In the current instrumentation it's easy enough to count how many image captions users (who have opted in to sharing usage data) are publishing by just counting how many action = 'saved' AND wikidataDescriptionEdit IN('new', 'existing') AND pageNS = 'File' events there are – for example:

{
  "schema": "MobileWikiAppEdit",
  "revision": 18115551,
  "wiki": "ruwiki",
  "event": {
    "action": "saved",
    "wikidataDescriptionEdit": "new",
    "anon": false,
    "pageNS": "File",
    "client_dt": "2019-12-10T16:59:00-0500",
    "app_install_id": "[REDACTED]",
    "session_token": "11372ffc-8c68-4434-b1ce-de391106cd90"
  }
}

T239706 asks for a breakdown by whether the caption was made from the article view or the gallery module, which is currently not possible.

Proposal: a source field which would have values:

• "article"
• "gallery"
• "suggested edits"

Note: "leadImage" is the source for captions added from the CTA that appears from an article.

(Feel free to edit if there's a place in the app that an image caption could be initiated from that's missing from the list.)

• I recommend this field to be a flexible string – not an enum – to make it easier to capture additional sources of edits in the future (e.g. "section").

• For some reason cardClicked events aren't generated when clicking on FeedContentType.SUGGESTED_EDITS cards (but seemingly only those, clicks on other cards seem unaffected).