How can businesses progress in chatbots adoption?

We are talking about a paradigm shift in human-technology interaction while referring to artificial intelligence based chatbots and virtual assistants. This is definitely true, however there are considerable constraints in adoption of these new interfaces by the masses. To overcome these constraints, Gartner suggests technology product managers to develop a task-based approach.

Gartner published a report called “Market Insight: Chatbot and Virtual Assistant Adoption Will Only Progress by Delivering Value and Trust” in 2018 focusing on the aspects that the product managers should consider to increase the adoption of their chatbots. In this blog, we will summarize this Gartner Report to help the businesses build their conversational AI strategies.

First of all, it is important to share an assumption by Gartner that by 2022, 16% of all consumer tasks will be fully automated, up from 6% in 2018. This automation, of course, will be enabled by the power of artificial intelligence. Gartner underlines that artificial intelligence is not a technology as it is widely assumed, but it groups a broad range of technologies, like semantics, machine learning, deep learning.

Do people know that they use AI-based services?

Within the first 5 years, we have witnessed many failures or poor conversational experiences because of varying reasons that diminished the engagement and adoption of the users in many sectors. Considering the phase that we are going through in terms of conversational AI, this is quite normal and expected. In addition, we need to take into consideration that people use services that are enabled by AI technologies without knowing it. In a recent Gartner survey on consumer perceptions of AI, 61% percent of the respondents use the services enabled by AI but not know what AI is or is not aware of that AI is the technology behind that service. The most common examples of these services are – using a chatbot, a virtual assistant, a driving route app, or getting music or shopping recommendation (“How Consumers Want Your AI to Interact With Them — Interfaces, Trust and Autonomy”).

Why do people use them? 

According to the report, the answer is simple, just because these services create an explicit value to them, making their lives easier. So the key for the user’s adoption is value and trust. If AI used in services deliver value to the user, then the user is more willing to give permission for greater automation and personalization.

Gartner suggests the product managers to clarify automation opportunities by examining the complexity and value of user tasks. For this, the report defines two dimensions that shape the approach to automation and use of chatbots: (1) complexity of the task (2) the value achieved by the task. 

Source: Gartner (August 2018)

In the table above the tasks are categorized within 4 groups based on their complexity and value to the user. The product manager can use such a table to make a clear roadmap for their chatbot strategy. It does not necessarily mean that every quadrant requires a distinct chatbot, but companies evolve the capabilities of a chatbot from one quadrant to another. 

What are the crucial points to be considered?

1- Develop chatbot trust and automation focusing on groups of tasks

There are many companies and platforms that create chatbots, most of them assert to be AI-based, using machine learning and deep learning technologies. However, most of the chatbots that we see around are rule-based or menu-based ones. The first rule is to be transparent. The chatbot you build has to deliver what is promised. If the chatbot is promised to be AI-based, it has to use AI technologies, for example allowing a free-flowing conversation. Announcing that the chatbot is a rule-based does not make it insufficient but on the contrary, shows the honesty and transparency in terms of the company’s approach. Gartner says that the customers really do not like marketing words, but prefer hearing the honest limitations of the product or channel that they use. This is much more relevant for the case of chatbots as it is obvious that this field is still improving. 

Gartner suggests that the easiest way to create a chatbot experience is to start with the “routine” low-complex, low-value tasks, such as grocery shopping. One of the most implicit trust element is sharing of data and the need for sharing data is minimum for these routine tasks. Gartner says that the users in emerging markets are more willing to share their data than the ones in the mature markets. If the user can get better experiences in these routine tasks, the level of trust increases. However this also requires well-defined data policies of the company to avoid data breaches.

2- Develop a segment based strategy as the automation sensitivities for users vary by use case, persona and location 

A recent Gartner survey taken in the U.S. and the U.K. showed that AI preferences for automation varied depending on (1) Task (use case/context) (2) Persona (attitudinal and demographic elements – “Survey Analysis: Five Digital Consumer Personas Critical to Understand Tech User Behavior”) (3) Location (culture) 

Therefore, the product managers should consider every segment of customers differently in terms of their propensity for automation. In general, people accept AI support for low-complex but high valued tasks. In terms of the persona, early adopters are less likely to automate complex tasks than later adopters, given their technology savviness (Survey Analysis: Five Digital Consumer Personas Critical to Understand Tech User Behavior). Culture is another parameter for trust. People in emerging markets are more likely to share data than those in mature markets. (“Survey Analysis — Balance Privacy Control to Improve the VPA User Experience”). So, the product managers should clearly define the segment that they are aiming and the value that they are offering for this specific group. 

As Cbot, we also suggest companies to start this conversational business journey by defining limited use cases or segments. After a certain level of experience, it is more convenient to enlarge the functionalities covered or the segments aimed. 

3- Consider the different rates of automation adoption for task 

It is explicit that messaging or chatting is a huge communication trend but, it has to be accepted that chatting with an AI chatbot is not same with chatting with a person that you know very well. Gartner says that given the pervasive use of chat applications, we would expect users will adopt chatbots as the next stage of the chat evolution. So it is important to know the likelihood of automation of the groups of tasks in the table above. Gartner found out that in mature markets there is a higher potential of automation for lower value tasks and less potential for higher value tasks. However, respondents from China and India were for greater automation across a wider range of tasks.

4- Delivering a simple, well-designed experience matters 

The product managers should honestly announce the boundaries of their chatbot and position it as an employee of their company. The chatbot has to be trained properly with the domain specific data and should improve itself over time. As the tasks get more complex, a multimode interface involving chat, voice, conversation, video or photos will leverage the experience. Therefore, designing the user experience, empowered with the necessary tools and features, aligned with the purpose of the chatbot is crucial importance for the adoptability of it by the users.

Gartner asserts that maturation in the aspects described below helps the chatbot deliver a better experience.

1- The orchestration of more than one specialist bots 

2- Being embedded at the edge of the devices for the speed and availability

3- Context awareness for better intent addressing 

4- Being able to handoff to a human

5- Using NLG (Natural Language Generation) to cover all the variations of the intents

6- Providing multimodal conversational experiences including voice, keypad, gesture and image

This is another point that we fully agree. We definitely think that customer experience matters, sometimes more than the technology. For an AI company, providing a simple, natural and seamless customer experience is something totally different from developing the most advanced technology. Therefore, the companies should focus on the customer experience and ensure that their vendor has the capabilities to enhance it.

Conclusion

Chatbot use shows us more or less people’s general approach to AI use. To increase the trust and adoption of this new interface, the product managers should consider certain aspects. Focusing on the group of tasks, developing a segment based strategy, taking into account the different rates of automation adoption and delivering an enhanced user experience are the most crucial points. We believe that trust will continue to be the most important parameter of adaptability as AI-based chatbots enlarge their functionalities and use cases. 

Let’s discuss about the report!