How to Build and Configure a Chatbot For Your Shopify Store
Over the last few years, you’ve probably seen a proliferation of the chatbot, though you may not have realized what it is. More and more websites have been implementing a live chat system into their websites. These systems often sit passively in the corner of the site, allowing users to contact someone inside the business at any time.
If you’re not a frequent user of these systems, you may not realize that many of them are currently powered by robots, rather than humans. They offer the same sort of benefits as an automated phone system, with a few similar drawbacks.
If you’re interested in making use of this communications channel, you have a lot of options, but you should make sure it’s what you really want to do first.
Comparing a Chatbot with a Messenger System
The first decision you need to make is whether you want to use a live chat system or a chatbot. Both options are available to you, but they have specific considerations you need to review to make the decision.
Live chats are an easy system that works similarly to a telephone system. You have operators standing by, ready to take the digital call, answering whenever someone on your website decides to initiate a conversation with the system. You can outsource this to a company that provides a call center-like answering service, or have it linked to your customer service team in your office.
Live chats have a few tangible benefits as a system:
- By connecting to a real human, users are able to explain complex issues and receive solutions faster than virtually any other method of service.
- Live chat is better for many users, particularly younger users, than alternative support methods like email or phone calls.
- Customer service agents can dynamically choose if a situation is right to push for a transition to a sales contact or otherwise manage expectations for a chat directly.
On the other hand, it has a few drawbacks as well.
- In order for live chat to work, an agent needs to be ready to answer, which means you need employee coverage around the clock, or else limit your service to business hours.
- If you have a significant volume of contacts, your agents may be overwhelmed, which can increase wait times and dissatisfy customers.
- Depending on your agents, complex issues may be difficult to handle without escalation to other employees.
Live chats can be very helpful from a customer service perspective, and they can be decent at generating leads and sales, but you need quite a bit of training for your teams to leverage them appropriately. The ongoing expense of 24/7 coverage can also be quite steep, depending on how you do it.
Chatbots, meanwhile, are conversational flowcharts triggered by keywords in what a user types to your chat system. While many companies describe them as “AI”, they’re nowhere near what you would consider an AI if you’re a fan of science fiction. They are, in essence, just a series of messages triggered by keywords, the same way a phone system might have a user press a button or speak a word to navigate through menus.
The main difference between a phone system and a chatbot is that the chatbots can hide some of what they’re doing. They can use lists of trigger words rather than a single “press 1 for A, 2 for B” sort of choice system. This makes the conversational engine feel smoother.
The pros of a chatbot include:
- Easy 24/7 coverage without needing to pay an employee to be available around the clock.
- The ability to offload most basic customer service questions to what is essentially guided self-service.
- The ability to handle large volumes of contacts without overloading.
Given that a chatbot is not a true AI, though, there are some flaws with the system.
- Complex issues can confuse the bot and require escalation to a real human.
- Some users don’t like having to talk to a non-human system and would rather chat with a real human.
- Chatbots don’t usually learn and adapt over time; they can only handle what you program them to handle.
All of this combined means that some situations are excellent for a chatbot (large volumes of simple questions) and others (complex questions or unique issues) are not.
There’s a third method, a hybrid system, that many businesses choose to use. This is a system that uses a chatbot to perform the initial contact, solving basic issues, and filtering out contacts that don’t require human intervention.
If something is too complex, confusing, or off-script for the bot, it connects the user to a chat agent. This option has many of the benefits of both a live chat and a chatbot, though it does still mean you need a live agent available at all times you want support to be available.
Determine the Purpose and Scope of Your Chatbot
Assuming that you’ve decided to use a chatbot system, either on its own or as part of a hybrid system, you need to determine what the purpose of your chatbot will be.
Generally, you’ll see two kinds of chatbots; support and sales. A support bot can offer assistance with most of the basic and intermediate customer support issues. You often see these systems live on support pages or accessible by clicking links on a support page. A sales bot is more frequently installed on the rest of the website, including landing pages, blog posts, and product pages.
You can certainly design a single bot to handle everything you can think of. A bot is only limited by two things: how much effort you’re willing to put into configuring it, and whether or not the system you’re using has a limit on chat flows and messages.
Some businesses prefer to use a chatbot solely to handle basic customer service issues and sales questions, such as questions about product features. The bot will be relatively simple and will either redirect a user to another method of contact if they need more than the bot can handle, or will transition to a live agent.
Other businesses strive to make the most robust and all-encompassing chatbots they can. These are generally built up over the course of months or years, with new content and new flows added, old content changed, and even split testing implemented for sales information along the way.
Our recommendation is to start small. The larger and more sprawling your chatbot becomes, the more issues you are likely to have with it, and often the more expensive the system you’ll have to use is.
Decide on a Method of Installation
There are three ways you can use a chatbot on a Shopify store.
The first is to use a third-party system such as Facebook Messenger. Messenger has a chatbot system, though it’s very basic, it’s familiar to millions of users because of how prevalent Messenger is today. There are other third-party systems as well, though they don’t have the familiarity of Facebook or the ease of use you might expect from such a system.
The second is to use a Shopify App. There are quite a few live chat systems that have Shopify apps to pick through. We don’t have any particular recommendations here; simply look them over, determine what their pros and cons are, and pick one that works for your purposes. Just remember that many of them have limits on cheap or free plans, and will still require custom configuration.
The third is to use a custom-developed chatbot system. Chat frameworks are available to use, and developing a custom system on top of them isn’t necessarily a tall order. You’ll still need the time and skills of a developer, and the money to pay them, but the end result is a more customized and generally less limited system than what you get from other options.
Develop a Chatbot Flow and Script
Once you’ve determined the system you’re going to use, you can start to develop your conversation flows and scripts. You don’t need to start writing out a copy yet; just concepts, questions, the information you want to be included, and so on.
To start with, look up what your most common customer contacts are about. Do they ask for product specifications? Do they have questions about pricing, shipping, return windows, warranties, or other processes? Do they have common support questions about your product? Every business has a list of common issues they address on a daily or weekly basis; this is a great foundation for your chatbot. Another great source of information is any FAQs or knowledge bases you’ve set up to handle simple self-service support.
Next, consider how deep into the weeds you want to go with secondary and tertiary support concerns. Is it worth the time to develop a conversation flow for a particular support question you only see once a month or only a handful of times a year? Some businesses will say yes, others will say no. It’s entirely up to you.
Another consideration you may have is whether you want your chatbot to encourage sales. If your chatbot is implemented site-wide, a lot of common questions will naturally come from users who don’t already own your product. They’ll have questions about your product, and you can have your bot answer them, sure. You can also use the bot to incentivize a purchase, by offering discount codes, coupons, trials, or other incentives.
Next, you will want to start mapping out the flow of a conversation. All chatbot contacts will start from a single welcome message asking the user what they need. This message will be your most frequently tested and changed message, as you develop ways to greet a user that encourages them to talk about their issue in a way the bot can pick up.
From there, you want to divide into categories of conversation and map out different issues and flows. Consider what a user can say that will trigger a given topic, what message the bot will respond with, and how the conversation can progress. Your system may also need to ask for additional information to help solve some kinds of problems, especially if you’re giving it the ability to perform common support tasks as well as provide information. This can be dangerous, though, so we recommend avoiding it without serious thought to account security.
Developing a chatbot’s conversation flow is an endlessly deep subject. You can spend months learning and reading about advanced communication flows, structured and unstructured conversational information, rigidity versus volatility, and other concerns. For business support or sales, you don’t need to dig that deep for the most common usage, but it’s certainly a fascinating area of study.
Illustration: A chat bot sending a webhook to register seats for an event automatically and returning a response
Once you have your conversations mapped out, you can start developing the exact copy your chatbot will use. Writing copy, you should focus on human-sounding live chat content, but make sure to disclose that the user is reaching an automated system initially, so they have the right expectations in mind.
Implementing a Custom Chat Solution
Many “out of the box” chat systems are not as plug and play as you’d think. Many of them allow custom styling, advanced conversion tracking, webhooks for your CRM software, and other features that require practiced hands to set up properly.
If you’re looking for a surface-level chat system for basic queries and you don’t have any requirements to store or use the data coming through your chats, then you may only require a basic chat system without any of these advanced features.
If your chats are high-value customers and if you can make value out of conversion and customer data, and if you’d like to tailor your chat experience to your website, it’s recommended that you work with an expert to implement these improvements.
When configured properly, you can even harvest the data your customers share to personalize your emails to them, and even send them ultra-targeted SMS messages. The initial setup takes a little time, but once you’re up and running, this is the kind of software that pays dividends for years.
Chatbots are surprisingly useful and efficient, but when they’re written poorly, it can have devastating effects. If your automatic failover to a live person is not coded properly, your customers could end up chatting with a bot endlessly and unable to get to a live human. This is detrimental to your user experience, and you may just see an increase in phone calls and emails as people abandon their chats. If your SMS messaging isn’t configured properly, you might end up making an embarrassing mistake like texting your customer “Hi %%customer_name%%!”.
User experience is everything.
There are many live chatbot frameworks available, but what matters is how robust your codebase and your conversation framework is. If you don’t have any experience setting up chatbots, you may run into some embarrassing issues like in the image above. Just run a Google search for “chatbot fails” to see hundreds of examples of poorly configured chatbots. You don’t want to be on those lists!
If you’re interested in a custom chatbot for your sales and support, or just want an out of the box system configured for your Shopify store, you’ve come to the right place. Drop us a line and we’ll chat about what we can do for you!
- Which Shopify Apps Can Help You Sell Subscription Products? - January 16, 2021
- 6 Ways to Build a Visual Product Customizer in Shopify - January 4, 2021
- How to Build and Configure a Chatbot For Your Shopify Store - December 16, 2020