Chatbots for business
There are many ways that any business can benefit from using chatbots as a channel to interact with customers. It’s easy to see this if you think about chatbots in two ways:
- As a form of automated messaging;
- As an application (even a graphical application) that lives inside a messaging application (such as Facebook Messenger, Kik or Telegram).
Automated messaging that lives inside an existing app has powerful and enormously beneficial businesses uses. Being able to use chatbots within messaging apps means there is no need to download the bot, nor to authenticate the user. There’s no standalone download, no separate authentication, no new system for the user to learn, or any other such friction. Bots automatically benefit from the fact that chat apps have a critical mass of users that can participate in the bot experience. They piggyback on this buy-in.
Imagine you received an email asking you to pay a bill. This email may contain a link to the system on which you could pay your bill. In order to do so, you need to open that system, log on, navigate to the relevant page and then go through the relevant steps required to pay the bill.
Now imagine you received a message asking you to pay your bill within a chat app. The message is sent from a bot that allows you to click a button in the chat app to pay the bill. If you click the button, a graphical screen will appear in the chat app with all the main details of the transaction already filled in. All you need to do is click on the pay button.
It’s easy to see that the second example is a much better experience for the user, and will have a much higher success rate. The bot experience allows the user to benefit from the fact that they are already a verified user of the chat app, and that the message requesting payment is tightly integrated with the functionality to pay.
There are countless sorts of examples like these that we can dream up. There are also examples where the user may be seeking information from the bot, and can access this information directly in the chat app. And there are examples where the user wants to perform an action directly in the chat app, rather than downloading and using an app to do so.
Before discussing different possible use cases for bots, it’s important to note that there are some limitations to their use.
Firstly, any new technology easily falls victim to vanity use. So you need to ensure that the bot actually solves a real problem for the end user. And then there still needs to be a compelling reason why a bot is better than an app or a website. The experience of the bot needs to actually be an improvement on using conventional tools, all things considered—including the setup experience.
Secondly, it’s easy to overcomplicate a bot, or try to make it do too much. For example, while natural language processing works really well for broad but superficial tasks, it’s impossible to get it to replace human agents for more context-oriented tasks. These limitations need to be understood and mitigated through the design of the bot, including using features that escalate the conversation to humans if necessary (human-in-the-loop or HITL functionality).
Chat interfaces are also suboptimal for completing many tasks—especially when the user may need to change their mind, or see a compact overview of many possible choices. A graphical interface (still controlled by a bot) is often a better option in these cases.
Here is a list of some good use cases for bots in different industry verticals. It is not exhaustive by any means—rather, it’s an illustration as to how bots might be used in different industries—and across industries.
Customer service is likely the most popular use case for chatbots right now. The technology is there for bots to easily answer questions that are posed in natural language. These systems can work very well, provide the chatbots are mainly limited to simple, frequently asked questions, and the conversations are reliably escalated to humans if the bot cannot answer appropriately.
The capabilities of chatbots in customer service can be expanded to go beyond just responding with information. They can actually allow the user to do things like booking appointments, tracking orders, or buying warranties, whether via text or voice commands, or through a graphical interface.
Financial institutions have already embraced chatbots in a big way, recognizing that they both reduce the cost of customer services, and provide many of those services in a much more convenient form. It’s already possible to do simple tasks such as make payments or balance inquiries via chat or SMS. More sophisticated tasks—purchasing insurance for instance—are becoming a reality. With a judicious use of HITL and graphical interfaces, financial institutions can offer their customers the ability to complete many of the most common transactions conveniently from within their chat application.
There are several possible ecommerce applications for chatbots:
- The obvious application of a chatbot for ecommerce is as the online equivalent of a store assistant, who tries to help you find the item you want. The chatbot would ask you what you would like, and try to assist you in finding it. This use case, however, has issues: search and filter functionality in online stores is typically robust and easy to use, so users don’t have the same problems they face in finding relevant items in a physical store.
- Rather than trying to replace search and filter functionality, which already works well, the chatbot could offer other services that support the buying experience. For instance, it might be possible for the chatbot to offer a more seamless payments solution from within the chat. It also might be possible for the bot to assist in answering questions about a specific item, once the user has already identified it through a broad search.
- Chatbots can also function at a higher level, especially on purchases where the customer might value convenience over value. For example, a chatbot may allow a customer to buy and send flowers much more efficiently than ordering off a website. This could be particularly true if all the relevant address and payment information was already stored within the chat app.
- The fact that chabots can be used in groups will likely result in some interesting use cases for ecommerce, especially in social purchase—perhaps as versions of the Groupon model.
- Finally, chatbots can be linked to geolocation, meaning that they can deliver discounts or information in highly contextual ways. For example, a bot could offer a discount voucher for a particular store when a user walks past it in the mall.
Marketing is also a current big use case for chatbots. There is clearly value for marketers in being able to reach customers inside their preferred chat apps. Currently the open rates on chatbots are much higher than for email because they still have some novelty value.
Although there may be some economic merit in the short term due to the novelty value of bots, in the longer term this channel is likely to become as saturated as email and others.
Nonetheless, bots have a big advantage over other marketing channels because they are not just a channel of communication but also of operation. Not only can a bot advertise a product or service, but a customer can convert directly using the bot, without leaving the channel. This is potentially very powerful for marketers.
The potential virality of bots, and the use of graphical interfaces, will open up opportunities for gamified, social marketing campaigns. Of course, there is also potential for marketers to leverage popular bots using affiliate or native advertising.
Besides the obvious, customer service type applications—such as answering administrative questions (in groups) and booking appointments—bots could be used to bring educational services within the chat app. Many of the experiences that are delivered by a standalone application today could be converted to bot experiences that can take place directly within a chat program.
For example, bots that teach a language could interact with students within a chat app. Students could interact with the bot, answer questions and record messages for it. This type of testing and interaction could be expended to other subject domains. Platforms like Duolingo could expand into chat, rather than having to rely on their own apps and restaurants.
It’s also possible that the social properties of chat apps could be exploited to create social educational experiences. Groups of students could compete to answer questions or create content.
Again there are many customer service opportunities here—reducing the cost of administration and improving the convenience for patients. Booking appointments with doctors and for tests is an obvious use case. Making the insurance process more seamless could be another huge application of chatbots.
Patients will be reluctant to download an app for each doctor or hospital they visit, so a chatbot is potentially a good solution in this space. Geolocated chatbots could guide people through hospitals and allow them to ask questions relevant to the section of the hospital they are located in.
Although AI in healthcare will become huge and will be used extensively to augment the capabilities of health professionals, chatbots will not replace doctors any time soon. They could, however, have an immediate impact in areas such as tracking prescriptions and medication use, and allowing doctors and patients to share health diaries.
There is a particularly good case for chatbots in the government sector. This is because citizens will be reluctant to download apps for all the government services they want to use, especially when they will typically only use these services very infrequently.
Instead of using apps, government departments can offer a wide range of services using chatbots. Citizens could get information on service usage, pay bills, and ask questions. They could also provide information for visas, passports and permits, and ask questions about the processes without leaving the chat platform.
As with government departments, hotel guests will be reluctant to download an app for the hotel they’re visiting unless they are very frequent guests. But there are many services that the internet makes extremely convenient. These could be provided by chatbots. Obvious use cases include asking questions about hotel facilities, and ordering room service.
It’s also possible to use geolocation here: chatbots would appear when required by the physical context. When the customer enters a coffee shop or restaurant, for example, the relevant bot could take their order. It could also be possible for guests to leave feedback or ask questions based on where they are. For example, they could leave feedback on the swimming pool facility when they are by the pool.
There are a huge number of ways that chatbots can be used to improve the customers’ experience of a business. It is likely that soon all businesses will offer chatbots as a way of not just communicating with businesses but a way of interacting with many of the services that the businesses offer.