The Judgment Line: Auditing a Swiggy Marketing Team Roster by Task
Objective: Given a 6-person marketing team's weekly task list, apply the lesson's 'draw the line at judgment, not task type' framework to sort each task into 'stays human' or 'AI-owned,' then name the 2026-era role that should own the AI-owned bucket.
You're the marketing ops analyst at Swiggy, auditing the growth marketing team's task list ahead of a restructuring proposal for next quarter.
Sort every task on the roster by judgment content, not by who currently does it or which channel team they sit in, then assign an owner to the execution bucket.
Before you start
What you'll need
Free path (everything below is enough to finish)
No account friction, filters and a formula column are all this audit needs
Paid upgrades (optional, faster/deeper)
Gives the named owner a shared queue instead of an ad-hoc spreadsheet handoff
The process
1 step
Step 01 of 01
The lesson's division-of-labor map draws the line at judgment: strategy, relationships, and brand-risk calls stay human; drafting, data pulls, and first-pass execution move to AI. Every role has both kinds of work inside it, so sorting by department instead of by task is the most common mistake.
The roster lists 14 weekly tasks across 6 people, from 'write client positioning memo' to 'pull last week's campaign CTR into the report.' Which stay human, which move to an AI workflow?
Procedure
- Import the export and freeze the header row.
- Add a 'bucket' column and mark each task 'human' or 'AI-owned' using the judgment test, not the current owner's job title.
- Flag any task that mixes both, e.g. 'write headline variants and pick the winner,' and split it into two rows.
- Total the hours in the AI-owned bucket to see how much capacity would free up.
HUMAN (5 tasks, 11 hrs/week) - Client positioning memo for Q3 pitch - Escalation call with a churn-risk restaurant partner - Sign-off on a campaign that touches a sensitive delivery-fee message AI-OWNED (9 tasks, 19 hrs/week) - Pull last week's campaign CTR into the weekly report - Draft 10 push-notification headline variants for A/B testing - First-pass competitor promo scan across 3 rival apps
Healthy
AI-owned bucket totals more hours than the human bucket, and every AI-owned task has a proposed owner for the workflow.
Unhealthy
Tasks are bucketed by which channel team currently does them, so 'content team' tasks all land in one bucket regardless of whether they require judgment.
What this means
A task-by-task split, not a role-by-role split, is what actually reveals where hours can move.
So what do I do about it?
| Symptom | Action | Effort |
|---|---|---|
| The AI-owned bucket has no named owner for the workflow or quality gate | Assign the bucket to a single role (per the lesson, usually an AI Marketing Operations Lead) before proposing the split to leadership | 30 min |
| A task like 'write headline variants and pick the winner' sits in one bucket | Split it: variant generation is AI-owned, picking the winner is a human judgment call | 5 min |
Final deliverable
A task-level split of the 14-task roster into 'stays human' and 'AI-owned' buckets, with total hours per bucket and a named role assigned to own the AI-owned workflow.
See a reference example
Squarespace, SMB team task audit (excerpt) HUMAN (4 tasks) - Renewal-risk merchant call, escalation tier - Final sign-off on template names implying a discount that doesn't exist AI-OWNED (7 tasks) - Draft 8 subject-line variants for the template-launch email - Pull template adoption rate into the Monday report Owner assigned: AI Marketing Operations Lead, prompt library + weekly QA gate.
Success criteria
You're done when you can:
- Every task on the roster is assigned to exactly one bucket, with mixed tasks split rather than force-fit
- The AI-owned bucket has a single named owner role, not left unassigned