fix: prefer company over uni on overlap during affiliation (CM-1216)

[skwowet]

Summary

Affiliation logic now treats the email on an activity as the primary signal for determining which organization should be credited for that activity. When an activity does not contain an organization email, overlapping university and company affiliations are resolved in favor of the company.

This behavior now applies consistently during both ingest (data_sink_worker) and bulk refresh (refreshMemberOrganizationAffiliations).

Context

Member activities can be affiliated to an organization based on work history (memberOrganizations), manual segment rules, and other fallback logic. That generally worked well for simple timelines, but broke down in a couple of cases:

• A member had overlapping university and company roles (e.g. student + internship).
• An activity contained an explicit organization email (user@company.com, user@school.edu), but date-based affiliation logic resolved the activity to a different organization.

The email used on an activity is first-party evidence of which organization the activity was performed on behalf of. It should take precedence over enrichment-derived timelines and generic tie-breakers (stint length, org size, source rank), unless a manual affiliation override has been configured.

What changed

Ingest (findAffiliation)

Affiliation resolution now follows this order:

1. Manual segment affiliations (unchanged).
2. Activity email domain — if the activity username contains an organization email, affiliate to the memberOrganization whose organization has a matching verified primary-domain.
3. Date-based matching — for activities without an organization email (plain handles, public inbox domains), affiliate using the organization active at that timestamp. When a company and university overlap, the company is preferred.
4. Existing undated fallbacks (unchanged).

Only the email present on the activity username is considered. Member profile emails are not used as a substitute when the activity itself does not contain an email.

Bulk refresh (prepareMemberOrganizationAffiliationTimeline)

Bulk refresh now mirrors the same behavior used during ingest.

To avoid activities being processed twice, refresh runs in two passes:

Pass	Activities	Logic
Email-based	Username matches a member organization's verified domain	Affiliate directly to that organization
Timeline-based	No email, or no matching member organization	Use date-based affiliation logic, preferring companies over universities when overlaps exist

Manual segment affiliations continue to take precedence through the existing segment-scoped timeline and skipManualAffiliationSegments guards.

Expected behavior

Activity signal	Result
Username contains an organization email (@company.com, @school.edu)	Affiliate to the organization with the matching verified domain
No organization email, company + university overlap	Company wins
Manual segment affiliation	Manual rule wins for that segment

[Copilot]

Pull request overview

This PR adjusts member activity affiliation to prioritize the email domain present on an activity’s username as the strongest routing signal, aiming to prevent “university vs company” overlap cases from incorrectly affiliating to universities when there’s no email-domain evidence.

Changes:

• Added DAL support for fetching verified primary domains per organization and introduced a heuristic to prefer company affiliations over university affiliations during timeline overlaps.
• Updated affiliation resolution to (a) route by activity email domain when present, and (b) otherwise fall back to timestamp-based member organization timelines with a company-over-university preference.
• Updated ingest-time activity processing to derive the affiliation domain from the activity username (when it is a valid email) rather than from the member’s verified email identities.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
services/libs/data-access-layer/src/organizations/identities.ts	Adds org-domain fetch + overlap preference helper used by affiliation logic.
services/libs/data-access-layer/src/members/segments.ts	Changes work-experience query to enforce verified-domain matching when an activity domain is provided.
services/libs/data-access-layer/src/member-organization-affiliation/types.ts	Extends timeline items with domain-based routing fields.
services/libs/data-access-layer/src/member-organization-affiliation/index.ts	Adds domain-partitioned refresh passes to keep email-domain activities disjoint from timeline-based passes.
services/libs/common_services/src/services/common.member.service.ts	Updates per-activity affiliation decision flow to prioritize activity email domain and apply company-over-university preference.
services/apps/data_sink_worker/src/service/activity.service.ts	Switches ingest-time domain extraction to use the activity username (when it’s a valid email).

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated 2 comments.

[Copilot]

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated 2 comments.

[Copilot]

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated 2 comments.