Feature Request

In the Location-Based Analytics module: The outer (summary) view displays deleted articles if they were viewed prior to deletion. However, in the drill-down view, details of deleted articles cannot be fetched. The article order in the drill-down view is retained from the outer view, even though deleted article details are unavailable. This behavior can cause confusion for users when certain articles appear in the summary but are not accessible in the detailed breakdown. This feature request proposes adding a clear informational banner in the UI to explain this scenario to users.

Customer reported that Eddy Analytics does not clearly differentiate between conversation count and query count. Currently, multiple queries within the same conversation share a single conversation identifier, which causes confusion in analytics and CSV exports. The customer expects conversations and queries to be displayed separately, where each conversation thread (only with multiple queries) is counted once under Total Conversations and each individual question is counted under Total Queries. An enhancement to clearly distinguish conversation level and query level metrics in the Analytics dashboard and export reports would improve clarity and reporting accuracy.

There is a discrepancy between the data displayed in the Overview section and the Drill-down section of Reader Analytics, particularly impacting JWT readers. Currently, the Overview aggregates data by index-type category (including reads, views, and time spent), whereas the Drill-down section does not include this aggregation logic. As a result, analytics counts for the same user differ across these two views. This inconsistency becomes more noticeable when applying filters such as custom date ranges, leading to confusion and reduced reliability of the analytics data. The expectation is that both Overview and Drill-down sections should follow consistent aggregation logic to ensure aligned and accurate reporting across all views.

Currently, on the tiles section in analytics we show count from all articles. However, on click of "View All" we show only active articles count in the drill down section. This creates a mismatch in the count displayed on both section.

There is a discrepancy in data which is observed for readers in the analytics section. When viewing the data (views, reads) in the readers section it shows a different value than viewing the data for individual readers even though we select the same filter.

Currently, links pointing to custom domains associated with the platform are being classified as external links. However, these custom domains are configured and managed as part of the same system environment. This feature request proposes enhancing the link classification logic to treat configured custom domain URLs as internal links instead of external links.

Currently, the link analytics tool validates links based solely on HTTP response status codes. If a link redirects to a login page or a restricted/private page that returns a 200 OK response, it is incorrectly marked as valid. This results in links that are effectively inaccessible to readers not being flagged, leading to inaccurate reporting and poor user experience. This feature request proposes enhancing the link validation logic to detect and flag such inaccessible links correctly.

We would like to see more information within the article analytics page. An additional tab alongside views, reads, likes, dislikes, feedback etc. that shows how many times an article was referenced in AI answers, and how many times the article has been clicked on from an answer. When clicking on an article title within the analytics to see who is included in the views, reads, likes, dislikes etc., any site contributors, admin, or owners are not included in the data. It would be helpful to see them included to reduce confusion when numbers don't add up. Including the feedback metrics when clicking on an article title to see reader info would help us track who is submitting feedback on which articles. Currently we have to search through other tabs in order to see who has submitted feedback and on what article.

In some cases, keywords that are available in multiple articles are being recorded under the No Results Found analytics. For example, the keyword "Porting" exists in several articles but is still captured as a “no results” search. This behaviour creates inaccurate analytics and confuses customers, since valid keywords should not be tracked as failed searches. The issue seems to occur when users apply certain search filters or advanced search combinations, and the filtered list of articles does not have the keyword searched in it, which causes the system to incorrectly categorise the search result.

We would like to see the following on the EddyAI analytics page: Being able to filter these analytics by custom dates, not just by month. It is difficult to get a full picture of the analytics with such limited windows. Dates of queries as part of the analytics bar, not just when you click on the query details, and being able to filter queries by date. Being able to filter analytics by user/reader info. Being able to filter analytics by query topic. When pressing the arrow on a query, it would be helpful to have the drop-down show the answer provided by EddyAI, not just the question again. There are too many steps to take to see what answer was provided and when it was searched.