The report of the analysis is available at this link. To view, please download the HTML locally and open it with any web browser. I will publish it publicly after privacy review.~

Full report at https://analytics.wikimedia.org/published/reports/content_translation/analysis_of_cx_suggestions_T369696_Jul2024.html

Summary
The key takeaway is that user interactions with the currently available suggestion features is low. This is primarily due to users who come from an external entry point where an article is already selected, they are automatically progressed to the translation start screen. 90% of the sessions took this path, which minimized exposure of any features present on the translation home. For new feature additions, suggestions, or others, it would be important to have more entry points where users reached the translation home without the article being already selected, or think about ways of bringing exposure to suggestion features for users coming through an external point.

Of all interactions with suggestions related to features, more than 50% of the events were triggered by users with 1000+ experience. In the previous funnel analysis, we observed that as user experience increased, they opened the translation dashboard directly more, rather than from an external entry point. Mostly for starting a translation, users with less experience accepted a suggestion shown because there was no seed article, whereas for experienced users, it was from suggestions related to their recent edit.

When users discarded a suggestion, in 72% of the cases they discarded the next suggestion as well, whereas 14% selected an article and proceeded to the translation. When users refreshed the whole suggestions list, 50% refreshed again, whereas 21% accepted an article from the new list of suggested and proceeded to translation. The frequency of users either discarding a suggestion or refreshing the suggestion for most sessions was only once or twice, which is good.

73% of users who selected an article to translation from the suggestions list, made at least one edit to the translation.

From the current available instrumentation, it is not possible to answer the all proposed questions. Some of the key steps in the user flow as such publishing the translation is not currently instrumented. For a comprehensive understanding of CX user flows, the tasks T357250, T357253, T357252, T357254, & T357255 should be prioritized.

@Pginer-WMF I have added an addendum to the previous analysis to further investigate users entering the dashboard from Special:Contribute entry point.

Summary
On mobile web, which is also the most commonly used access method on the four wikis under consideration, the contribute page accounted for 12% of the times the content translation dashboard was opened. While the frequent languages and the content languages selector were the most commonly used entry points, the contribute page was also substantially utilized. On desktop, it was used in about 19% of cases.

Interestingly, the contributions page entry point was most used by users with 100–999 edits to open the translation dashboard home page, with a significant difference compared to other groups. The use of the contributions page entry point increased as users gained more experience, which may indicate that as users become more engaged with contributing to the projects, they might visit the Special:Contribute page more often. However, after reaching a certain level (1000+ edits), its usage decreases. As observed in the funnel entry points analysis, users with more experience tend to open the dashboard directly, compared to other entry points.

In the majority of cases (55%), users who opened the translation dashboard through the contributions menu proceeded to start a translation by selecting an article from the search results. Their interactions with suggestion-related events (discard/refresh) were low, with approximately 10 events recorded. In 12% of the cases, they accepted a suggestion (either because it was related to a recent edit or due to the absence of a seed article) to start a translation. Note: This means that the users directly accepted a suggestion without discarding or refreshing.

Compared to the overall funnel, users who opened the dashboard through Special:Contribute had substantially less back-and-forth movement between various steps. For example, moving back and forth between the home and the translation start screen. They also had significantly higher editing activity after opening the translation editor. In 93% of the cases, they added at least one translation.