(DE1.1): If we surface the availability of edit suggestions on articles to logged in readers through a controlled experiment, there will be a ≥# increase in the proportion of distinct registered users with ≤100 cumulative edits who publish at least one constructive edit (mobile).

What is it?

As a logged-in Wikipedia reader who has edited before, does not do so regularly, and is arriving on-wiki seeking to learn about something/someone (broadly defined), I would value knowing when there are clear, actionable, and personally relevant improvements available within the article I am reading so that I can easily decide which (if any) I'd like to act on in this moment without needing to seek them out myself.

What does done look like?

[acceptance criteria]

Timeline & milestones

[August 1, 2026] - Launch controlled experiment

Open Questions

1. Will the count of suggestions be specific to the entire article? The section you're viewing? Something else?
2. How should the UI behave when an edit has happened in the time between now and when suggestions were last computed?
- Related: how should the UI behave when something occurred (e.g. an outage) that causes the system to miss edit event data?
3. How (if at all) will we account for the possibility that one of the suggested sources is wrong because the logic within one of those sources changed? E.g. template changed, a suggestion/check configuration was changed? Etc.
4. What (if any) data will we retain about the suggestions that are generated? For how long? Context here.
5. What (if any) ability will we have, in Phase 1, to know where (e.g. what section(s)) suggestions are present within?

Cross-team dependencies & consultations

Dependencies

Editing – introduce a technical framework for the interchange of edit suggestions
Editing – generate suggestions asynchronously, at scale
Search – TBD
SRE –
Data Platform – create an Edit Suggestion dataset
Reader Experience: align on UX

Consultations

Growth, ML, ModTools, Connection: consult about suggestion technical framework and data stream

Existing Artifacts

What does it block?

[add teams, products, projects, features, etc that will remain blocked if we take no action]

Related Objects
Search...

Status	Subtype	Assigned	Task
Open		None	T360489 [EPIC] Offer edit suggestions as a service
Open		None	T416531 [EPIC] Surface Edit Suggestions outside of VE
Open		None	T431076 [EPIC] Surface specific edit suggestions within articles (Phase 2)
Open		ppelberg	T428816 [EPIC] Surface edit suggestion counts to logged-in editors on articles (Phase 1)
Open		None	T409589 [EPIC] Expose the presence of edit suggestions to people while they are reading Wikipedia
Open		None	T410081 [Suggestion/Signal] Surface information about readability
Open		None	T428843 Suggest people upload photos on articles where images have been requested
Open		ppelberg	T429554 Publish project page for surfacing suggestions to logged-in editors
Open		ppelberg	T430712 [EPIC] Launch controlled experiment of surfacing suggestion counts on articles to logged-in editors
Open		None	T426778 [EPIC] Streamline the suggestion browsing and publishing experiences
Open		Esanders	T425597 Suggestion Mode feedback toasts block me from advancing to next suggestion
Open		dchan	T423465 [SPIKE] Investigate creating a top-level model class for Edit Check
Open		GGoncalves-WMF	T428814 [EPIC] Ready server-side suggestions PoC for use in production
Open		ppelberg	T432266 [SPIKE] Define scale requirements for suggestion count controlled experiment (logged-in users)
Open		ppelberg	T432272 Add instrumentation to log suggestion counts shown in read and edit modes
Open		None	T432274 [Suggestion Counts on articles] Report on leading indicators
Open		None	T432288 [SPIKE] Determine cache invalidation approach for on-wiki config changes
Open		None	T433607 Provide a way to query the LAC without triggering a compute job on a cache miss
Open		MNeisler	T432427 Publish experiment plan: read mode suggestion counts (logged-in)
Open		Ottomata	T432733 Server Side Edit Suggestions - Linked Artifacts Cache precompute job
Open		bmartinezcalvo	T412373 Suggestion Mode: Prioritize suggestions shown to newcomers in Suggestion Mode for the first time
Open		None	T432622 Prompt people to publish after completing first suggestion
Declined		None	T432759 Conduct usability testing of new post-Suggestion publishing prompt
Open		bmartinezcalvo	T412389 Suggestion Mode: Conduct moderated usability testing to understand how users use Suggestion Mode
Resolved	BUG REPORT	medelius	T413280 Suggestion mode - Don't expand any cards until clicked
Open		None	T428193 Suggestion Mode: Entry point in read mode
Open		cjming	T431493 [DE 1.1.9] An instrument that produces edit_survived events
Open		None	T433530 Backfill Edit Suggestion Counts in Linked Artifacts Cache for technical pilot
Open		None	T432286 [MILESTONE] Run technical pilot to evaluate Headless VE resource requirements
Open		Esanders	T428811 [EPIC] Deploy headless VE
Open		Esanders	T428491 [EPIC] Build a headless mode for EditCheck that scrapes suggestions and outputs them via an API
Resolved		Esanders	T431326 Decide on mechanism by which we will trigger headless VE
Resolved		Esanders	T431492 Server Side Edit Suggestions - Headless VE - LAC gRPC lambda interface
Open		None	T431495 Server Side Edit Suggestions - Headless VE - request tracing
Open		Esanders	T431541 Headless VE: Allow the user to pass in which wiki to fetch from with each request
Open		Esanders	T431547 Headless VE: Create config endpoint so check account configs can be cached
Open		Esanders	T431726 Consider performance improvements for headless VE
Duplicate		None	T431497 Deploy Headless VE in k8s for technical pilot
Open		Ottomata	T433514 Headless VE - Docker image pipeline
Open		Eevans	T431499 Linked Artifact Cache - connect to Headless VE suggestion counts lambda
Open	Spike	Esanders	T431602 [SPIKE] Identify candidate suggestions that can be extracted from VE
Open		None	T431603 Investigate making TextMatchEditCheck run independently of VE
Open		None	T432046 Propose parameters for headless VE pilot
Open		None	T432708 Application Security Review Request : EditCheck server-side suggestions
Open		None	T432715 Satisfy SCROLL soft launch requirements: EditCheckHeadless
Open		None	T434109 Create Headless VE Helm chart & deployment
Open		dchan	T434129 Create a wikitech page for EditCheckHeadless
Open		dchan	T434134 EditCheckHeadless: Ensure maintainance contact details are correct
Open		jhathaway	T433024 Implement sampling requirements for Headless VE technical pilot
Open		None	T431320 Resolve tension between prominence of 'Help' and 'Suggestion Mode'
Open		None	T432747 [SPIKE] Define SRE support level and roles for production components of server-side suggestions

Event Timeline

ldelench_wmf created this task.Jun 10 2026, 9:05 PM

Restricted Application added a subscriber: Aklapper. · View Herald TranscriptJun 10 2026, 9:05 PM

ldelench_wmf added a subtask: T428814: [EPIC] Ready server-side suggestions PoC for use in production.Jun 10 2026, 9:06 PM

ppelberg updated the task description. (Show Details)Jun 17 2026, 5:12 AM

ppelberg added a subtask: T409589: [EPIC] Expose the presence of edit suggestions to people while they are reading Wikipedia.

ppelberg added a subtask: T428193: Suggestion Mode: Entry point in read mode.Jun 17 2026, 5:14 AM

ppelberg created subtask T429554: Publish project page for surfacing suggestions to logged-in editors.Jun 17 2026, 10:19 PM

ppelberg renamed this task from [EPIC] Surface availability of edit suggestions to logged-in readers (controlled experiment) to [EPIC] Surface availability of edit suggestions to logged-in editors (controlled experiment).Jun 18 2026, 3:53 PM

ppelberg updated the task description. (Show Details)Jun 18 2026, 5:04 PM

Will the count of suggestions be specific to the entire article? The section you're viewing? Something else?

I have been it would be per suggestion type, but for the entire article.

What (if any) ability will we have, in Phase 1, to know where (e.g. what section(s)) suggestions are present within?

I think none. This assumes that we have a data model that can associate suggestions with specific content, which is for Phase 2.

What (if any) data will we retain about the suggestions that are generated? For how long? Context here.

Retain where? ;)

If for use in product feature: this would complicate things, but we can think about it if you really need it.

If for offline use by internal WMF for analysis, etc: We will produce the edit suggestion data changes to Kafka and the Data Lake.

ppelberg updated the task description. (Show Details)Jun 30 2026, 3:59 AM

ppelberg renamed this task from [EPIC] Surface availability of edit suggestions to logged-in editors (controlled experiment) to [EPIC] Surface availability of edit suggestions to logged-in editors on article.Jun 30 2026, 4:32 PM

ppelberg renamed this task from [EPIC] Surface availability of edit suggestions to logged-in editors on article to [EPIC] Surface availability of edit suggestions to logged-in editors on articles.

ppelberg created subtask T430712: [EPIC] Launch controlled experiment of surfacing suggestion counts on articles to logged-in editors.

ppelberg renamed this task from [EPIC] Surface availability of edit suggestions to logged-in editors on articles to [EPIC] Surface edit suggestion counts to logged-in editors on articles (Phase 1).Jul 3 2026, 5:38 AM

ppelberg mentioned this in T431076: [EPIC] Surface specific edit suggestions within articles (Phase 2).Jul 3 2026, 5:50 AM

ppelberg edited parent tasks, added: T431076: [EPIC] Surface specific edit suggestions within articles (Phase 2); removed: T416531: [EPIC] Surface Edit Suggestions outside of VE.

ppelberg mentioned this in T432184: [SPIKE] Articulate strategy for lowering likelihood of suggestion false positives.Tue, Jul 14, 10:43 PM

ppelberg removed a subtask: T428814: [EPIC] Ready server-side suggestions PoC for use in production.Tue, Jul 14, 11:04 PM

ppelberg created subtask T432286: [MILESTONE] Run technical pilot to evaluate Headless VE resource requirements.Wed, Jul 15, 7:30 PM

Esanders mentioned this in T432715: Satisfy SCROLL soft launch requirements: EditCheckHeadless.Tue, Jul 21, 11:08 AM

Ottomata mentioned this in T432733: Server Side Edit Suggestions - Linked Artifacts Cache precompute job.Tue, Jul 21, 2:01 PM

GGoncalves-WMF added a subtask: T432747: [SPIKE] Define SRE support level and roles for production components of server-side suggestions.Tue, Jul 21, 4:30 PM

ppelberg removed a subtask: T428193: Suggestion Mode: Entry point in read mode.Fri, Jul 24, 6:03 PM

gopalakrishna_a subscribed.Tue, Jul 28, 6:58 AM

Since the wikis are still TBD, so one question while that is open.
How many of the checks work outside English? Revise Tone uses a BERT and Reference Check looks language-dependent. While Add-a-Link and Add-an-Image already run on many wikis. If most of the inventory is English-only, a wiki with two check types and a wiki with six are not running the same experiment.
I am a volunteer and have done mobile contribution research with CIS-A2K in India. Report published on diff. If Telugu or Kannada are in scope, I can run moderated sessions with editors under 100 edits on their own phones, in the same window as the experiment, reported against the metric in the hypothesis.

Johannnes89 subscribed.Wed, Jul 29, 6:22 AM

[EPIC] Surface edit suggestion counts to logged-in editors on articles (Phase 1)Open, Needs TriagePublicActions

Description

Hypothesis