We have been working on ChatBot projects and the analytic is one of the key issues for the success of the projects. Hence we decided to try Chatbase in our projects.
First of all you need basic coding skill to integrate your ChatBot to Chatbase. But as I said, it is not too complicated, you just need to post messages and some tags to the Chatbase.
Enter to Chatbase website and create a bot to get an API key to send data to Chatbase.
Now you should make some changes in your ChatBot code to post related data.
curl -X POST \
-H ‘cache-control: no-cache’ \
-H ‘content-type: application/json’ \
“api_key”: “[your api key]”,
“message”: “Hello, bot!”,
As you see above, you should send message and intent to the Chatbase in the body of your post call.
API Key: The key you had in step 1.
Platform: It can be web, app, facebook or other platforms. If you state platform, you can create platform based reports.
Message: It is the message of the user. You can use variable to make a dynamic request.
Intent: If you use NLP for your ChatBot, you should be familiar with intents. Intent is important to group different messages in the reports. You should post intent found by your NLP tool to Chatbase.
user_id : User id should be unique. For Facebook it is Facebook ID and in other channels there should be a unique ID for user identification.
After sending message and intent to the Chatbase, you should define steps cutting flow of ChatBot. In Chatbase the messages not replied by ChatBot is calling as “not handled”. Not handled messages are important to find problems in the chat flow.
Tracking linked URLs are another important key for the reports. Hence URL tracking will provide crucial information about the success of the ChatBot.
In above example, you can see the easiest way to track a URL in Chatbase.
Using Chatbase you can easily reach number of sessions and users, cohort analysis, session follow, intent related messages.
In the above graph, you can see session flow of one of our demo projects. I personally liked to see this flow, it is useful to make comments about the performance of the ChatBot.
Pros & Cons of Chatbase
- API is working properly and there isn’t an important problem in technical side.
- Not handled message report is usefull to understand the problems of ChatBot.
- URL tracking is working well and usefull.
- Session flow is one of the best reports to see the big picture.
- You can create custom funnels as seen below according to intents of user. It is convenient to test different funnels.
- It is not possible to integrate your ChatBot without coding skill.
- As of first impression in November 2017, it is not as easy an integration with Facebook Chatbot like Dashbot (another ChatBot analytic tool).
- As of first impression in November 2017, it isn’t possible to make user specific analysis in ChatBase. We need to make customer specific analysis and reach messages of a specific user. We can do this analyze in other ChatBot analytic tools such as Dashbot.
My first impression about ChatBot is quite well. I think you can try Chatbase in your projects.
Who are we?
hizliYOL Technology is Turkish ChatBot development company. We are developing enterprise ChatBot projects for different companies in Turkish and English.