If you think back to your time at school, there was always that one kid the teacher (charitably) called the 'Chatterbox". Like a classroom superhero (or supervillain, depending on how close you were sat) they could talk non-stop, with a seemingly never-ending social battery! But when they’re gone, we miss them. Why IS that? Well, it’s in our nature to want someone to talk to us, to hear us out. And while a 'chatterbox' is rarely much help in the listening department (because they're full noise), chatbots, powered by cognitive AI (artificial intelligence), are more than capable of handling a two-way conversation.
In 2011, Gartner predicted that by 2020, customers will manage 85% of their relationships with the enterprise without interacting with a human. And here we are in 2021, with chatbot technology spread across hundreds of channels, allowing us to vastly improve all customer interactions and increase the number of potential business opportunities!
Take the Jenny chatbot by GetJenny, as an example. It is always up and running 24/7 to successfully handle customer support requests across Slush’s website and mobile app. The Jenny chatbot also increased the conversion rate by 55% for Slush in just a year!
Chatbots are a bona fide way to provide increased opportunities for your customers to convert through your website. They also reduce the cost of customer service by 30%.
47% of digitally mature organisations say they have a defined AI strategy to improve user experience. Are you one of them? If the answer's no, read on to find out how you can implement a successful chatbot strategy by following five simple steps:
- Gather information about your target audience from a variety of sources
- Plan the type of chatbot, and what the bot is going to do to meet customer expectations
- Select a platform & build your bot to create a great chatbot experience
- Check if the chatbot works & improve it further
- Launch your chatbot & monitor its activity in terms of customer engagement and user types
Step 1: Gather information about your potential customers
To kick things off, you should start by gathering information. Be the spy kid. Find out who your customers are, what they want, which words they use to get their idea or query across and get familiar with the customer experience in its full capacity.
Many businesses already have data on their customers from different channels. They understand and define customer intents with the help of customer journey maps.
Customer journey maps start with awareness and end with the final purchase, followed by brand loyalty (all things going cushty). Analysing different checkpoints during this customer journey will give you insight into what served as a pain point for the customer. You will then be able to leverage this information to build your bot (in step 3).
Customer intents drawn from journey maps tell us:
- The pain points of the customers.
- The roadblocks that require a real-life person from the organisation to jump in to remove them.
- The kind of conversations that satisfy the customers.
- If the customers want to make requests, talk about incidents, or ask questions related to your brand of products or services.
- The products that attract the highest volume of interactions from customers.
This information helps us understand our customers better and, in turn, creates sensical and helpful conversations as part of the chatbot!
Step 2: Decide what the bot is going to do to meet customer expectations
After we've got the customers expectations down-pat, we move on to planning and deciding how our chatbot is going to satisfy the needs of our potential customers as they move through their customer journey. Now, let’s look into six ways a chatbot can meet customer expectations:
- Narrow scope. The key here is to not spread ourselves too thin. Answering technical questions is difficult in comparison to answering simple questions. So to ensure our bot provides correct answers and ensure an effective conversation flow is maintained to allow users to have a positive experience, we keep the scope as narrow as possible. Will we be doing less? Sure. But it's a matter of quality over quantity, and we know we can always widen our scope or purpose in the future when we have a better understanding of customer behaviours.
- Consistent responses. We want our chatbot to have a personality that doesn’t sound fake and aligns with our brand's tone of voice. After all, the whole point of it all is to create a real-life experience for customers that feels as close to a normal human interaction as possible.
- 24/7 availability. Customers are impatient and don’t like waiting around for hours to receive a response from human-mediated customer service. Chatbots are the ultimate workaholics and never grow tired of answering user queries straight away, whenever your customers need assistance!
- Identifying the intent. Chatbots should decipher the customer’s query and bring up appropriate responses from the database.
- Streamline experience. The conversation should be smooth and delight our customers.
- Multiple languages. International clients find it easy to communicate with a bot that understands their native language.
Step 3: Select a platform & build your bot
Once we define our target audience and chatbot objectives to meet the customer expectations, the next logical step is to build a bot using the questions of the customers, responses, network alerts, and API (Application Programming Interface).
Now, which platform should you use for building and managing our chatbot? Well, if you're a Hubspot user - you're sorted. But if not, no trouble! There are plenty of chatbot development companies out there, but our favourite is the Ayehu automation platform.
Ayehu is user-friendly and brings you the best results in a matter of just minutes. Goodbye to hours of coding and scripting - you beauty! Even McKinsey&Company vouch for it, calling it the “best-in-class solution” when selecting the right platform for chatbot automation (with both simple and more complex problem-resolution capabilities).
Other chatbot solution providers include TARS, ManyChat, Botsociety, Landbot, Chatfuel, Smartloop, so do your research!
Step 4: Check if the chatbot works & improve it further
Our bot learning should not come at the expense of real customers looking for answers. It’s best to test bot metrics with staff members and select users before putting it out there for the world to see. Running it by them first will help the chatbot learn to decode new phrases and link them to pre-defined intents. It will also discover aspects that hinder user experience, like loopholes, bugs, and new intents that we were missing before. Remember, the sky’s the limit when it comes to improving AI (artificial intelligence) and bot learning features.
Is this enough for officially launching the chatbot? Well... not quite.
Who is going to help the staff learn and adapt to new technology and get to grips with new features? They also need to understand how to work with the chatbot tool and identify their new responsibilities (i.e., monitoring the efficiency of their chatbot). For this, consider testing by a broad customer base for a short period of time. It will equip the staff with skills to carry out the final step on our list!
Step 5: Launch your chat box & monitor its activity
Now that we’re through with the building, learning, and testing of our chatbot, it’s time to finally launch it in the market! Channels you can launch your chatbot on include the client website, Facebook Messenger, WhatsApp business account, Slack, Telegram, etc.
Once again, our work is not done here (spoiler alert: it never is - you should always be looking for ways to create the most successful customer experience). After launching the new service, we must monitor the activity of the chatbot in terms of its performance, while being bombarded with hundreds and thousands of intents demanding quick and valuable responses.
Metrics such as AHT (average handle time) and MTTR (mean time to resolve) just won’t cut the mustard for a multi-faceted channel that juggles multiple queries simultaneously and where time is a variable factor, given the nature of intent is different for every customer.
The following three engagement metrics give us a better understanding of user actions, and the areas we should look into to optimise user experience and improve customer engagement.
- Successful outcome or completion rate. This metric tracks the number of successful conversations between the customers and the chatbot. A successful conversation means that the customers are happy and satisfied with the chatbot’s responses.
- Bounce rate or fallout rate. The bounce rate measures how often customers turn to some other channel (or a real-life agent) after finding it difficult to work with the chatbot.
- Reuse rate. When we enjoy a service, we keep on coming back for more. The reuse rate measure tracks how often customers return to the bot again after a delightful, successful experience.
Monitoring these relevant metrics helps us gather more data which we can feed into the chatbot’s learning process for future references.
Time to get started
Well, there you have it! Now that we've whizzed through the five steps of implementing an effective chatbot strategy, you should have a good idea of how you can get the ball rolling to build your additional customer service offering.
And of course, if you have any issues with your HubSpot marketing automation, don't think twice about dropping us a line!