Page MenuHomePhabricator

Create a dashboard to monitor impact and performance of different editing modalities
Open, HighPublic

Description

This task involves the work of enabling product leadership to grasp the impact of the structured editing modality contributor teams are investing in.

This modality is an effort to reduce the toil and confusion new(er) volunteers must currently overcome to contribute constructively to Wikipedia.

Requirements

Dashboard

  • Create a Superset dashboard that enables us to answer the following questions on an ongoing basis:
    • Among new(er) volunteers, what proportion of NS:0 edits are made through structured editing experiences, compared with open-ended editing experiences (see "Definitions" below)? How is this proportion changing over time?
    • Across all experience levels, what proportion of NS:0 edits are made through structured editing experiences? How is this proportion changing over time?
    • How is the total volume of NS:0 edits changing over time for the following categories?
      • All human edits
      • Edits by new(er) volunteers
      • Edits made through structured editing experiences
      • Edits made through structured editing experiences by new(er) volunteers
    • In the past 90 days, how many cumulative edits was a given structured editing modality acted upon?
      • In this context "a given structured editing modality" refers to the Edit Check, Edit Suggestion, or Task type. E.g. Reference Check, Paste Check, Tone Check, Suggested: revise tone, Suggested: add links, etc.
      • Ideally, this could be implemented as a table to start
  • The dashboard should support:
    • Viewing each metric by platform: mobile web, desktop web, both
    • Grouping/filtering each metric by whether edits were reverted (or not): mobile web, desktop web, both
  • Each metric should show:
    • The current reporting period value and y-o-y offset

See Newcomer tasks share of edits as a helpful point of reference.

Definitions
Structured editing experiences/modality
We will consider an edit as belonging to this "structured modality" if:

  • Edit was made to NS:0
  • Any of the following is true:
    • ≥1 Edit Check was activated during the course of the edit session
    • ≥1 Edit Suggestion was acted on during the course of the edit session
    • Edit was initiated and/or completed via Newcomer homepage (read: structured tasks / suggested edits)
WARNING: Peter is confirming the accuracy and exhaustiveness of the above.

New(er) editors
We will consider "new(er) editors" as people for whom any of the following is true:

  • People who are logged in and have published ≤100 cumulative edits at the time of publishing

Edits made via open-ended experiences
We will consider "edits made via open-ended experiences" as all human edits made via full-page editing experiences (e.g. VE, source, NWE).

Event Timeline

ppelberg updated the task description. (Show Details)
ppelberg updated the task description. (Show Details)

[unasked-for advice as part of my quest for better data] Do we have a more precise indicator of "made through structured editing experiences" than "≥1 Edit Suggestion was seen during the course of the edit session"? I'm assuming that means that someone clicked on the Suggestion to open up the details panel? But I wonder if we could go a bit further as I imagine folks will open up the Suggestions just out of curiosity. I know at least some of the suggestions have an action button and no button - could there be edit tags based on interaction with these instead? Or perhaps a more implicit "you opened the Suggestion and edited the element that was attached to it?"

As a note, agreed that triggering Edit Check is enough for it to count as relevant (given that it is quite tightly tied to the specific edit being made and far from a universal occurrence). Same with the Newcomer homepage tasks.

[unasked-for advice as part of my quest for better data] Do we have a more precise indicator of "made through structured editing experiences" than "≥1 Edit Suggestion was seen during the course of the edit session"? I'm assuming that means that someone clicked on the Suggestion to open up the details panel? But I wonder if we could go a bit further as I imagine folks will open up the Suggestions just out of curiosity.

@Isaac: great spot and question.

Following some offline discussion, @MNeisler and I have similarly come to think that, in the context of Edit Suggestions, someone needs to have acted upon the suggestion rather than simply having seen it.

I've updated the task description to reflect this.

And hey, "unasked-for advice" is something I/we very much welcome and more than that, value! As always, thank you for paying attention and lending your input :)

ppelberg moved this task from Editing-Q4-2025-2026 to Editing-Q1-2026-2027 on the Editing-team board.

Per today's offline discussions, we're going to work on this dashboard at the start of Q1 and narrow the scope of the initial implementation. See T428818 for more context.

Note: I've updated the description (T415718#12067790) per offline discussion with Selena wherein she helpfully identified the value in also tracking sessions when an Edit Check/Suggestion was activated.

Rationale: sometimes people internalize the feedback and make a change rather than use the UI affordance(s).

@ppelberg I'm starting work on this task and have a few clarifying questions noted below.

OPEN QUESTIONS:

  • Can you clarify the difference between the following two metrics listed in the task description? Based on the definitions provided, it seems to me that "edit suggestion/check was activated/triggered" aligns very closely with "structured editing experiences".
  • Across all experience levels, what proportion of NS:0 edits are made in which an edit suggestion/check was activated/triggered, compared with open-ended editing experiences? How is this proportion changing over time?
  • Across all experience levels, what proportion of NS:0 edits are made through structured editing experiences? How is this proportion changing over time?
  • Should this dashboard be limited to permanent, registered users? In other words, we would exclude logged-out and temporary accounts.
  • I noticed there is a current subtask T428818 to create a phase 1 of the structured editing experience dashboard. Can you confirm if I should still prioritize this subtask first?

@ppelberg I'm starting work on this task and have a few clarifying questions noted below.

OPEN QUESTIONS:

  • Can you clarify the difference between the following two metrics listed in the task description? Based on the definitions provided, it seems to me that "edit suggestion/check was activated/triggered" aligns very closely with "structured editing experiences".

Good spot. I think these two bullet points were trying to distinguish between a particular subset of structured edits – those made using an Edit Check or Suggestion Mode – and open-ended experiences.

With this said, I don't think that right now we need that level of granularity. Accordingly, I've removed the first and clarified the meaning of the second in the task description.

  • Across all experience levels, what proportion of NS:0 edits are made in which an edit suggestion/check was activated/triggered, compared with open-ended editing experiences? How is this proportion changing over time?
  • Across all experience levels, what proportion of NS:0 edits are made through structured editing experiences? How is this proportion changing over time?
  • Should this dashboard be limited to permanent, registered users? In other words, we would exclude logged-out and temporary accounts.

Can we include all three (registered users, logged out, and temp accounts) and have the ability to filter based on them? If this adds complexity, can we start with permanent, registered users and have the ability to filter by cumulative edit counts (e.g. 1-4, 5-99, 100+)

  • I noticed there is a current subtask T428818 to create a phase 1 of the structured editing experience dashboard. Can you confirm if I should still prioritize this subtask first?

Yes, please! I assume if we start there we can iterate on it over time to eventually deliver on what this task describes. Of course, please let me know if that's not the case.