When it comes to analyzing your chatbot performance, there is no better support than the Statistics Module. The tab in Chatbot Creator full of transparent graphical diagrams which enables you to monitor chatbot performance and the behaviour of its users. The Statistics Module allows you to check what content or offer is interesting for your target group and pay attention to unexpected or expected events in your chatbot. As a result, you can draw conclusions and take the right steps so the chatbot works best for most people who use it.
Imagine that you have organized a campaign to bring new users to the chatbot and you want to analyze if this was successful. Or that you have sent a push notification with product updates with links guiding to your e-commerce store and you want to check whether the users opened them. This kind of knowledge is really important for your solution and even more important for the business. You can get it in the Statistics Module.
I’ll start from important chatbot user metrics. These are statistics that allow you to analyze your chatbot performance – whether it has gained new users, whether they return, how much time the interaction usually lasts. In short, they show if, when and for how long people interact with your chatbot.
This shows the total number of people who interacted with your chatbot. You can observe how the number increases. Here you will see if your promotional campaigns bore fruit and if your chatbot audience has grown over time.
How many people have interacted with your chatbot during a given day? How many of them have just started using your chatbot? You’ll find the answer here.
The Daily Activity Dashboard is divided into new users and active users so you can evaluate your actions and check how they influence interactions during a given day.
“Messages” is the dashboard with the number of messages exchanged between the user and the chatbot.
Here you can see how many sessions your users have purchased. And what the time of the average session is.
This shows how our users got to the chatbot. Here you can check whether people found your chatbot in the way that you had planned.
This additional metric is useful when you run a competition in your chatbot. Here you can check how many people sent in an entry to the competition on a given day and on which day they sent the largest number of entries.
In this part you’ll find more specific metrics showing how people use your chatbot. Which block they choose, how often they ask for the moderator’s help, which elements they click on.
In statistical Events, you can add parameters that gather information about users who arrived in a specific place in your chatbot. Some Events are set by default inside the Chatbot Creator platform:
→ elements sent by users to the chatbot like “vulgarisms” or “images”
→ enabling Moderator mode – tells you how many of your chatbot users have searched for help of the consultant.
If there are many such people, you should think about the cause. Perhaps you didn’t explain how your chatbot works well enough?
You can also set events in the chatbot structure by yourself to check:
→ the number of reservations or appointments booked via your bot;
→ the number of products ordered via your bot;
Statistical events that are shown in the “Events” table inside the Chatbot Creator platform can be defined in the block structure by adding an action (before a message) or in the button.
Shows how many times a specific block has been displayed to the users. Thanks to this metric you can see which blocks are the most “popular” in your chatbot.
Here you can see how many times a given link has been opened by users in general. It will show you if the users click on buttons with links to external pages of your choice. This statistic is especially useful if you want your chatbot to support sales in your business.
The chatbot needs to be useful and respond to its users adequately – that’s the basis. The “detect intents” table, I’m describing below reveals whether it works in your chatbot’s case. What’s more, by showing users inputs and unrecognized phrases, it lets you improve your chatbot.
The system detects and shows you which phrases entered by users are recognized and not recognized by a chatbot in the form of an excel table. You don’t have to look at every conversation one by one. Just download the list. Based on that, you can improve your chatbot so it understands your users better.
Check out our short tutorial about the Statistics Module inside of Chatbot Creator.
More chatbot tutorials you'll find inside the Chatbot Creator App
Once you know what separate metrics mean for the chatbot, you can pass along and discover which numbers to compare to evaluate whether your chatbot is doing its job.
Of course, depending on your chatbot’s goal you can choose different types of data and make your own comparisons. However, it is good to start from the basic ones.
Combining data about a number of sessions in the chatbot with a number of users during a given day you can see how many times one person interacted with the chatbot.
[ When your goal is to engage customers, these stats will tell you whether they are likely to go back to the chatbot ]
If you set a specific event you can cross the number that is displayed in an Event chart with the number of active users. Then you can see how many people who have interacted with the chatbot during a given day (or period of time) performed and action that you planned. You can also find out whether the path that should guide people up to this point is effective. You can see how many people who talked to made and appoitment or ordered a product.
You can check how events work in the Health and Beauty Chatbot Template in the Chatbot Creator app.
Comparing those two metrics shows what is the number of people who interacted with your chatbot in general and those who accomplished adding an entry to the competition. You can count an average number of entries per person.
Another crucial part is the chatbot customer flow. Check how users navigate the chatbot and what their most common paths are. What buttons and features are used the most often.
Which blocks have a bigger click rate, when do users usually ask for contact with a moderator or end an interaction with the chatbot.
Since you have the customer journey in the offline world, you can think about the same concept in the online, chatbot world. Taking that into account, you can map how many users:
Then try to answer the question of which obstacles they faced, when the moment they resigned from completing an action occurred, what else you can do to make it easier for them.
The information provided above allows you to optimize and develop the chatbot.