The social campaign process
One process from a marketing department's catalog, written out in full. This page has the document itself, the records one run left behind, and a simulation that plays that run from the trigger to the close.
the document
The process document
Every blank in the format is filled in. Version 6 rewrote the process around
the thirteen activities the manager actually does, and the amendment history says
so. Sign-offs are marked never: brand and legal stay with people no
matter how good the agents get.
PROCESS: social campaign id: mkt/social-campaign v6
owner: Head of Social effective: 2026-08-20
trigger: a campaign brief is approved in mkt/campaign-planning
concurrency: runs may overlap - up to six live at once, one per campaign
goal: a campaign launched on every channel the brief names, run for its
window, and reported against what the brief asked for
phases:
take-in-brief - human: read the brief, pull out what the campaign
must accomplish
owner: Manager after: trigger
set-parameters - human: pick platforms, audience and dates, and put
a number on it. Leadership confirms the budget
owner: Manager after: take-in-brief
brief-creatives - convenes briefing: the team hears what is needed
and by when, together
owner: Manager after: set-parameters
settle-direction - convenes bake-off: two or three concepts, one
chosen, with the reasons said out loud
owner: Manager after: brief-creatives
produce-posts - runs build-by-talent: copy, images and video,
each sized and worded for its platform
owner: Manager after: settle-direction
get-sign-offs - convenes approval: brand and legal sign, as named
signers, against a version of the posts
owner: Legal reviewer after: produce-posts
automation: never
practical-side - human: tracking links, the landing page check, the
ad campaigns, the posting schedule
owner: Manager after: set-parameters
launch - human: the manager says go. The scheduler and the
ad platforms take it from there
owner: Manager
after: get-sign-offs + practical-side
watch - human: track results against the brief
owner: Manager after: launch
engage - human: answer comments and messages
owner: Community manager after: launch
optimize - human: move ad money, swap stale creative
owner: Media buyer after: launch
wrap-up - runs collect-and-report: the numbers from every
channel, reported against the brief
owner: Manager
after: watch + engage + optimize
record-learnings - convenes debrief: what to repeat, what to avoid
owner: Manager after: wrap-up
run-scoped:
spend - runs allocate-and-reconcile owner: Manager
from: set-parameters until: run close
status - runs collect-and-report owner: Manager
every: 7 days
from: launch until: run close
policy:
no claim ships without a citation in the claims register
sign-off is a human gate and is never delegated to an agent
a post changed after sign-off goes back through sign-offs. There is
no minor-change path
amendments:
v2 - claims checking moved out of legal review, so legal reads the
exceptions rather than every asset (debrief, run 12)
v3 - claims checking raised from supervised to autonomous after
twenty clean runs (debrief, run 31)
v4 - the readout moved from 7 days to 14. A week was too early to see
anything on the slower channels (debrief, run 38)
v5 - the weekly status, the budget watch, and community response
written down as run-scoped (debrief, run 47)
v6 - rewritten around the thirteen activities the manager actually
does. Setting parameters, the practical side, the launch
decision, the live window and the learnings had been compressed
into six phases or left out entirely (activity walk, 2026-08-20)
The phase graph has two branches and a window. After the parameters are set,
the creative chain and the practical side run at the same time, and launch waits
for both. After launch, watch, engage and optimize run at once for as long as the
campaign is live, and wrap-up waits for all three. Order comes from the
after: lines rather than from the numbering.
the records
The records a run produces
Each activity leaves whatever record its own kind of work produces. The
briefing, the bake-off, the approval and the debrief leave their CONVENED and
DONE records. Produce-posts and wrap-up leave a plan and a JOB DONE each, because
they run work patterns. A human: activity leaves one line saying
what happened and who did it. The run opens and closes around all of them:
RUN: mkt/social-campaign v6 · run 52
started: 2026-08-20 by Manager
trigger: campaign brief CB-118, approved in mkt/campaign-planning run 9
RUN DONE: mkt/social-campaign v6 · run 52
outcome: completed
result: the campaign ran its window on 4 channels; posts reached v3;
reported against the brief
open: one flagged reply is waiting on an answer from support
A run keeps the version it ran under. Run 47, two versions ago, carries v4 in its records, and an auditor reads those records against v4 rather than against this page.
Between them, those records answer what an auditor asks for. The process document says what the process is, and the run records show that it ran. Because each record was written as the work happened and names who acted and when, nobody has to reconstruct any of it afterwards.
the simulation
A simulation of one run
One run is played from start to finish on its own page, with the sign-off changes and a mid-flight creative swap included.