Where do chatbots belong in your customer care toolbox?

Chatbots, digital and virtual assistants have had their fair share of technology hype. But why haven’t we seen the wide adoption of them across many industries?  

There are of course many reasons that users and decisionmakers were disappointed by the early results. Chatbot technology has evolved massively since the early attempts in 1960s and first mass market implementations in 2011 with Apple’s Siri. Still, many myths surround chatbots, and these probably have something to do with continued underinvestment in the development and deployment of such solutions.

In a recent blog post, the Technology and Services Industry Association (TSIA) debunked eight myths about the use of chatbots in B2B setting. These misconceptions have certainly contributed to companies not paying enough attention to chatbots. A widespread myth claims that customers don’t want to use bots because they don’t expect a bot to get the right information, perform the right action or be as efficient as a live agent would be.

At the dawn of chatbot adoption that might have been true, but both the technology has evolved since, and the customers are also more open to instant chat interactions, rather than engaging in long phone conversations.

The other false impression is that simple search options  can provide alternatives to complex chatbot systems. While unified search engines can perform certain actions based on single-word or short phrase queries, users have higher expectations and more complex needs to find the right information that helps them resolve issues than a limited-input, instant-response search can offer.

Chatbots integrate a conversational experience, enabling two-way communication between the users and the AI system. This improves the accuracy and efficiency of the interaction via incremental user inputs. An AI-based chatbot can follow a logical path and narrow down options, rather than rely on trial and error.

So, if you are convinced a chatbot is a way to solve your challenges in offering guided technical support, there are a few things to pay attention to, as we have learned in our development process.

Based on the last few years of offering and improving the Nokia Digital Assistant, we have identified three critical areas that significantly impact the success of a customer service chatbot solution.

Does your chatbot get the right data?

The amount of information available has been growing at an exponential rate. According to R. Buckminster Fuller, the rate of data duplication until 1900 was once every century, approximately 40 years ago it was doubling every year, and now, based on the latest estimates, we are doubling the volume of captured data every 12 hours. Especially in the case of innovative technologies such as 5G, we can expect that exponential pace to continue.

But do all those data points transform into structured information and useful knowledge? Keeping up with the exponential data growth is a major challenge.

In the context of providing technical support based on best available information, you need to figure out a way to build, maintain, curate, and evolve data sources. And to prepare for further growth and plan for future applications and use cases, you need to plan for the integration of additional data sources and formats.

The Nokia Digital Assistant structures data from over 20 data sources covering all Nokia technologies, anything that support engineers would need to deal with. These data sources go far beyond product information and include technical notes, product and parameter information, and live network data. We have prepared our solution to ingest any type of dataset, so we can introduce new information sources and formats. This capability is not limited to Nokia's own sources - we integrate customer data, creating a personalized knowledge base of network related information.

Does your chatbot have to talk to be effective?  One bot or many modules?

Building on good data foundations is critical, and so is having an intuitive and responsive user experience.

Enabling effective data extraction is key to user adoption. Support engineers will regularly return to a user interface that is easy to manage and built in a language of their choice.  

Chatbots don’t necessarily need to talk. Imagine having to talk in a noisy environment, repeating technical information and error or product codes multiple times. While the popular chatbots were initially focusing on voice interaction to drive ease of use, clearly understanding users’ requests is much more important. A bot needs to be effective before it is efficient.

How did we address these usability aspects? We have built an intuitive system that runs in any modern web browser and on major mobile platforms so users can interact with it as they please. Overall experience delivered to the user is sustained by conversational components.  Instead of building a monolithic, massive chatbot, we designed a combination of smaller, modular components. Each software component of the bot can deal efficiently with different information sources and functionalities. This modular approach allows Nokia Digital Assistant to expand its knowledge in a scalable and faster way, lowering the maintenance effort and ensuring future scalability of the system Users get more features, in much shorter time.

This brings me to the final point – learning by doing. And by that I mean the bot’s machine learning algorithms learning from repeated interactions with users. The more support engineers converse with the chatbot and put it to the test, the more conversation flows get added, the better the system becomes.

Activate your user community

An active and responsive user base can significantly improve your chatbot solution. User curated data plays a huge role in online content, and for a good reason. There’s a general misconception that we always know what the user wants to do, that we can script it. This might lead to a few satisfied users, but you’ll be missing out on constant feedback and improvement loops that can benefit the whole community of users.  
We started rolling out the Nokia Digital Assistant in phases, reaching out to more and more users, while enabling them to use the newest features. This has contributed to the global adoption of the Nokia Digital Assistant. The digital assistant became commercially available just after the COVID-19 pandemic hit in early 2020, since then over 150 CSPs have deployed the digital assistant, and over 1300 active users log in regularly.

Our user surveys show 80 percent positive feedback and we continue to get regular user ratings based on conversations conducted with the Nokia Digital Assistant. The efficiency of searches is also a good way of evaluating a chatbot. A survey conducted by SearchYourCloud revealed that ‘Workers took up to eight searches to find the right document and information’. Based on our measurements of user interactions, the Nokia Digital Assistant is able to reduce the number of searches to fewer than five – a reduction of about 40 percent.

But the best feedback of all is to see our users coming back over and over again! You could be one of them. If you currently have a software support services contract with us – the Nokia Digital Assistant is included in the benefits. Reach out to your dedicated Care Program Manager to make sure you’re taking full advantage of the chatbot’s functionality.