual assistants (VAs) in the form of visual “avatars” (animated human faces or figures) can provide a captivating and interesting alternative to frequently asked questions (FAQs) or searches on a company’s website. They use natural language text analytics, which allow customers to type a free-form question instead of choosing from a list. The VA replies with answers, points to links on the web page, or asks more questions to arrive at the right topic or answer.
VAs typically interact in text and present visual answers. Some applications offer the option to hear the avatar speak the information. The avatar may be static, or show gestures and different facial expressions when the query is not understood (concern) or when the correct response is given (happy). Some even use humor to engage customers. Developing the name, image and voice of the avatar to offer the right representation of a brand and approach to customer interactions can be nearly as important as the technology and information behind them. For a little fun (and a great example), go to www.goarmy.com and meet Sgt Star®, the U.S. Army’s virtual guide who makes finding answers on its website much more inviting, with a dose of culture built into the learning process.
The growth in adoption of VAs is exploding. Gartner projects that, by 2015, at least 1,500 large organizations will facilitate half of all online customer self-service searches through a VA. Companies are having great success with this technology. Aetna states that its avatar-driven VAs have reduced calls to its technical support by 29% (Forrester TechRadar, Q2 2013).
But does just having an avatar increase the accuracy of the answers given? Of course not. Like any technology, careful planning and appropriate investment of time and resources yield results. Successful implementations focus on limited conversations specific to the organization’s products or services. Moreover, a good avatar is highly dependent on the structure, content, accuracy and relevancy of your knowledge base (KB). If you cannot provide useful answers to your customers, all the money spent on cute and interesting avatars will not reap rewards and, even worse, may erode your brand. Beyond the KB, you need some intelligence built into the interpretation of the question and natural language text analytics. Not everyone asks the same way, and many use poor grammar or shorthand. Figure 1 shows the steps to accept a customer inquiry (via a variety of channels), interpret the need, and resolve it through self or assisted service.
Here are some additional considerations which may increase your success in deploying VA technology:
- Give customers a choice of topics or short FAQs in addition to the free-form entry option.
- If the VA misses several questions, offer escalation paths to a live agent via voice or chat.
- Don’t get carried away with trying to make your avatar appear too human (like a video). Keep it light. Avatars with facial expressions and gestures reveal a personality and tend to keep customers’ attention better.
- Choose a system that allows you to turn off the audible avatar voice. Get feedback from your customers and decide which medium of information dissemination (voice or text) is better received.
- Tune the application, including the interface and the outcomes; test and seek feedback constantly.
Done right, VAs can capture customers’ attention and lead to a better overall assisted service experience. And, these techniques can be translated to your other self-service channels.
The proliferation of smart phones is prompting companies to deploy advanced features in their mobile applications, including voice-assisted service. Some organizations are even incorporating the same avatar from their website into their mobile apps, leveraging the same interface and KB. “Nina” was the helpful voice assistant who I mentioned at the beginning of this article. It is a system from Nuance that is currently utilized by USAA, Geico, U.S. Bank and others to provide smartphone-assisted service similar to that of iPhone’s Siri. Most companies using this software are in the financial services and insurance industries, and many are still in the testing phases of deployment. If your mobile app is mature and has achieved measured success, this may be your next opportunity. Keep in mind that the same dependencies on the KB and intelligence exist for your mobile app, along with the added challenges of natural language speech recognition when allowing voice interactions through the mobile device.
Video Agent Assist
Video has beckoned for the customer-facing contact center several times over the last 20 years, but has never taken hold previously. The struggle has been to overcome obstacles to adoption—for the center and the customer. To start, few centers are well-suited in their design and facility structure, not to mention dress codes, distractions, etc. In addition, many customers either don’t have video capability (or know how to use it), or prefer not to interact on video. Corporate IT gets nervous thinking about the bandwidth and security impact on its network. However, one-way video has come a long way and seems to be gaining traction.
Amazon now has the “Mayday” button on its Kindle Fire to connect with a customer service rep via one-way video. The agent can draw on the customer’s screen to provide directions and even take control of the device to fix a problem. Similarly, in the UK, a potential Apple customer can select the “Ask Now” button when viewing iPad or iPhone products online, and an “Apple Specialist” agent appears on a video window, along with a prepared presentation of the product. Customers purchasing the product through the Specialist can schedule a free online session to help set up the new device.
As big-name companies begin to use (one-way) video, we could see this as a “game changer” that enhances customer expectations and comfort with the technology. Using video can create high-value contact and connections that would otherwise be missed. Think personal bankers in financial services, disease management counselors in healthcare, and personal shoppers in consumer products. Consider piloting video assistance into your web and/or mobile app if you see benefits for your business.
Web collaboration is making a comeback. In its improved state, web collaboration uses subtle approaches to coach rather than control, coupled with technology advances that make it more useful than ever. Mature solutions have overcome compatibility issues between competing browsers, and many no longer require software downloads in order to function. Some applications require a security code that the agent provides so that the customer is confident that the agent (and nobody else) is viewing or accessing their screen. Furthermore, certain pages or fields on the customer’s screen can be masked so agents can’t see or control them. Finally, some technology providers have offerings that require little or no code to implement the service, making it easy to deploy.
Here are three co-browse methods:
- Over-the-shoulder: Customers click a “talk to an agent” button on the website, which initiates a phone call between the customer and an agent. The agent is able to view the customer’s web page and cursor. The agent highlights sections of the web page on the customer’s screen but does not take over the customer’s cursor. This method is good for sales.
- Watch me: After connecting as described above, the agent clicks on “show my screen” to enable the customer to view the agent’s screen. The agent demonstrates the site navigation so that he or she can get to the right place. This method is useful for customer service.
- Take over: The agent clicks a button to view and control the customer’s page. Once the customer “accepts,” the agent can drive the customer’s desktop (not just the web page). This application is most appropriate for technical support.
Co-browsing has been around for over 10 years but is still considered a niche technology. Watch for offerings and adoption to increase. If you are new to web collaboration, try the “over-the-shoulder” or “watch me” methods first. These are more comfortable for your customer and a more likely fit for most situations.
Behind-the-Scenes: Live IVR and SMS Virtual Assistants
Want to make your speech recognition and SMS interactions more effective? Some vendors have increased accuracy by incorporating the most basic technology: humans, with the added benefit of more personalized responses. Here, we highlight two technology firms that add human oversight to assisted service.
Interactions Corporation (www.interactions.net) has a unique way to improve the accuracy of natural language speech recognition in IVRs. Their patented technology uses a “proxy” device between the IVR application and the speech engine. It determines whether the customer’s spoken phrase is simple or complicated. If simple, it passes it to the Automatic Speech Recognition (ASR) engine. If complex or likely to result in low match rate, it sends the phrase to a live agent (Human Assisted Understanding) to reply to that phrase. The human agents sit behind the scenes, interpret the spoken statements, and push prompts back to the customers. They do not speak to or have direct contact with the customers.
Text4Assist (www.text4assist.com) uses similar analytics for self-help using SMS text to the call center. The company launched in Europe, but is developing applications for North America. Text4Assist has spent a considerable amount of development time working with clients to integrate with CRM systems and develop the algorithms to determine whether the SMS KB engine or human assistance should respond to the customer’s text.
These technologies rely on the same concepts described earlier: good natural language processing and a robust KB. You also need to consider if they will support information interactions or transactions, and the implications for identification and verification (ID&V). The IVR example has high value in determining a customer’s need and routing to the right agent, potentially providing some information, as well. In either case, the opportunity to support transactions could be greater if the customer is already logged in to a specific account (ID&V is already complete).
These new solutions are a hosted/managed service of sorts, and can have different pricing models than traditional technology solutions as a result. The “human” side of the equation can be staffed by the client, or through hosted third parties. The cost could be based per transaction, per-minute or by success rates (including customer satisfaction results), for example.
Assisted Service Done Well Increases Brand Loyalty
Contact centers are moving beyond “self-service” into a more nuanced “assisted service” concept, which brings entirely new ways of looking at human assistance. Done well and targeted to the right audience, avatars, mobile assistants, video and web co-browsing provide a more satisfying customer experience. These technologies are making self and assisted service more available and useful to the customer. They can also benefit the organization through both cost savings and revenue increases. Virtual agents cost less than “real” agents, and a happy, self-assisted customer increases brand loyalty.
Be sure to carefully examine the tools and market providers, and deploy the right assisted service technology for your industry, customers and environment. As is true of all new technologies, develop a solid business case, run a comprehensive pilot and measure success prior to complete commitment and adoption. And don’t worry, the IVR is not going away!
– Reprinted with permission from Contact Center Pipeline, www.contactcenterpipeline.com