How to measure feedback or performance
Measuring performance of a chatbot can be challenging. When is a bot succesful? Is your virtual assistance succesful when there is no handover to a human? Or did your end user dropped out because it performed poor.
In this article we will show you how to create a simple example of feedback flow within Flow.ai.
The Feedback Flow
We start the Feedback Flow with an event. After asnwering an FAQ we instruct the bot to continue with the Feedback Event by dragging and dropping the Event Reply into the FAQ flow.
The Feedback Flow will be triggered and the timer is used to delay a reply by the bot. We continue with a follow up “Was I able to help?”. You can add Quick Replies to improve CX if supported by your channel. From here we’ve got two options:
1. New Flows
If the user answers “Was I able to help?” with the Quick Replies or simply by typing input we trigger a new event and corresponding Flow. If the answer is positive, the bot was able to help, we trigger Feedback Positive and if the answer is negative, we trigger Feedback Negative.
2. Add Tags
Instead of creating more Flows you could also try to keep things more compact by adding tags. If the positve branch has been triggered the conversation can be tagged as Positive and the other way around.
Tag each branch by dragging and dropping a Tag onto your canvas. Then add a name for that tag so we can use it later on.
Our design is finished and we can now let the bot do it’s thing and data will come in. The next step is to analyze the incoming data.
Data Export for Flows and Tags
Within the Analytics view you can select a data range and export data. You will receive an export of your data. Within the data you will be able to see which flows have been triggered and how often it occured. The same process allows you to get insights in the tags
Another way to measure performance is to add an analytics integration from the Intregration section. You will be able to see your flows and tags in your existing analytics tools.