Skip to main content

ClickUp QA Reviewer SOP

1. Purpose and Scope​

Purpose: Operational playbook for the QA Reviewer role — the person verifying that a client-facing or audited ticket actually meets its acceptance criteria + Evidence bullet before it closes. The rules layer (what to check, what counts as blocking, how to tag defects) lives in SC Contractor Task Quality — Review Standard. This SOP is the how — daily workflow, view setup, worked examples per domain, escalation ladder.

Who this is for:

Scope boundaries:

  • In scope: QA-Review subtask workflow (pre-release client-facing check) + post-hoc sampling workflow (Monday weekly audit). Per-domain feedback patterns. Escalation when defect rate trends adverse.
  • Out of scope: the rules themselves (Review Standard §Acceptance Criteria confirmation through §Approve vs. clarify), the time-tracking flow (timesheet approval SOP), workspace admin tasks (workspace setup SOP §9).

2. The two QA contexts​

The QA Reviewer operates in two distinct contexts, each with its own trigger and cadence. Both apply the same rules (per the Review Standard); the workflow shape differs.

ContextTriggerCadenceView used
QA-Review subtaskCreator added a qa-review-<parent-slug> subtask on a client-facing parent ticket; subtask was assigned to you and Assignee marked the parent ready for reviewContinuous — react when the subtask appears in your queueqa-review-queue (§3)
Post-hoc samplingWeekly Monday cadenceWeekly — random ~10–15% of previous week's closed ticketsweekly-sampling-window (§3)

Both feed into the same tagging discipline (qa:work-failed-review / qa:work-missed / qa:work-partial) and the same escalation ladder (§7).


3. View setup (do this once)​

Two saved views power the QA Reviewer's daily work. Build both at workspace level so they persist across sessions.

3.1 qa-review-queue — the pre-release queue​

Already documented in ClickUp Views and Filters Guide §6.5. Summary:

  1. Scope: Everywhere (workspace-wide).
  2. Filter: name contains "qa-review-" OR name contains "am-review-" (matches both current and legacy patterns during the transition).
  3. Filter: assignee = me.
  4. Filter: status is not complete.
  5. Sort: due date ascending; then parent task priority descending.
  6. Columns: Name | Parent | Status | Priority | Due date.
  7. Save: qa-review-queue, share Private.

3.2 weekly-sampling-window — the post-hoc sampling pool​

Run every Monday morning to surface the previous week's closures.

  1. Scope: Workspace (Client Delivery + SC Operations).
  2. Filter: status = complete AND date_done > today − 7 days AND date_done <= today.
  3. Filter: tag NOT contains qa:work-missed (these were already flagged in earlier sampling — don't re-pick them).
  4. Sort: random (use ClickUp's shuffle if available, OR sort by date_done and pick every Nth — see §5.2).
  5. Columns: Name | Client | Assignee | Date Done | Tags | Estimated Effort | Time Tracked.
  6. Save: weekly-sampling-window, share Private.

4. The QA-Review subtask workflow (continuous)​

When a parent ticket's QA-Review subtask appears in your queue:

4.1 The walk​

For every subtask, in this order:

  1. Open the parent ticket (not the subtask) — read the ## Acceptance Criteria block + the Evidence bullet.
  2. Open the deliverable named in the Evidence bullet (deploy URL, PR, document, screenshot, log link).
  3. Walk every AC bullet against the deliverable. Mark each: met / not met / partially met. (Per Review Standard §Acceptance Criteria confirmation.)
  4. Make the decision per the diagram above + Review Standard §Follow-up vs. send-back.
  5. Tag per Review Standard §Quality tagging.
  6. Close (or not) the QA-Review subtask per the path you chose.

The parent ticket cannot move to complete until the QA-Review subtask is closed — this is invariant per Ways of Working §2 Invariant 3. The Assignee waits on you.

4.2 Subtask lifecycle​

Note: "send back" doesn't move the subtask backward in ClickUp; it stays at in progress while the Assignee fixes the parent. The QA Reviewer re-checks once Assignee signals readiness again.

4.3 SLA target​

  • Response within 2 business days of the subtask appearing in your queue. Faster is better; slower than 2 business days is a signal to delegate.
  • Clarification responses count as response — you don't need to make the approve/send-back call within 2 days, you just need to engage.

If you're tracking against this SLA: filter qa-review-queue by date_created > today − 2 days and check the count. Anything pre-dating that window deserves a same-day response.


5. The post-hoc sampling workflow (weekly)​

5.1 When​

Every Monday morning. Block 30 minutes on the calendar. The cadence is per Review Standard §Post-hoc sampling.

5.2 How to pick the sample​

Open the weekly-sampling-window view (§3.2). Two approaches:

  • Random: use ClickUp's shuffle (if available) and take the first ~10–15% of the result. For a typical week of 20–40 closures, that's 3–6 tickets.
  • Stratified random: bucket by Assignee first (one per active contractor), then random within each bucket. Better signal when contractor mix is skewed.

Add targeted spot-checks on top of the random baseline when:

  • A new contractor's first ticket of the week. (Always sample.)
  • A high-value client deliverable shipped this week.
  • A domain where defects have trended up (per the escalation ladder §7).

Never replace the random baseline with spot-checks — bias destroys the telemetry value.

5.3 The walk​

Same as §4.1 (open ticket, read AC, open deliverable, walk each bullet). The fact that the ticket is already closed doesn't change the bar. Use Review Standard §Post-hoc sampling for the defect-handling decision (reopen vs. follow-up).

5.4 Tagging defects found in sampling​

  • Always apply qa:work-missed on a defect found in a closed ticket — that's the whole point of the tag.
  • Leave a one-line comment on the original: qa:work-missed — sampled YYYY-MM-DD, <one-line defect>, action: reopened (or action: follow-up filed <task_url>).

5.5 Tracking the sample​

At the end of the Monday session, post a comment on a recurring weekly sampling task (or in the team Slack channel):

Sampling YYYY-MM-DD (week of YYYY-MM-DD):
- Tickets in window: N
- Sampled: M (~X%)
- Missed defects (`qa:work-missed`): K (tickets: <links>)
- Tickets not ready (`qa:ticket-not-ready`): J (tickets: <links>)
- Notes: <anything worth flagging — patterns, domain trends, contractor calibration signal>

This is the audit trail. Without it, "I sampled this week" is unverifiable.


6. Per-domain worked examples​

Common defect patterns by domain, with the right tag + action.

6.1 WordPress​

Defect pattern: custom code where a plugin was specified. Example: AC says "use Smart Slider 3 for the carousel." Deliverable uses a custom-coded carousel. Action: send back with Blocking: Carousel uses custom code — swap for Smart Slider 3 per AC #3. Tag qa:work-failed-review.

Defect pattern: works on staging, broken on production. Example: deliverable URL is staging; client only sees production. Action: clarify — ask Assignee to either deploy to production or update Evidence bullet to point at production URL. No status change.

Defect pattern: page-builder lock-in. Example: AC didn't specify but deliverable uses Elementor when site is otherwise Block Editor. Action: follow-up (the lock-in is real but not in the original AC) — close original, file follow-up to migrate. Tag original qa:work-partial.

6.2 Copywriting / brand content​

Defect pattern: voice deviation. Example: AC says "match SC brand voice." Deliverable uses sales-y language with em-dashes for emphasis (per email-writing-style anti-patterns). Action: send back with Blocking: voice deviates from sc-brand-voice.md §3 (em-dash for emphasis, "absolutely!") — revise per voice doc. Tag qa:work-failed-review.

Defect pattern: missing CTA. Example: AC says "include a CTA to the demo funnel." Deliverable copy has no link. Action: send back with explicit AC violation. Tag qa:work-failed-review.

Defect pattern: factual claim without source. Example: deliverable claims "trusted by 50+ businesses" — no Evidence bullet supporting the number. Action: clarify — ask for the source. If not substantiable, send back with the claim removed.

6.3 GHL (workflows / funnels / automations)​

Defect pattern: trigger-tag misnamed. Example: AC says "trigger on tag lead-from-demo." Deliverable uses demo-lead. Action: send back with Blocking: tag name doesn't match AC — lead-from-demonotdemo-lead. Tag qa:work-failed-review. (Especially important because tag drift breaks downstream reporting.)

Defect pattern: untested in subaccount. Example: workflow deployed to client subaccount but Evidence bullet only shows screenshot from SC test subaccount. Action: clarify — request a successful test run in the client's actual subaccount with screenshot of the contact's automation log.

Defect pattern: webhook with no failure handler. Example: AC didn't specify error handling; deliverable webhook silently drops failed calls. Action: follow-up (out of original scope) but flag in approval comment. Tag qa:work-partial if the silent-drop is consequential; otherwise approve clean.


7. Escalation ladder​

When a single contractor's qa:work-failed-review rate or qa:work-missed rate trends adversely, the QA Reviewer escalates rather than absorbing the rework cost silently. qa:ticket-not-ready is not counted against the Assignee; a Creator whose ticket-not-ready rate trends adversely is raised with the Workspace Owner instead, because the fix is better tickets, not a contractor warning.

Thresholds are operational dials, not fixed rules. A 4-week rolling window is the right cadence for SC's task volume; the percentages are heuristic. The Workspace Owner can adjust per contractor based on engagement context (e.g., new joiner gets more headroom than a 12-month partner).

Levels in detail:

  • Level 1 — Calibration chat (informal). 30-minute 1:1 with the contractor. Walk the defects, name the pattern (e.g., "every WordPress task this month has had a custom-code defect"), and set a 2-week recheck. Don't escalate further if the pattern resolves.
  • Level 2 — Documented warning. Per Partner Engagement Framework. Contract addendum naming the defect pattern and the 30-day improvement window. Cc Workspace Owner.
  • Level 3 — Engagement off-ramp. Per Partner Engagement Framework §Termination. Workspace Owner makes the call. Always capture lessons (what did we miss in selection, what defect class is generalizable) and plan replacement before terminating.

Anti-patterns:

  • Silently absorbing rework week after week. The contractor never sees the pattern; you burn AM hours indefinitely.
  • Jumping straight to Level 3 on a single bad week. Sampling noise is real; the rolling window exists to filter it.
  • Escalating without documentation. Level 1 conversations are informal; Level 2+ MUST be written.

8. Common questions​

Q: Assignee marked parent ready for review but didn't @ me in the QA-Review subtask. Should I check the subtask anyway? A: Yes — the assignment is the signal, the @ is courtesy. Check the qa-review-queue view daily.

Q: I want to approve but I have one small advisory item. Do I add a tag? A: No. Tags are for blocking defects (qa:work-failed-review) or scope shortfalls (qa:work-partial). Advisory feedback gets a labeled bullet in the approval comment — no tag.

Q: The Assignee is on vacation. Their parent has open AC items that I'd normally send back. What do I do? A: Leave the QA-Review subtask in in progress with the blocking bullets clearly posted. Notify the Assignee's delegate per Role handoff. Don't approve to "clear the queue."

Q: I'm sampling and a closed ticket has no ## Acceptance Criteria block at all. A: Construct AC retroactively from the task description + chat history. Post as a comment ("Reading AC retroactively as: 1...2...3..."). Then walk through. Tag qa:work-missed if the AC walk now fails. Either way, tag qa:ticket-not-ready with the comment qa:ticket-not-ready — acceptance_criteria: none at creation: the missing block is the Creator's defect, recorded separately from whatever the delivery did.

Q: I'm the AM AND the Creator on a ticket I authored. Can I QA-Review my own ticket? A: No — the Creator and QA Reviewer roles should be separate per ticket. If you're the only available reviewer, defer the QA-Review to a delegate (per Role handoff) OR file as your own follow-up with a comment naming the self-review conflict.

Q: Contractor's time-tracked hours are way over estimate. Do I send back? A: No — time tracking is calibration data, not a failure condition. Apply the light-touch from Review Standard §Time-entry validation: >50% over with no explanation, ask in a comment. Approve the ticket regardless if AC are met.


9. References​

10. Revision History​

DateVersionChangeAuthor
2026-06-141.0Initial release. Operational SOP for the QA Reviewer role: two QA contexts (subtask + sampling), view setup, walk procedure for each context, per-domain worked examples (WordPress, copywriting, GHL), 3-level escalation ladder, common questions. Layers on top of the existing Review Standard; cross-links heavily without duplicating rules.Symphony Core Systems Team
2026-10-221.1Missing-AC sampling finding now tagged qa:ticket-not-ready instead of a sampling-notes mention. Escalation ladder states that ticket defects count against the Creator, not the AssigneeSymphony Core Systems Team