Hi Megan, I was cleaning up my backlog and assigned this to you :)

I've reviewed the data logged in editattemptstep and visualeditorfeatureuse and confirmed the new events instrumented in T243641 are logging mostly as expected.

@DLynch Just one question/potential issue:
All init_mechanism = 'tooslow' events to date have been logged with init_type = section. Is this expected? Or should this init_mechanism type also occur for editing initiated for the whole page?

All other data appears to be logging as expected. See a summary of checks and some initial findings summarized below.

QA Check Summary
Note: Data reflects events logged from 1 November 2023 through 11 December 2023.

• Confirmed we are logging both the context-show and fallback-confirm actions associated with the fallback prompt feature. There are more context show events than fallback confirm events as expected.
• Users in about 5% of the editing sessions where the fallback prompt was shown elect to use the fallback editor.
• About 1.3% of all edit attempts during the reviewed time period included a fallback prompt event.
• Confirmed that fallback feature events only appear when the user is editing on mobile (platform = 'phone') and using VisualEditor (editor_interface = visualeditor)
• The fallback prompt typically only appears once in an editing session. Only about 2% of sessions were shown the fallback prompt more than once.
• I reviewed the order of events for a sample of editing sessions that included a fallback prompt event. The order of edit events appears as expected.
• init_mechanism = 'tooslow' in editattemptstep only appears with wikitext and mobile editing sessions.
• Confirmed that for all init_mechanism = 'tooslow' events, there is an immediately-preceding init event with the same page_token / session_token.
• As expected, the median load time for sessions shown the fallback prompt (4.3 seconds) is significantly higher than the median load time for sessions not shown the prompt (1 second)

As part of this task, I've also updated the vefu data dictionary with the new events added to track use of this feature.

Thanks @DLynch.

I believe we can resolve task this now as events have been verified. Note further analysis of this data will be done in T347437