When I published my last post, many readers were asking me to provide more details about Conversational Interfaces (CI). Therefore, today I want to go through some basic concepts about CI, highlight the main differences from other chatbot approaches like NLP or Voice UI. Finally, I will show some interesting examples.
As the bot market has passed the stage of hype and started to mature, many people realize that Chatbots are not going to replace Apps anytime soon.
Sure there are some early successes, but overall I think the industry is still trying to find the killer app/bot. (Though we don’t want a bot that can actually kill.)
Many companies are experimenting with different approaches to take chatbot to the mainstream. We can observe three major trends:
From NLP to NLU
One of the main point that Facebook uses to promote Chatbot was that Natural Language Processing (NLP) technology was good enough to understand all kind of user request. However, if you have interacted with a chatbot you know, it´s far from true.
Hence there is a new trend to evolve from NLP to NLU Natural Language Understanding. Many tech companies are throwing lots of money and resources to develop hard techs like reinforcement learning, AI negotiations capabilities to allow machines have a better understanding of user intent.
Voice Interface is Hot
There is a lot of criticism about how Chatbot adds more taps and friction than conventional UI; consequently, it’s not convenient for users.
Like Golden Krishna stated, “the best interface is no interface,” many people are considering voice interface as an excellent approach to reduce friction of Chatbot.
Furthermore, with the proliferation of Smart Speakers like Amazon Echo, Google Home, Apple Homepod the use of voice is starting to become a new trend.
Hybrid Approach with Conversational Interfaces
As companies realize that NLP has still a long way to go, there is a new approach to solving problems from UX perspective. Instead of relying only on natural language interaction, with CI we can have a richer and more dynamic user experience.
In this category, the best example would be the bot platform of Slack, where they are betting big on UI elements like images, interactive buttons, and message menus. Facebook is also betting on the interface side by using web views inside of a messenger chatbot.
Let’s dig deeper into Conversational Interface. First of all a simple definition:
A CI is a hybrid UI that interacts with users combining chat, voice or any other natural language interface with graphical UI elements like buttons, images, menus, videos, etc.
The Benefits of Using a Conversational Interface
Increasing user attention
In today’s world users have less attention span than a goldfish, they are easily distracted by all sorts of things like mobile apps, emails, slack notifications, etc.
In a standard GUI, users receive all the information at once and are usually confused by multiple inputs.
What´s interesting is that in the case of CI, information is provided progressively under user´s command. It also provides one clear call to action for each user interaction with the system. In this way, we could increase user attention and provide information only if needed. Consequently, we will get better user engagement.
Less user frustration
As I mentioned before, NLP technologies are still in their infancy. On the other hand, users have a very high expectation in AI (which in part is due to those Hollywood sci-fi movies we saw in the past xD).
Therefore users have very low tolerance about the error rate a chatbot.
With CI we can easily solve this problem by limiting user inputs to just a few options. Instead of asking openly what the user want, we can proactively offer him some choices so he could get what he needs with fewer taps. Some people might say that this limits user freedom, therefore, has less value. However, the reality is it’s better to allow fewer options than to piss off users not understanding their request. Also, IMHO user interface exists to simplify user’s choices so they could get what they need without over thinking.
Better Cost effectiveness
Another great point about CI is its cost effectiveness. Implementing an NLP technology is quite expensive, a vendor could charge 10k-20k for an initial implementation. What’s more, there are usually hidden costs associated with human agent support and AI training during first 6–10 months after the deployment.
In CI the price is usually a little bit lower (5k-10k), as it based on web technology. Additionally, it stands out concerning post-deploy cost versus NLP. Once deployment is made, CI can work autonomously since day one without many (or any) human assistance. It does need continuous improvement to make the user interaction frictionless but usually at a fraction of the cost of NLP´s AI training.
Examples of Conversational Interface
K2 Agency — K2 bank bot
This is a platform built by K2 Agency, who specializes in designing and building fin-tech product.
I won’t go into details about it because K2 has already done an excellent job describing how they made the product.
- The importance of error management in a chatbot. I.e. how can user change a data after sending it.
- How to leverage keyword recommendations to simplify user’s mental work
- They made the bot available for multiple devices (PC, mobile, IoT), input systems (text or voice) and platforms (Amazon alexa, FB messenger).
Typeform — conversational article
Another super interesting use case, created by Typeform the most user-friendly forms builder ever existed. It´s a post that combines conventional text based content with small chatbot interactions which provides extra information about certain topics. You can check it out here:
Like the previous case, the Typeform team has also done a phenomenal job describing the bot building process.
- You should think a lot about how to add value to readers with the chatbot and avoid being perceived as distractions/intrusive.
- Some interesting points about how to make the content more personalized and unique for each user. I.e. if someone like cats show picture of cats in future interactions.
- Design the conversational experience thinking about possible troll actions from readers. I.e., what´s your name => my name is Dinosaur xD.
Landbot.io — Landing page as a chatbot
You can see the example here.
The final use case is about building conversational experiences for landing pages. Obviously, I might be biased here, but IMHO this is the killer app for chatbots. I already explain the reasons in my previous article:
- CI by nature focuses better the user attention which helps to drive engangement.
- The potential of CI is to personalize each user interaction and always offer the most relevant information.
- Being able to qualify lead in real time is a huge benefit for companies to optimize their full funnel conversion.
In my opinion, CI is still in its early day, so we will see how things will evolve. Maybe conversational article will take off, or maybe we can combine Voice UI with CI as a visual support. It´s certainly a perfect time for entrepreneurs and bot enthusiasts to experiment new uses cases with CI. In a future post, I will describe how to design and build conversational experiences.
If you like this post please show me some ❤️ with your awesome claps 👏
Source: Chatbots Magazine