Last year, (October 9, 2013 to be exact) "X" company posted a blog with the opinion that self-service technology is declining. Essentially, the blog post aimed to challenge Virtual Agent (VA) vendors with a list of reasons why VAs are not the future of customer service. If these statements hold true, the following companies are wrong: Leading analyst firms, the top device manufacturers (Apple, Google, etc.), IBM and its Watson division, and more.
Let's take a look at the list of reasons why said company disagrees with Virtual Agent technology, and set the record straight.
Claim 1: Virtual Agent setup takes months since questions and answers are manually added, managed and incorporated into scripts.
Truth: Training a Virtual Agent to represent your brand takes care. A true Virtual Agent does not require 'scripting.' Natural Language Processing (NLP) coupled with an Inference Engine allows the Virtual Agent to drive the conversation. It's not necessary to write up every single possible phrase that the end user will ask your company. Patented technology takes the scripting out of VA creation. Further, a typical VA can be ready to go in 6-8 weeks.
Claim 2: The accuracy rate is not as high, the tool does not learn anything automatically, etc.
Truth: Traditional Artificial Intelligence (AI) learns on its own. AI is at the core of a true VA. The reason why customers do not opt for automatic learning is the same reason why you wouldn't let call center agent learn on his own. Similar to your best agents, VAs require care, and training. For example: A company in the PC industry uses a VA to solve issues related to BSOD or Blue Screen of Death. The root of this problem can be related to multiple options. Is it a driver? Is it a new installed software? Is there a single solution that can be learned on a whim? Tools exist in the back end to 'turn on' automatic learning. Rote memory watches end user behavior, and provides the most relevant solution based on its confidence level. In addition, accuracy results are based on statistical analysis. The reporting is conservative and based solely on votes from the end customer. Virtual Agents prove their success by deflecting up to 50% of calls when positioned
correctly and achieve (on average) 85% success rate.
Claim 3: The cost to create and manage answers is high(er) and could be up to $1 per contact because of the amount of manual work needed.
Truth: noHold reports on this metric for many of its customers. Specifically, a Fortune 500 company in the networking industry reported that the cost per contact is less than the price of a gumball for each session or conversation with the VA. Another company in the PC industry was within the same range.
Claim 4: Conversations, questions, and answers must be pre-written, pre-programmed according to a certain predicted script.
Truth: As mentioned in claim 1, there is no scripting.
Claim 5: VAs do not support automatic translation.
Truth: Yes, they do. The reason most VA vendors do not typically offer 'automatic translation' is for this reason: Global languages are structured differently. Automatic translation is great, and easy to do however, lacks accuracy and does not take the NLP into account. A true NLP looks at the roots of words and how groups of words are related. In order to translate in this environment, tokenization and lemmentization are key. According to online resources, tokenization is useful both in linguistics (where it is a form of text segmentation), and in computer science, where it forms part of a lexical analysis. In response to Lemmatization: In many languages, words appear in several inflected forms. For example, in English, the verb 'to walk' may appear as 'walk', 'walked', 'walks', 'walking'. The base form, 'walk', that one might look up in a dictionary, is called the lemma for the word. The combination of the base form with the part of speech is often called the lexeme of the word. Lemmatisation is closely related to stemming. The difference is that a stemmer operates on a single word without knowledge of the context, and therefore cannot discriminate between words which have different meanings depending on part of speech. However, stemmers are typically easier to implement and run faster, and the reduced accuracy may not matter for some applications.(wikipedia.com, 2014)
Claim 6: VAs are usually placed on the home page or certain landing pages as it is too complex to create different conversation scripts for each page. As a result, VAs are used mostly for lead generation or routing.
Truth: Virtual Agents are light weight, and again do not require scripting. A single Virtual Agent can be diagnostic and interactive (offer troubleshooting) for complex issues, and at the same time, be conversational. Many trusted brands position the same VA on a Facebook site and community forums that they use on support pages and contact us pages.
In summary: According to respected analyst firms, as well as trusted brands, the Virtual Assistant market is expected to grow exponentially. Whether for customer service, technical support, or as a butler for your device, adoption is undeniable. A true Virtual Agent uses NLP, and can troubleshoot issues just as well as it can answer conversational questions. Powerful reporting tools empower award winning brands to listen to the voice of the customer while providing a snapshot that can be shared company-wide.