How To Article

How To Get Started With Chat Automation

By April 2, 2019 May 9th, 2019 No Comments
How To Get Started With Chat Automation

Rules-based automation and artificial intelligence have expanded the horizons for business processes and continue to alter them. However, taking a step towards chat automation is overall more difficult to achieve because of the real-time nature of chat messages.

However, introducing message automation to your customer chats provides much-needed relief to customer service representatives and reduces the overall cost of your customer service operation. It will allow your customer support staff to take more time to solve issues with higher complexity that require empathy and/or technical experience.

As with any endeavor that has high potential cost savings, the process of message automation is not one that you can just flick on and off like a light switch. It is a process of trial, error, and optimization. It is essential to go through this process because it is easy to disappoint a user or a customer in the real-time messaging environment.

We've developed five necessary steps to follow. To follow these five steps you'll need:

The steps are:

  1. Evaluate Your Customer Conversations
  2. Identify Patterns
  3. Experiment With A Light Onboarding Automation
  4. Train Your Smart Replies To Answer Your FAQs
  5. Analyze, Optimize & Implement

1. Evaluate Your Customer Conversations

The first step to message automation is understanding the chats your customer service representatives and salespeople are having with your customers or users.

The goal of this step is to get a vague idea of the potential pay off you can achieve by automating some conversations.

For the initial research, ask the business development team, sales department, and customer representatives for anecdotal data. You'll be surprised how bothered employees can be about repetitive messaging.

Record the answers employee's give to these three critical questions:

  • What are the most common first message or messages that they receive in the chat? We'll call these onboarding messages.
  • What are the most common problems your customer support team responds to for customers or users that your company has already chatted with? We'll call these Frequently Asked Questions (FAQs).
  • How often are the responses given for either onboarding messages or FAQs exactly the same?

At this stage, you'll have a pretty good idea of which topics you should conduct more research about by looking at past chat conversations. If your staff say they don't experience any repetitive conversations, you may not benefit from message automation just yet.

2. Identify Patterns

The second step to practical chat automation involves finding parts of conversations that you can automate using chat automation rules (for Onboarding Messages) or AI (for FAQs). To do so, find sample chats of onboarding messages & FAQs that you learned about from your team in Step 1.

The best way to find samples of onboarding messages and FAQs is to look directly at chats. You may try to go through emails to find this information, but email conversations will not be useful because the way people write in a real-time chat is vastly different to email.

If you've got live chat set up on your website, checking those conversations is an excellent place to start. If you have messaging app business accounts set up already, you'll be able to find records of your chats in the following places:

Once you've got about 10 to 20 examples of chat conversations for each onboarding message & FAQ you'd like to automate you're ready for the next step.

3. Experiment With A Light Onboarding Automation

For most businesses, we see a few questions or topics that are so prevalent that it's best to preempt it for all new contacts (subscribers).

An onboarding automation helps to welcome new contacts into the conversation. It can be a message to educate a contact about an issue or a question that the contact can answer that lets you help them faster.

When choosing what kind of message to send or question to ask with your first onboarding automation consider the chat automation matrix below.

This is a chat automation matrix.
Chat Automation Matrix

Make sure to start automating messages where you have a high probability of being right and a low cost of being wrong.

Our first experiment was just one simple question. What language do you prefer to chat in English, Spanish or Chinese? It's very easy to make a simple onboarding automation like this for your own company.

1. Navigate to Settings > Add Field and create a field with the contacts preferred language.

Create a Custom Field to Record the Data
Create a Custom Field to Record the Data

2. Navigate to Surveys > Add Survey and create a survey to ask a contact their preferred language.

Create A Survey To Collect The Data
Create A Survey To Collect The Data

3. Navigate to Automation > Add Rule and Create a Rule that sends the survey to the contact when they start the conversation.

Create A Rule To Launch The Survey
Create A Rule To Launch The Survey

4. Navigate to Automation > Add Rule and create a rule that assigns the contact to one of your Rocketbots platform users depending on the language they choose.

Create A Rule To Assign Someone Based On The Survey Response
Create A Rule To Assign Someone Based On The Survey Response

That's it. You've created your first automation.

A word of caution, we tried our first experiments with onboarding automations in a closed environment that didn't include our real users. However, we quickly learned this method wasted our teams time (they were the testers) and usually didn't reflect on what actually happened in the real world. Experiment with very light automation first, like the one we described above and test it on your real contacts.

If you're ready for something more complex, we've created also created a guide to complex onboarding automations.

Now that you've done something to automation contact onboarding, let's take care of FAQs.

4. Train Your Smart Replies To Answer Your FAQs

Once you've gone through your chat logs and have a good idea what your frequently asked questions are, you'll need to make canned responses. Canned responses are just standardized responses that make answer common simple questions.

Remember, your canned responses should not cover every case or version of the question, just the most common ones.

In our case, the most common questions are about connecting channels. People will ask, how can I connect WhatsApp to your platform? How can I connect Line to your platform?

The answers we give to these questions are quite simple. We point them to the documentation. Getting smart replies to start answering your FAQs is easy.

1. Add your canned response to Smart Replies by using it. Rocketbots remembers the responses you use and offers them up as smart replies, when a similar question comes up.

Use Your Canned Reponse By Answering a Relevant Question

2. Train Smart Replies by answering questions using the same Smart Reply every time. It's crucial to use the exact same response, this will train Smart Replies much faster. As you type in the input bar, you will also search through Smart Replies, making it easy to use the exact same one.

Typing In The Input Bar Will Search Through All Smart Replies

3. Keep using the Smart Reply even when it comes up as the top result. By continuously using the same Smart Reply you will raise this responses confidence level to above 90%. Once this happens, you can set Smart Replies to send automatically.

After 2 or 3 Interactions You Canned Response Will Become The Top Answer

4. Activate automated smart replies when you're ready. We recommend starting at 90% confidence.

Once The Response Confidence Is High, Automate Smart Replies

That's it. Now that you've trained one Smart Reply, just repeat the process for common questions and soon enough you'll have 20% to 50% of your FAQs answered automatically.

5. Analyze, Optimize & Implement

Nobody expects you to get it right the first time. Chat automation is a trial and error process that requires organizations to test better variants of their answers repeatedly and analyze consumer feedback to automated chats to improve their design.

The key to a successful chat automation system is the ability of the organization to record, evaluate, and improve, based on their findings.

Iaroslav Kudritskiy

Iaroslav Kudritskiy

Iaroslav is the Chief Operating Officer at Rocketbots overseeing growth and operations. He is a graduate of the MBA program at the Chinese University of Hong Kong and his previous roles include positions at Kodak Alaris, Xaxis & Light Reaction.

Leave a Reply