The Tasks dashboard presents you with content in your knowledge base that needs your attention. The default dashboard filters apply, but all widgets are also filtered to provide the most relevant information.
All the information available in this dashboard is taken from other dashboards with different filters applied, so if you would like to investigate some of these issues for your project more in-depth, you can make use of the original dashboards.
Read more about the different tasks in this dashboard below.
Task: Check Recognition/Rewrite - These Answers Get the Most Negative Feedback
This table shows the top 5 Q&A, dialog and FAQ answers with negative feedback. You can drill into the user questions and feedback comments to get insight in what causes the large amount of negative feedback.
This table may be empty if no answers have received negative feedback, or if the answers that did receive negative feedback were not shown often enough to be relevant.
On dialogs and FAQ, your customers may be interpreting the (correct) answer as uninformative or to be using the wrong tone. In that case, consider rewriting the answer to make it more appealing in general. This can also be the case for Q&A answers, but here you should also check that the answer is actually an answer to the questions being asked. If it isn't, two things might be happening:
The recognition is OK: (Nearly) this exact user question is on the Q&A and it was supposed to get this answer, but it doesn't match the intended question. Move the question to another Q&A with a better answer (if available) or create a new Q&A with a new answer.
The recognition is not OK: The user question matches on some words of a question in the Q&A, but it wasn't supposed to end up there. If users regularly ask this question, you should add this user question to Q&A with a more suitable answer.
If you wish to investigate more deeply into Q&A or answers that are receiving negative feedback, check the Answers or Feedback dashboards.
Note: This table does not include the answer Conversational AI Cloud gives when a Q&A question is not recognized at all. The Tasks: Q&A recognition dashboard includes suggestions for improving that recognition and the Q&A questions dashboard can also provide you with user questions without any recognition.
Task: Rewrite - The Links in These Answers Are Hardly Ever Clicked
In this table you'll find the top 5 Q&A, dialog and FAQ answers including links that don't get many clicks. You should consider rewriting these answers to either:
Provide the information in the answer itself and eliminate the need to follow the link.
Make the function of the link(s) in the Answer more clear. Either the answer or the visual (i.e. the link text) itself should ensure people feel compelled to click the link.
The widget allows you to drill into some of the questions and feedback on the answers to help you evaluate which course of action to take.
If you wish to investigate Link Clicks further, use the answers dashboard.
Task: Review - These FAQ Are Clicked Less Than 1% of the Time
This table shows you the top 10 shown FAQs with a click percentage smaller than 1%. These FAQs were clicked less than once for every 100 times they were shown.
You should consider whether these FAQs are shown on the correct pages or that they relate to the right Q&A. To determine this, the dashboard allows you to drill into the category the FAQ is assigned to. You should also consider the phrasing of the FAQ question: is this question, phrased this way, something your customers would be wondering about?
We determine whether FAQs are shown through requests to the system to retrieve FAQ. If you are not showing FAQs anywhere but you still see data here, please check with the person responsible for the implementation of Conversational AI Cloud on your website or application.
To investigate FAQ further, use the FAQs Dashboard.
Task: Review - These Dialogs Have a High Dropout Rate
In this table you can find an overview of (at most) the Top 5 started dialogs for which the dropout rate in a session is higher than 60%. Your users apparently have trouble finishing these dialogs.
You should review the dialog flow to find the large drop-off points and possibly adjust either the answer or the structure of the dialog at that point. Please use the 'Go to Diagram and Nodes' option to see the dialog flow for this dialog.
There are various reasons why people might not continue in a dialog:
You are offering too many options: Your customers get lost in the number of choices you present. In the flow, you will likely see some options that are never or rarely used. Consider removing these options or restructure the dialog so you're only offering a few options at a time.
You are offering too few options: Your customers don't see an option that matches their needs. Check the feedback comments on the Dialog or check the regular Q&A questions for what people are asking on this subject. Add the missing options, but be careful not to offer too many.
Continuing in a Dialog is not necessary: If users find the information they're looking for before reaching one of the final nodes in the Dialog, there will be no need to continue. For example, in a Dialog where the final question is 'Did this succeed?", users who succeeded are unlikely to continue. Restructure the Dialog to avoid this kind of question. Also, if you are offering links in your Dialog, but the Dialog has not yet reached one of the final nodes, your customers may be clicking the link instead of continuing the Dialog.
If you wish to investigate other Dialog flows, you can find these on the Dialogs dashboard.
Task: Rewrite - These Dialog Options Are Too Long
We've found that generally, offering very long texts as choices in dialogs means your users will not go through the dialog as easily as they could. To that end, this table presents you with some of your dialog options that are very long. You should consider shortening them.
You will also notice long dialog options when you investigate the flows of badly-performing dialogs.