Skip to main content

ClickUp Ways of Working

Purpose​

The single doorway for every Symphony Core team member who creates, updates, or reviews tickets in ClickUp. Captures the rules and habits that make our workspace coherent: status hygiene, client identification, evidence, time tracking, tags, subtasks, assignment, views, and where each rule's authoritative source lives.

This is the breadth layer. Each section is short and points to the depth layer (the canon — SOPs, standards, working rules) when you need the full text.

Scope and audience​

  1. This document end-to-end (~20 min).
  2. Markdown Task File to ClickUp SOP — how tickets are authored.
  3. SC Contractor Task Quality — Pre-Work Standard — what makes a ticket reviewable.
  4. SC Contractor Task Quality — Review Standard — how reviewers act on completed work.
  5. ClickUp Views and Filters Guide — building and using saved views.
  6. ClickUp Timesheet Approval SOP — if you log billable time.

Roles​

Every ClickUp ticket is touched by up to four roles. Two are per-ticket; two are workspace-wide. Each role may be filled by a human or by an AI agent where noted.

RolePer-ticket or workspaceWhoWhat they do
Creatorper-ticketHuman (Account Manager, Workspace Owner, contractor) OR AI (Claude when authoring tickets from a markdown task file)Authors the ticket. Owns the description, acceptance criteria, Evidence bullet, tags, and initial routing (list, priority, assignee). The Creator is the one accountable for the ticket being well-formed at creation — see Pre-Work Standard.
Assigneeper-ticketHuman (one — see §8 single-assignee rule)Owns execution. Logs time, advances status, attaches Evidence, completes the work. Exactly one per ticket; if two people need to act, that's two subtasks.
QA Reviewerper-ticket (when present)Human (default: the Account Manager) OR AI (Claude when reviewing markdown-defined deliverables)Verifies that the work actually meets the acceptance criteria + Evidence bullet before the ticket can close. For client-facing tickets, this role is materialized as a qa-review-<parent-slug> subtask (see §7) and the parent cannot move to complete until the QA-Review subtask is closed. For internal/non-client tickets, the role is optional — self-attestation at close suffices and the AM samples post-hoc per the Review Standard.
ClickUp Adminworkspace-wideHuman (Workspace Owner)Owns workspace configuration: ClickApps, status templates, custom fields, space/folder/list structure, tag taxonomy, member/guest permissions, automations. Orthogonal to the per-ticket triple — the Admin does not act on individual tickets in this role. See SC Task Management Workspace Setup SOP.

Specialization: the timesheet approval flow uses QA Reviewer as the role-holder who reviews submitted timesheets for billing — see ClickUp Timesheet Approval SOP §3. Same role, narrower deliverable (the timesheet itself).

Legacy naming: before 2026-06-14 the per-ticket QA role was called the "AM-review subtask" because the AM was the default human. Existing am-review-* subtasks stay valid as-is (backward-compatible); new tickets use the qa-review-<parent-slug> name and treat the AM as the default-but-not-only-possible QA Reviewer.

Role handoff — vacation, sick, delegation​

Roles assume the role-holder is available. When they're not (vacation, sick day, end-of-engagement), pre-decide coverage rather than handling it ad-hoc.

Creator (per-ticket). Once the ticket is created, the Creator role is mostly fulfilled — the ticket's description, AC, and Evidence bullet are static artifacts. If the Creator goes on vacation, no handoff is needed for the ticket itself. If a follow-up clarification is needed and the Creator is unreachable, the Assignee or QA Reviewer can edit the description (note the edit in a comment naming the original Creator + reason).

Assignee (per-ticket). When an Assignee will be unavailable for more than one business day on a ticket they own:

  1. Update the ticket comment with: "OOO <start-date> to <end-date>. Current state: <one line>. Next step: <one line>. Reassigning to <delegate-name>."
  2. Reassign the ticket to the delegate. Per single-assignee-per-ticket, the ticket has exactly one Assignee — not two simultaneously.
  3. On return, the original Assignee can reassign back if they're picking it up, OR leave the delegate as Assignee if the delegate completed it. Don't reassign-back just to "own" a done ticket.

QA Reviewer (per-ticket, when present). The QA Reviewer is materialized as the qa-review-<parent-slug> subtask's assignee. Same OOO pattern applies — reassign the subtask to a delegate QA Reviewer for the OOO window. Default delegate: another AM, or the Workspace Owner if no second AM exists. The delegation rule: the delegate must hold the QA Reviewer competency for that domain (a WordPress QA review needs a WP-literate delegate; a brand-voice QA review needs a brand-literate delegate). If no qualified delegate exists, leave the subtask in TO DO with a comment naming the return date — better to delay the close than to rubber-stamp.

ClickUp Admin (workspace-wide). The Admin role is structural, not per-ticket. For OOO over one week:

  1. Designate a backup Admin by promoting another Member to Admin (Workspace Settings → Members). See ClickUp Permissions Matrix §5 for the lifecycle steps.
  2. Document the OOO window on the recurring workspace-admin tracking surface (or in the team Slack channel).
  3. Demote on return to keep the active-Admin count small (≤3 per Permissions Matrix §2).

For Workspace Owner OOO: the second Owner (if two exist) covers. If only one Owner exists, an Admin acts in a limited capacity — Admins can do everything except billing changes and ownership transfer. Don't add a second Owner just for OOO coverage; that's a succession-level change.

Anti-patterns:

  • Leaving a ticket assigned to an OOO person "they'll pick it up when they're back." This blocks the next step for the duration.
  • Reassigning during OOO without the context-handoff comment (current state + next step). The delegate spends 15 minutes reverse-engineering what was already in your head.
  • Adding the delegate as a Watcher and saying "they have visibility now." Watchers aren't owners; ownership doesn't transfer through visibility.

1. Meta-principle: rules first, automation second​

Every ClickUp Automation must enforce a rule that is already documented in canon — the documented procedures: this playbook, a SOP, a standard, or a working rule. New rules land in writing before any automation is built or changed. Don't infer rules from observed automation behavior — automation is downstream of writing, never the other way around.

Why this matters. Automation that encodes undocumented rules creates shadow policy: no one can read what the system is actually enforcing. The rule is the contract; the automation is a derivative of the contract.

In practice.

  • See a behavior you don't understand and someone says "the automation does that" → push back. Ask which doc the rule lives in. If it doesn't live anywhere, the rule needs to be written before the automation continues to run.
  • Want to propose a new automation? Write the rule first in the relevant canon doc (often this playbook). Open a separate change to add the automation referencing that rule.
  • This applies symmetrically to ClickUp Automations, GHL workflows, and symphony-flow triggers.

See working rule rules-first-automation for the durable form of this rule.

2. Status invariants​

Canonical status definitions live in SC Task Data Standard §1. Don't re-state them here; do internalize the invariants below — these are the rules that catch the failure modes we've actually seen.

The five statuses​

TODO → IN PROGRESS → COMPLETE, with BLOCKED and CANCELLED as off-ramps.

Invariant 1 — time logged means the work has started​

A task with non-zero Time Tracked cannot remain in TODO. Logging the first time entry (live timer or manual entry) implies the work has started; the status must advance to IN PROGRESS in the same motion. If the work is already done, go straight to COMPLETE with the Evidence bullet satisfied.

Worked example. A task in Client Delivery > acme > active-work shows status TO DO but has 3h 45m of tracked time on it. Wrong state. The person who logged the time owns the fix: bump to IN PROGRESS (or COMPLETE if the work is done) before the next status sweep.

Why it matters. Status drives every saved view, dashboard, and reporting cut. If status lies, the views lie. The weekly timesheet approval pass treats any TO DO with time on it as an anomaly — fix in flight, not at month-end.

Invariant 2 — BLOCKED needs a blocked-on:* tag​

A task in BLOCKED status must carry one of the blocked-on:client, blocked-on:vendor, or blocked-on:hosting-support tags (see SC ClickUp Tag Taxonomy §External-dependency state). The tag names what we're waiting on — status alone doesn't.

Invariant 3 — QA-Review subtask gates parent completion​

A parent task that carries a qa-review-<parent-slug> subtask (or the legacy am-review-<parent-slug> form) cannot move to COMPLETE until the subtask itself is closed. See Markdown Task File to ClickUp SOP §3c. Role: QA Reviewer (see §Roles).

Per-client folder architecture (where tasks live)​

Each active client gets one folder under Client Delivery. Each folder has two lists:

  • active-work — every task in every status; the Closed-group hide on complete keeps finished work out of default views.
  • reference — standing reference items only, never work in flight.

Cold storage uses ClickUp's native task archive on a monthly UI sweep — see Workspace Setup SOP §5.6.

3. Client identification — tag vs custom field​

We use both a client/<slug> tag and a Client custom field. They have orthogonal jobs:

Where the task livesUse thisWhy
Client Delivery > <slug> > active-work (per-client list)Client custom field is canonical; client/<slug> tag is optionalField gives dropdown integrity and powers the "By Client" view
Client Delivery shared list (e.g., client-delivery-operations)Both: Client field and client/<slug> tagList doesn't encode the client; tag carries it for cross-space reports
SC Operations space (touching a client)client/<slug> tag (field doesn't exist here)Tag is the only client identifier available in SC Operations

Drift rule: when both are populated on a task, the tag slug must equal the Client field value. No drift.

Authoritative sources:

4. Acceptance criteria and the Evidence bullet​

Every closeable ticket has acceptance criteria. The last AC bullet names the concrete artifact attached at close — screenshot, deploy URL, log link, PR URL.

Evidence: screenshot of the updated header in the staging env attached to this ticket

Blocking for client-facing tickets; advisory for internal work. Without the Evidence bullet, validation reduces to "the contractor said so." See pre-work standard validation-evidence-rule.

5. Time tracking​

Applies to anyone billing hours through ClickUp (currently Project Partner vendor teams; SC Internal Team Members stay on Google Sheets per Timesheet and Invoice Submission SOP).

  • Native ClickUp timer or manual entry with explicit start/end. Either is fine.
  • One task per entry. If you switched tasks mid-session, that's two entries.
  • Description required on every entry. Built header CTA + applied design-system class is good; WP work is not.
  • 0.1-hour granularity (6-minute increments). Decimal hours, not clock minutes.
  • End-of-day batch logging is fine; end-of-week is not — descriptions get vague and entries become unverifiable.
  • First time entry on a TO DO task → bump to IN PROGRESS in the same motion (loop back to §2 Invariant 1).

Full submission flow and AM approval mechanics: ClickUp Timesheet Approval SOP. Per-contractor companions live at 10-people-operations/contractors/<contractor>/<contractor>-timesheet-quickstart.md.

6. Tag hygiene​

The space-level tag list is a closed taxonomy. Only tags listed in the SC ClickUp Tag Taxonomy are valid on tickets. New tags require a PR amending the taxonomy before they appear in ClickUp.

Quick reference for the families:

FamilyUse
client/<slug>Client identity (see §3)
domain:*Subject-matter area (wordpress, seo, dns, billing, ghl, design)
qa:work-failed-review / :work-missed / :work-partial / qa:ticket-not-readyDefect telemetry — applied per review standard
blocked-on:*External dependency causing a BLOCKED status (see §2 Invariant 2)

What is not a tag: status, priority, assignee, due date, time estimate, one-off context. Use the structured field or the description.

7. Subtasks and the QA-Review pattern​

For any ticket whose primary output is client-facing (published copy, brand assets, anything the client will see), the Creator adds a qa-review-<parent-slug> subtask at ticket creation. The subtask follows the standard flow (to do → in progress → complete). The QA Reviewer (default: the Account Manager — see §Roles) is the assignee.

The parent ticket cannot move to complete until the QA-Review subtask is closed (§2 Invariant 3). This is how SC catches defects pre-client without a workflow-wide review status.

Internal/non-client work flows TODO → IN PROGRESS → COMPLETE on contractor self-attestation backed by the Evidence bullet. Post-hoc sampling by the AM catches escapes across both paths.

Legacy compatibility: before 2026-06-14 this pattern was named am-review-<parent-slug>. Existing am-review-* subtasks remain valid (do not rename them retroactively). New tickets use qa-review-*. Views must match both — see Views Guide §6.5.

Full mechanics: Markdown Task File to ClickUp SOP §3c.

8. One assignee per ticket​

Tickets have one assignee. Multiple assignees split accountability and break "who owns this" — they're disabled at the workspace level. If two people need to act, that's two subtasks or two separate tickets, not one ticket with two assignees.

See working rule single-assignee-per-ticket.

9. Views and saved filters​

Use the saved views; don't rebuild ad-hoc filters in the UI every week.

ViewRecipeUse
workspace-all-openViews Guide §6.7Every open ticket across every client — the daily situational-awareness view
workspace-completed-30dViews Guide §6.8What closed in the last 30 days; also the surface for the monthly native-archive sweep
<client>-completed-this-monthViews Guide §6.9Per-client completed work for the current calendar month
timesheet-<contractor-slug>Views Guide §6.1Per-contractor time-entry export
active-work-<client-slug>Views Guide §6.2Per-client open work, sorted by priority/due-date
blocked-all-clientsViews Guide §6.3Cross-client blocked-ticket triage
quality-flagged-15dViews Guide §6.4Quality-tagged tickets in the rolling 15-day window
qa-review-queueViews Guide §6.5Personal QA-Review subtask queue (matches both qa-review-* and legacy am-review-*)
capacity-next-7dViews Guide §6.6Workload check for the upcoming week

Naming convention and sharing rules are in Views Guide §5 and §4. If you find yourself rebuilding the same filter twice, save it under the convention. Avoid list-name filters — see Views Guide §2 anti-pattern note.

10. Comments vs description​

WhereWhat goes there
DescriptionCurrent-state snapshot of the ticket: scope, acceptance criteria, links to canon. Updated as the work evolves — not just at creation.
CommentsAudit trail of discussion, decisions, blockers, hand-offs, client quotes. Append-only by nature.

Status changes don't replace description updates. If the AC changed mid-flight, edit the description and drop a comment naming what changed and why. The description should always read as if you're starting today.

11. Ticket liveness — cadence, dwell, and going quiet​

Sections 1–11 cover writing a ticket well and closing it properly. This section covers the part in between: what owning a ticket obligates you to do while the work is in flight.

The point is communication, not surveillance. The comment exists so the next person can act without asking you first. A ticket that has gone quiet is indistinguishable — from the outside — from a ticket nobody is doing, and that ambiguity is what costs weeks.

The cadence​

SituationWhat the Assignee owes
Ticket is IN PROGRESSA progress comment at least every 3 business days. "Still on this, blocked on nothing, expect Thursday" is a valid comment. Silence is not.
Ticket is assigned and TO DOStart it, or comment saying why not, within 5 business days of assignment.
Ticket is BLOCKEDA blocked-on:* tag (§2 Invariant 2) plus a chase comment every 5 business days naming who was chased and when.

A comment costs thirty seconds. The alternative — reconstructing state from a 60-day-old ticket — costs an hour and usually a call.

Dwell ceiling​

A ticket sitting in IN PROGRESS for more than 10 business days must be split, moved to BLOCKED with a tag, or re-scoped.

Long dwell is a scoping defect, not a diligence defect. A ticket that takes three weeks of elapsed time was almost always mis-sized at creation, or is waiting on something that should have been made explicit. Treat the ceiling as a prompt to fix the ticket, not as a performance signal about the person holding it.

There is deliberately no WIP limit. Concurrency has not been the failure mode here; dwell and silence have. A cap that never binds is a rule that teaches people rules don't matter.

When you are stuck, move to BLOCKED​

This is the rule that matters most, because it is the one most often skipped.

A blocker recorded only in a comment is invisible. No saved view queries comment text, so a ticket that reads IN PROGRESS while waiting on a credential, an approval, or an answer will keep reading as active work indefinitely — and nobody chasing the blocker will know it exists.

If you are waiting on someone else: change the status. Comment as well, but the status change is what makes it visible. BLOCKED is not an admission of failure; leaving a stuck ticket in IN PROGRESS is.

The same applies in the other direction. A question asked in a comment is a hand-off, and the person it is addressed to owes an answer on the same 3-business-day rhythm. An unanswered question is a blocker with no owner.

Closing is part of doing​

Work is not finished when the work is finished — it is finished when the ticket reflects it. If you complete the task, attach the Evidence bullet's artifact and move the status in the same motion. A ticket left open after the work is done misroutes everyone reading the board.

What happens when cadence slips​

Escalation is graduated and impersonal:

  1. First miss — the ticket surfaces in the weekly stale sweep (/audit-stale-tickets). No action beyond a nudge comment.
  2. Second consecutive miss — raised on the next scheduled call, as a scoping question first: is this ticket still right?
  3. Third — reassigned, or parked (no assignee, no due date — the parked-backlog convention). Parking is a legitimate outcome, not a punishment; a ticket nobody is going to do this quarter should look like one.

Detection is automated so nobody has to police it by memory: signals S7 (assignee silence, 7 days), S8 (orphaned assignment) and S9 (declared done, never closed) live in the SC ClickUp Ticket Audit Playbook, which measures last assignee touch rather than date_updated — the latter bumps whenever anyone edits a ticket, including the Creator and the audit itself.

12. The hygiene calendar​

Recurring rituals that keep the workspace honest. Each is defined in full by its owning doc — this table exists so you can see the whole rhythm in one place, not to restate procedure.

CadenceRitualRoleDefined in
OngoingQA-Review queue: respond within 2 business daysQA ReviewerQA Reviewer SOP §4.3
Weekly (Mon)Stale sweep — assignee silence, orphaned assignments, done-not-closedWorkspace Owner/audit-stale-tickets + Audit Playbook
Weekly (Mon)Post-hoc quality sampling of 10–15% of last week's closuresAccount ManagerReview Standard
WeeklyTimesheet approval batchQA ReviewerTimesheet Approval SOP §5
MonthlyNative-archive cold-storage sweep of completed ticketsClickUp AdminWorkspace Setup SOP §5.6
MonthlyEngagement review per contractorWorkspace OwnerPartner Engagement Framework §3
QuarterlyTag audit — live tags vs. taxonomyAccount ManagerSC ClickUp Tag Taxonomy
QuarterlySaved-view maintenanceClickUp AdminViews Guide §7
Per client, ad hocDeep ticket audit for one clientAccount Manager/audit-client-tickets <slug>

The weekly stale sweep is the one that backstops §11. Everything else assumes tickets are broadly honest; the sweep is what makes that assumption safe.

13. Where to learn more​

TopicCanon
Status modelSC Task Data Standard §1
Tag taxonomySC ClickUp Tag Taxonomy
Ticket authoringMarkdown Task File to ClickUp SOP
Time tracking + approvalClickUp Timesheet Approval SOP
Workspace/space setupSC Task Management Workspace Setup SOP
SC Operations spaceClickUp SC Operations Space Setup SOP
Pre-work quality (AC, evidence, tags)Pre-Work Standard
Review quality (quality tags, sampling)Review Standard
Views and filtersClickUp Views and Filters Guide
Which views to use by roleClickUp Views — Quick Reference by Role
Ticket templateClickUp Thin Ticket Template
Staleness signals + audit dispositionsSC ClickUp Ticket Audit Playbook
Durable behavioral rules.claude/working-rules.md

Revision History​

DateVersionChangeAuthor
2026-06-081.0Initial consolidated playbook. Captures the meta-principle (rules first, automation second), status invariants (including the new Time Tracked > 0 ⇒ ≥ IN PROGRESS rule from sc-task-data-standard §1), client identification, AC + evidence, time tracking, tag hygiene, AM-review subtasks, single assignee, views, and comments-vs-description. Designed as the single doorway for every internal-team user.Symphony Core Systems Team
2026-06-081.1§2: added "Per-client folder architecture" sub-section reflecting the 2-list simplification (active-work + reference, no completed-work or archive lists). §9 Views: rewrote as a table naming the three new workspace/folder views (workspace-all-open, workspace-completed-30d, <client>-completed-this-month) alongside the existing recipes, with a pointer to the list-name anti-pattern note. Driven by Workspace Setup SOP v1.2.Symphony Core Systems Team
2026-06-121.2Sprint 1 of Rohit's Remarkable-tablet feedback: §Scope wording → "Team Members"; §1 canon defined inline on first use; §2 Per-client folder architecture sub-section: removed historical commentary (the "retired 2026-06-08 because…" sentence), describes current state only; §5 Time tracking: "individual retained contractors" → "SC Internal Team Members"; §9 Views table: quality-flagged-this-quarter → quality-flagged-15d. Driven by 2026-06-12 sprint backlog S1-2 through S1-5.Symphony Core Systems Team
2026-06-141.3Sprint 2: new §Roles section (4-role model — Creator / Assignee / QA Reviewer / ClickUp Admin, Human-or-AI annotations). §2 Invariant 3 + §7 renamed AM-review subtask pattern → QA-Review subtask pattern; subtask name becomes qa-review-<parent-slug> for new tickets, existing am-review-* stay legacy. §9 view name am-review-queue → qa-review-queue. workflow_owner: CEO → workflow_owner: Workspace Owner (per reference-roles-strictly). Driven by 2026-06-12 backlog S2-1 + S2-4.Symphony Core Systems Team
2026-06-141.4§Roles: added "Role handoff — vacation, sick, delegation" sub-section (per-ticket OOO pattern for Creator/Assignee/QA Reviewer + workspace-wide pattern for ClickUp Admin/Workspace Owner). Cross-links to new ClickUp Permissions Matrix.Symphony Core Systems Team
2026-08-111.5New §11 Ticket liveness — the first written cadence contract for work in flight: comment every 3 business days in IN PROGRESS, pick up or explain within 5 of assignment, chase a blocker every 5, and a 10-business-day dwell ceiling framed as a scoping prompt. Adds the move-to-BLOCKED rule (a blocker recorded only in a comment is invisible to every saved view), the reciprocal answer obligation on questions asked in comments, closing-is-part-of-doing, and a graduated escalation ending in parking. No WIP limit — deliberately omitted; concurrency was not the observed failure mode and a non-binding cap teaches that rules don't bind. New §12 hygiene calendar consolidating seven existing rituals scattered across as many docs, plus the new weekly stale sweep. Former §11 renumbered to §13. Driven by the 2026-08-11 workspace ticket audit (100 assigned open tickets, ~59 unmoved for 30+ days, 6 assignee comments across one contractor's 49).Symphony Core Systems Team