This page displays user feedback about your chatbots, highlighting their effectiveness and identifying those that are potentially harmful to users. You can only see data for the current period. If you want to see data for another period, select the filter icon at the top of the page and change the date range.
Metrics
Like the Experience Insights page, you can see some quick metrics about your chatbots.
Total users |
Returning users |
Conversations |
Messages |
| The number of unique users sent or received a message during the selected period. | The total number of unique users who have sent or received a message during the selected period and haven't sent or received a message in the prior six months to the period's start date. | The total number of conversations that had a message sent or received during the period. | The total number of messages sent or received during the period. |
Hours Save |
Cost saved |
Sentiment |
Feedback |
| This is the total number of messages generated with positive or no feedback. It excludes negative feedback. This is calculated as $4m saved with 100k messages generated. | This is the number of hours saved times $40, according to the calculation mentioned. | This is the mean average of all feedback comment sentiment scores for feedback given during the period. If this shows a negative number, the overall sentiment is negative. The score can range from -100% to 100%. | This shows the number of positive and negative feedback given during the period. |
Most Effective and Least Effective
These blocks display trends regarding the chatbot's effectiveness in serving different topics. The terms are not direct 'keywords' from the text. Instead, they are identified using topic modeling or clustering techniques, which enables them to be self-identified without any prior training. By default, both views are sorted based on the mean average sentiment, determined by the feedback comments submitted through the feedback form related to that particular subject. It is important to note that sentiment is not based on the number of "thumbs up" or "thumbs down" ratings but rather on the overall tone and sentiment expressed in the comments. The topic with the highest positive or negative feedback sentiment is given priority in the default view.
Each item in the list displays a sentiment bar, with the middle representing a neutral sentiment (0), the far left representing the most negative sentiment, and the far right representing the most positive sentiment. Sentiments can range from -100 to -16, considered assertive, -15 to 15, regarded as energized, and 16 to 100, regarded as balanced. The chart displaying the 'least effective' options includes those deemed harmful, untrue, or unhelpful.
You can change the default sorting to "Most Ratings," which displays the most thumbs up or down. For the Least effective, you can also choose to sort by "Most harmful," "Most untrue," and "Most unhelpful."
You can select a row to see snippets of feedback comments relating to the topic. These are not full comments, just relevant snippets that contain positive or negative sentiments. You can see if the comment was harmful, untrue, or unhelpful. Personal identifying information is masked automatically when the feedback comment is submitted.
Most used bots
This list displays all of your bots sorted by the number of users they have. Each item on the list includes the bot's name, whether the bot is Generative AI or Rules-based, the number of users, conversations, and likes and dislikes. You can select the filter icon to change the sort order.