3 challenges facing virtual assistants

The upper echelon of chatbots, in terms of A.I.-powered intelligence, virtual assistants (VA) have become a hot topic in customer experience.

Presenting a significant opportunity to increase revenue and cost savings, virtual assistants:

[aditude-amp id="flyingcarpet" targeting='{"env":"staging","page_type":"article","post_id":2038833,"post_type":"guest","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"bots,","session":"D"}']

• Extend the reach of customer interactions
• Reduce live agent workforce needs (But they DON’T eliminate live agents — we still need them to oversee and train the A.I.-powered agents.)
• Boost agent productivity and reduce Average Handle Time
• Optimize digital performance (insights gained through analyzing transcripts provide rich visibility into self-service improvement opportunities.)

Despite the obvious benefits, though, roughly only 1 in 10 enterprise brands in the U.S. has deployed a virtual agent. Why haven’t more brands jumped on the VA bandwagon?

For starters, many companies have discovered the unique challenges that frequently limit VA experience and success. Let’s look at a few of these challenges and see what can be done to overcome them.

Challenge #1: Lack of accuracy and personalization

When a VA is not able to answer questions accurately, it’s because it lacks the proper context or doesn’t understand the intent of the question. 

Solution: To enable engaging conversations that result in an effective self-service resolution, the virtual assistant must:

• Allow the consumer to use natural language; this includes slang, typos or misspelling, synonyms, unstructured content, or complex grammar
• Reference context from the entire conversation
• Personalize responses to the user based on context (account balance, navigation history, search term, etc.)
• Troubleshoot using conversational strategies (clarifying questions, incomprehension strategies, disambiguation strategies to clarify customer intent through related answers, etc.)
• Support common social responses
• Guide the consumer through an appropriate journey with step-by-step assistance
• Integrate with backend systems to access pertinent account/product information and complete transactions

The VA’s ability to answer questions relevantly only happens with rigorous optimization, involving both humans and machine learning. Continuously ensuring solid quality control strategies will also help manage the risk of the VA learning undesired bad behaviors. Bottom line: virtual assistants should not be designed as a set-it-and-forget-it solution. 

Important: Brands should continue to collect data from all visitors but be careful not to treat their customers like a number. Furthermore, they must ensure that their automated strategies are connected with their live agent strategies.

Challenge #2: Lack of behavioral targeting

Virtual assistants tend to be implemented with the expectation that they can be everything to everyone, but they will have difficulty engaging the consumer with the right experience if not given behavioral targeting rules. 

Solution: Analytics and business rules help companies understand customer intent. Some of the areas of customer behavior analyzed are:

[aditude-amp id="medium1" targeting='{"env":"staging","page_type":"article","post_id":2038833,"post_type":"guest","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"bots,","session":"D"}']

• Page or content area viewed
• Time on page, content area, or site
• Previous engagement on site
• Campaign lead source
• On-site qualification
• Opt-out activity
• Visit to site
• Search activity (On/Offsite)
• Cart value or contents
• User-entered value
• Form field activity
• Site abandonment
• And many more 

Key to successful real-time customer segments targeting:  a sophisticated intelligent rules engine that leverages real-time analytics and business goals to choose between live agent, guides, or a virtual assistant on a per conversation basis — based on user behavior, profile, and live agent availability.

Challenge #3: Flawed and untested design choices

There are several types of virtual assistant designs in the marketplace, with brands implementing a style based on the purpose it will serve. In advertising, a human face increases attention and engagement. In social settings, assistants with anthropomorphized features increase human engagement, likeability, and trust. For contact centers and sales, personalizing and humanizing the engagement increases customer satisfaction and success rates; therefore it’s common practice for VA design to personify the VA with an avatar.

However, chat A/B testing has consistently demonstrated that having no avatar or human face visible during the conversation is actually the best practice; studies show that an image is distracting and leads to negative influence resulting from user bias toward the image. What really matters for users is humanizing and personalizing the experience through the right content to increase success rates and customer satisfaction. 

Solution: Once a brand decides on its design and conducts A/B testing, it needs to keep in mind that, regardless of the design, content is the key to successful communication.

[aditude-amp id="medium2" targeting='{"env":"staging","page_type":"article","post_id":2038833,"post_type":"guest","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"bots,","session":"D"}']

Overcomers win

The truth of the matter is that well-designed and optimized self-service engagements lead to substantial cost savings and customer satisfaction, while poorly executed self-service engagements increase expense and create customer frustration. Brands that have overcome the challenges of automated customer experience integration saw virtual assistant use jump from 28 percent in 2012 to 58 percent in 2015. This statistic will only grow as the technology progresses and more companies take that important “virtual” step in digital customer experience transformation.