How to influence users with chat bots

Chat bots have been around since the time of Netscape Navigator and the first major push of the consumer web

In 1994 the Eliza was created by Joseph Weizenbaum of MIT, who referred to his creation as the chatterbot. From the humble beginnings of taking in key phrases and using them to fabricate conversation, to complicated interactive customer service bots used by e-commerce juggernauts like Amazon, chat bots have seen a variety of functionality expansions and improvements over the past 30 years. Here we look at ways chat bots can positively influencer user experience and strengthen relationships.

What is a chat bot?

According to Oracle, a chat bot is simply a piece of software that mimics and parses language for the purpose of furthering a conversation. It can be as simple as the Eliza program, or as complex as a personalized digital assistant that uses machine learning to understand how to respond to future queries based on previous interactions.

What are the primary use cases for chat bots?

Many industries utilize chat bots in an attempt to both conserve resources and service users in a 24/7, always on digital world. They can be found in niche roles for customer service, technical support, scheduling, sales and social proof, and relationship management.

Does influence matter in chat bots?

Absolutely. Each use case will dictate the consumer type being dealt with and will vary depending on who a company is looking to influence to ensure that specific consumer behavior factors that affect decision making are taken into consideration for the purpose of reaching goal success metrics.

Let's look at a few examples:

1.) Customer service

Utilized by several of the large customer service platforms, the chat bots can be as simple as Eliza's keyword understanding that matches up to an expected support article listed in a company's knowledge base or frequently asked questions (FAQ) section, but they can be significantly more complex. Zendesk created an Answer Bot that doesn't just chat with end users to provide possible answers, but actually attempts to handle basic support tickets. The bot is conversational and attempts to not only provide education, as one would expect with a chat bot, but also to assist in actions.

Where does influence fit in? Using helpfulness ratings from previous interactions, a customer service chat bot can both provide a peer level influence signal in the form of: "151 others that mentioned [keyword] found this FAQ article the most helpful", to guide users to the fastest right answer as well as acting as a feedback loop to improve FAQ materials rated poorly to better suit the queries posed.

2.) Technical support

The next step up from general customer service queries such as resetting passwords or checking on an order status is to handle complex technical interactions that might require developer assistance. However, when users encounter bugs they may not be too happy to interact with a bot, with 40% of polled individuals per Zendesk preferring to interact with a human agent. In these instances, a technical support chat bot's role is to gather information such as the process undertaken prior to encountering an issue, assisting in grabbing a screenshot and system environment details that a developer may find helpful, and assuring the user that the issue is being directly assigned to be fixed with an elevated severity.

A company using information that gathers chat bots as part of its technical support operations is influencing users to provide as many details as possible in a structured format that will help save support and development staff time on rote activities, trains users that solving technical problems can be done with minimal time investment on their end, and ultimately can lead to better product outcomes when the feedback is received on a regular basis.

3.) Scheduling

Do you remember the last time you had to call and schedule an appointment for your car, a haircut, or plumber? More often, service-oriented companies that require appointments have been opting for either basic calendar scheduling services or scheduling chat bots. The biggest difference with the scheduling chat bots, which are often powered through an API to a calendar scheduling service is they allow the user to ask questions related to who might be available to provide a service, expected wait times, things to do while waiting, or even tease out upsells before a service is scheduled.

Users interacting with scheduling chat bots can be influenced by showing data in the scheduling process. This includes expected wait times, supply and demand details (if applicable), upsell opportunities that complement the service, and various incentives (in order to guide users towards appointment times more advantageous to a company, reducing overflow labor costs).

4.) Sales and social proof

Depending on the product, a sales chat bot might be a glorified scheduling chat bot to setup a demo. However, they can be also used for information gathering purposes to act as lead generation in a conversational manner that is more personalized than simple form fills. This is due to the ability to ask leading questions that guide the prospective customer into a more useful sales path for a human salesperson closing the deal later on. More complicated sales chat bots can even go so far as to qualify a lead to determine which sales rep should be working it, the priority to the point of handing off into a direct phone call from the chat window, or initiating a self-service trial for leads deemed unqualified to require human interaction but still worthwhile to work.

Influence matters greatly within sales chat bots, primarily via social proof. This can take the form of showing a user in real time what similar users have purchased based on information provided, guiding towards a particular product or purchase route, but can also act as a way to build trust by helping a lead to save money by providing the exact subscription plan or product quantity they need. Such trust and rapport leads to easier relationship management.

5.) Relationship management

Recently updated review guidelines from the FTC make it clear that most governments will be moving towards policies of transparency when it comes to reviews. In the past, some relationship management chat bots would solicit for reviews only in the instances of a positive experience, while guiding poor experiences towards support.

While it makes sense to address problems and ask for referrals, it is expected that as an industry we'll see an evolution in relationship management chat bots. A most likely scenario will involve randomized solicitation requests for reviews inserted into chat flow that are independent of perceived satisfaction levels, and there being a greater emphasis on solving customer service issues before customers feel negatively towards the company.

Relationship management is intertwined with influence as it is a request for a user to influence others. While the processes may change as the industry adapts to regulation, companies that tweak their customer service processes, resulting in more highly satisfied consumers, are going to reap the rewards that come from being able to ask a greater percentage of their base for referrals and reviews.

When should you involve a human?

Despite their many uses to influence users, chat bots are not the be-all and end-all of user interactions. It is important to know when a chat bot needs to involve human interaction. For instance, a well-developed chat bot will use a degree of either natural language processing or basic keyword parsing to determine if a user is using phrases associated with anger, annoyance, frustration or displeasure, and attempt to initiate a handoff to a human agent via live chat, or schedule a phone call back as a method of de-escalation. This way, a chat bot can be used to assist in the interactions that matter most to a user, create the outcome everyone wants, save time and money, and show the user that the company can still offer human assistance when it’s required.