Q&A Search Results
By changing the ‘Impact of user feedback to User Score’ value under Q&A configuration, you can let customers’ Q&A selections affect future Q&A search results. Let’s see how it works.
First, go to the Knowledge Base > Q&A and click the configuration icon to open the Q&A configuration. Scroll down to the Impact of user feedback to User Score.
The default value, for this setting, is zero (slowest), which means the customer’s selection doesn’t affect the Q&A search results. You can adjust the value as per your requirements. You can also reset all User Score values here.
Now let’s see how this changes the search result. When you enter a question in the search bar under Q&A, you’ll see the preview result for the Q&A search with Score and User Score.
The score value indicates how similar Alli’s AI thinks the Q&A is to the search query, before any feedback from customers. On the other hand, the User Score is the adjusted value of the Score after taking into account customers’ Q&A selections. You can notice how the User Score increases when customers select the Q&A after similar search queries and how it decreases if customers choose ‘None of the above’ when the Q&A is displayed as one of the options.
Since the Q&A with the highest User Score is displayed first, you’ll see that the order of the options changes as the User Score changes. If the User Score is high enough and is above the Direct Answer Threshold, then the answer will be displayed directly to customers with similar search queries.
Pretty smart, isn’t it? Please note that the selection in the user flow preview also affects the User Score. This means that if you don’t like certain Q&A search results, you can adjust the result manually by “teaching” the AI in the user flow preview.
Document Search Results
Now customers can give feedback to individual Documents search results, so the feedback adjusts the search result accordingly next time when a similar search is made. Let’s see how it works.
After a customer enters a search query to the Answer with Documents node in the conversation with Alli, thumb up / down icons will be displayed for each answer candidate. The customer can give positive feedback by clicking on the thumb up icon, negative by the thumb down icon.
When a customer gives a feedback to a certain answer candidate for a certain query, it affects future search results for similar queries. For example, let’s say a customer gave this feedback to ‘How often should the examination be performed?’ query:
In the dashboard, under the Knowledge Base > Documents tab, you can see that the Adjusted Score on the answer for that query went up, and it’s affecting the order of the search results, both on the dashboard and the conversation on the bot.
If you try a similar query, you can see that the Adjusted Score for the similar one is affected as well.
A few notes about the user feedback for Documents search:
- The amount of change in the adjusted score depends on the original Score, the number of feedbacks, and how similar the search query is with the original query that received the feedback.
- One customer can give multiple same feedback to one search result.
- When there’s any positive feedback, negative feedbacks won’t affect the Adjusted score.