May 6, 2021
If you read literature about customer support, especially as it relates with self-service support, you frequently find the expression “user intent”. But what is a user intent? We define it as the objective that a consumer wants to achieve when performing a search on the internet, browsing a website, or contacting a company service department. With the rise of automatic systems to let users self-serve in their relationship with companies, how intents are discovered by the system and managed has become paramount.
There is no doubt that having oneself understood quickly when looking for something is important, but typically there are many ways to express a need, a complaint or a desire and people will use them all (even assuming they know exactly what they are looking for, which is not always the case).
So, one of the biggest challenges for automation services is to map these expressions and identify the real motivation behind them.
Of course, when calling a company, users could be directed to speak with a human agent who will quicky determine the user’s need, but this is expensive and does not scale, so automated systems have been available for decades to “qualify the call”, discovering the user intent, and route the call to the most appropriate service. For many years this meant menu-based systems interacting with users through tones, but more recently Artificial Intelligence systems have become more and more able to converse with people, reducing the steps necessary and delivering a much more pleasant, agile, and accurate customer experience.
So, interest and investments in Conversational AI able to discover the user’s intent are increasing significantly in the various sectors in which it operates: from the development of algorithms and intelligent systems to advertising and Inbound Marketing strategies. But how, in fact, does the intent discovery work, how can this technology contribute to serving your company? This is what we discuss here.
Last week I watched a webinar and demo by a company providing tools and solutions for...
Talking with ChatGPT over the phone is cool. But we can also make it useful. In the past few...
How do intelligent systems identify intents?
When a user reaches an intelligent conversational virtual agent, the experience is very different from an old IVR, although the objective of the first part of the call is the same – identifying the caller’s intent. A conversational AI Virtual Agent will start the conversation with an open question, something like “Good morning, you have reached Company X, how can we help you?”. Users are thus free to express their need in any way they like.
When they speak, their sentence is first transcribed from voice to text by the Virtual Agent, then sent to a conversation engine for analysis. Engines can be of different types, but all of them compare the sentence with a knowledge base of possible requests, related to the capability of the service. Of course, the number of possible intents is not infinite and in fact they are the same that can be served by an older IVR system. So, the “domain” on which the Virtual Agent searches for the intent is limited, and this facilitates the search.
It is possible that the caller has already specified all that the system needs to know to correctly identify the intent, but very often this is not the case. For instance, a system to automatically book an appointment will need to know the name of the user, the type of appointment, the location, the date, and the time. No-one would say all this in the initial sentence. But a well-designed Virtual Agent will be able to narrow down the possibilities gradually to get to completely identify the users’ intent and service their need.
One advantage of Virtual Agents over old IVR systems is that the pieces of information can come on any order. So, continuing with the appointment booking example, the user may say “I would like to reserve an eye doctor appointment”, and the Virtual Agent can then ask them what day they want it, where, and what is the preferred time (assuming that the user’s phone number is already in the database and so the system knows who it is talking with). But the user could also say: “I need an appointment tomorrow”. In this case the Virtual Agent would reply: “what type of appointment? We cover ophthalmology, dermatology and radiology”, and then proceed to collect the rest of the information.
The conversation will continue until the Virtual Agent has collected all the information necessary to provide the service, or if there are complications, at least the Virtual Agent will be able to forward the call to the correct human agents – in this case, the ones serving the Ophthalmology department.
The advantages of an autonomous voice service
The ability of Virtual Agents to identify intents quickly and precisely provide several advantages for companies, especially ones with a high volume of customer interactions and several departments. To start with, it is not only the intent that the Virtual Agent identifies, but also what it is called “entities”: the pieces of information that make providing the service possible. In the example above, the intent is the type of appointment that the user seeks. Entities are instead the date and time, and location. Having the complete set of information often enables the Virtual Agent to complete the service without contacting a human agent, thus saving time and money.
Even when the service is not provided completely by the Virtual Agent, interactions are routed to the correct human agent queue with a much higher precision, greatly reducing the percentage of calls that have to be transferred to another department. This also saves money and time, not to mention providing a better customer experience.
Finally, the ability of Virtual Agent to collect most of not all the information necessary and transfer it to human agents together with the call also helps keeping the duration of calls shorter and save money.
Other advantages include the ability of the service to scale to meet demand, much faster than what a human agents-based contact center can scale. If a service peak is coming, due to the season or a scheduled event, or even in emergencies, it is easy to just increase the number of Virtual Agents that come in perfectly prepared and trained as the ones already in use. This will buffer the traffic increase on the “real” contact center, as a high percentage of the peak calls will be resolved in self-service mode and the peak on human agents will them be smoothed.
Virtual Agents also remove the limitations of day and time, since they work 24 hours a day, 7 days a week. Continuous service prevents demands from being “dammed” at the ends of the week and contributes to customer satisfaction, as most of their needs are met at any time and without waiting on the phone.
Virtual agents acceptance
Virtual Agents of all types are becoming more and more common, and thus accepted by the general public. People are increasingly used to controlling computer services by voice, from smart speakers to search engine searches to interactions with virtual agents over the phone. So, Virtual Agents are accepted by the public immediately, and gladly. A conversational experience not only contribute to a more human service, but also make the consumer’s routine more practical. Many problems can be solved without the customer having to click on a single button.
Understanding of the customer journey
For a company to be able to provide an excellent customer service, it needs to know its customers and their journey, while seeking support – the path taken from the first contact to when their need is met. This understanding allows organizations to deliver exactly what their customers are looking for at each stage of the journey.
Until the recent past, it was not possible to obtain this knowledge with satisfactory precision since the available data was very limited, especially on the telephone channel. The best companies could was to record calls and then select a sample to analyze, which was expensive if they wanted a more complete picture or necessarily incomplete.
A Virtual Agent however works on text, and so all calls are transcribed. This allows to use text analysis tools, also based on Artificial Intelligence, to monitor consumer behavior in detail and gain valuable insights.
About Interactive Media
Founded in Italy over 20 years ago, and with offices in Brazil and the USA, Interactive Media is at the forefront of Conversational AI technology and processes hundreds of millions of customer service conversations a year with its Virtual Agents, in different countries and languages.
Now that you understand the importance of the user’s intent, it is time to see up close how Virtual Agents interpret and use it in service. Get in touch with us and get to know OMNIA, our complete solution for the development, deployment, training, management and monitoring of Omnichannel Virtual Agents.
Last week I watched a webinar and demo by a company providing tools and solutions for conversational customer service. Interactive Media, where I work, is in the same sector and I wanted to scoop out a competitor, see what they have and how they are presenting their...
Talking with ChatGPT over the phone is cool. But we can also make it useful. In the past few months tech people worldwide have been talking almost exclusively about Open AI’s ChatGPT. It’s the first large language model chatbot to make a splash, and what a splash! It...
As the market for smart speakers falters, what are the Big Three (Amazon, Apple, Google) going to do? Alexa, should I bring an umbrella out tomorrow? This is a question that owners of smart speakers have been asking since 2013, the year when Amazon released its first...
Interact with us
Receive our exclusive content: