Ideally, we want to collect users votes on a document and page ids associated with it. Based on this, I've created a first draft of the schema here: https://www.mediawiki.org/wiki/User:SSethi_(WMF)/Sandbox/user_feedback_schema.json.

Is this the right approach? Is there anything that I'm missing here?

The script for the gadget is getting ready here in case that might help understand what the flow for this process would be like cc @chelsyx

Hey @srishakatux , to help us prioritize this task, can you provide and organize the task description in the following format?

• What's requested. If you know what data you want, be specific! If you have a question but don't know how it can be answered, say what you've already tried.
• Why it's requested. You don't have to write an essay, but give enough context that the analyst can interpret, adapt, and prioritize your request.
• When it's requested. If you have a hard deadline, say what it is and what it's tied to.
• Any other helpful information, like relevant documentation.

See "How to request analysis" section at https://phabricator.wikimedia.org/project/profile/3313/

Thanks @chelsyx for sharing this information! I've updated the task description accordingly.

If you need any data from this schema to be kept beyond the 90 day window, you need to add it to the whitelist:

https://github.com/wikimedia/analytics-refinery/blob/master/static_data/eventlogging/whitelist.yaml

@fdans Thanks! We don't need to keep the data beyond 90-day window for now. Does the schema look good to you and is it ready to be published on meta-wiki and tested?

@srishakatux we reviewed the schema during our tasking meeting and it looks good! Feel free to put it in metawiki.

@srishakatux something to think about is overall participation versus overall impressions of survey. Example: "how many people have seen this question and have not answered" this will help you weight the data a bit better as 90% of "no" answers when only 1% of people have actually answered question is of little meaning . 90% of "yes" answers with 90% participation (impression-wise) is however very meaningful. Hopefully this makes sense, let us know otherwise.