fix(signals): merge survey answers per question in the surveys scout recipe

[posthog]

Problem

• The surveys scout can miss survey feedback. When a later event in a submission omits an answer, the scout reads that answer as empty.
• Before this change, response-querying.md kept only argMax(uuid, timestamp) for each submission and read the answers from that one event.
• Partial-response events and manual captures can put each answer on a different event. The scout then dropped the earlier answers from its rating trends and theme reads.
• This is a defect in the scout's query recipe only. Stored responses and SDK events are correct.

Origin

• Scout: a custom scout
• First signal: 2026-10-13
• Inbox report: open
• Task started by: auto-start, after the report was rated P3 and ready to fix

Changes

• The recipe now groups survey sent events by submission and keeps the latest nonempty value of each question (argMaxIf(answer, timestamp, answer != '')). This is the same merge that fetch_response_rows() in the surveys responses API uses.
• The recipe keeps the existing rules: the modern and legacy key coalesce, the $survey_id filter, the uuid fallback when $survey_submission_id is missing, and $survey_iteration as a grouping boundary.
• The recipe reads completion separately: a submission is complete when one or more of its events is not marked $survey_completed: false.
• The recipe gives a rule for deleted answers. An empty value and an omitted key are the same, so an earlier answer stays. A cleared answer can stay, but a given answer never gets lost.
• The rating-trend and open-text queries now read from one per-submission base subquery.

Case	Old recipe (latest event)	New recipe
Answers split across events	keeps the last event's answers only	keeps every answer
Cumulative events	correct	correct
Changed answer	later value	later value
Later event has an empty answer	empty	earlier answer
No submission ID	one row per event	one row per event

How did you test this code?

• Ran a synthetic HogQL fixture through execute-sql. It had complementary partial events, cumulative events, a changed answer, a later empty answer, and legacy events without an ID. The merge gave the expected value in each case. The old latest-event read returned empty for the later-empty case.
• Ran the exact rating-trend SQL against events with a placeholder survey ID to confirm that it compiles.
• On one survey with multi-event submissions, the merged row count matched the count of distinct submission keys, and the completion flag separated complete from partial submissions. Only aggregate counts were read.
• posthog/test/repo_invariants/test_signals_scout_tool_citations.py passes, and Prettier is clean.

Test rationale: This change edits only a Markdown prompt reference. No test reads the content of a scout reference, so the fixture check above stands in for a unit test.

Release status

• No feature flag controls this change

Docs update

None. The edited file is the scout's own reference. lazy_seed mirrors it to teams that have not changed their copy.

🤖 Agent context

Autonomy: Fully autonomous

Agent: PostHog Desktop (Claude Code), claude-opus-5-5

• Started from an inbox report about the surveys scout recipe. Checked scout notes and the scratchpad first, and found nothing that blocks this area.
• Ported the merge from products/surveys/backend/responses/fetch_rows.py and did not design a new one. This keeps the scout and the product in agreement.
• Left products/surveys/backend/util.py (get_unique_survey_event_uuids_sql_subquery) unchanged, because the report scopes the fix to the scout reference.
• Duplicate search: gh pr list --state open --search "survey scout response" found no PR for this defect.
• All fixture data is invented. Nothing from the session is in this PR.

---

Created with PostHog Desktop from this inbox report.

🤖 Generated with Claude Code

[trunk-io]

😎 Merged successfully - details.

[github-actions]

🤖 CI report

⚠️ Trunk lane — backend Python lane

This PR is assigned to the backend Python lane. It runs backend Python tests and may merge in parallel with PRs in other lanes.

[stamphog]

Approved.

Contained scout-reference update with no open review concerns or identified showstoppers. The platform-attested Inbox provenance exempts the machine author from the bot-author refusal rule.

• PostHog Review (Standard) reviewed the current head: no findings.

Gate mechanics and policy version

Gate		Result
prerequisites	✓	all clear
deny-list	✓	no deny categories matched
size	✓	0L, 0F substantive, 110L/1F incl. docs/generated/snapshots — within ceiling
tier	✓	T0 auto-approve: T0-deterministic (110L, 1F, single-area, fix)
stamphog 2.5.0		.stamphog/policy.yml @ a08353f · reviewed head a08353f

[andrewm4894]

/trunk merge

[deployment-status-posthog]

Deploy status

Environment	Status	Deployed At	Workflow
dev	✅ Deployed	2026-10-14 06:20 UTC	Run
prod-us	⏳ Pending
prod-eu	✅ Deployed	2026-10-14 06:34 UTC	Run