Simulation: Manage a Flipkart Campaign Through Its Learning Phase
Objective: Make weekly budget, creative-volume, and refresh-timing decisions across a 4-week simulated Flipkart Advantage+-style campaign launch, and see the realistic consequence of each choice on learning-phase completion and CPA.
You're running paid social for Flipkart's private-label electronics line, launching an AI-optimized campaign for a festive sale window.
At each weekly checkpoint, decide how to respond to the dashboard. Each choice has a realistic, lesson-grounded consequence that carries into the next week.
Before you start
What you'll need
Free path (everything below is enough to finish)
Free, and sufficient for a single campaign's weekly checkpoint tracking
Paid upgrades (optional, faster/deeper)
The free path (manual creative production tracked in a sheet) works at low variant counts, but hitting the lesson's 16-variant matrix and a 3-4 week refresh cadence in parallel is where a paid generation tool starts saving real production time.
Purpose-built for generating full visual creative plus copy at the volume the learning phase and refresh cycle require
The process
Simulation
Week 1 — Launch
Day 7Day 7 of launch. The dashboard shows 31 conversions logged against the 50-conversion learning-phase target, on a $220 electronics accessory.
Conversions: 31/50 | Daily budget: $35 | CPA so far: $19 | Learning phase: not exited
What do you do at the end of week 1?
Final deliverable
A 4-week decision log showing which option was chosen at each checkpoint and the resulting CPA/ROAS trajectory that followed.
See a reference example
Airbnb 'Host Spotlight' campaign — decision log (excerpt) Week 1: doubled budget to exit learning phase on schedule (optimal) Week 2: built the full 16-variant matrix (optimal) Week 3: waited a week to confirm fatigue before refreshing (acceptable), lost ~6% CTR in the interim Week 4: routed the new batch through human review before scaling (optimal), caught one off-tone variant before launch
Success criteria
You're done when you can:
- Chooses a defensible option at all 4 checkpoints with reasoning tied to the specific lesson rule
- Can explain the realistic cost of the costly option at each stage, not just identify the optimal one
- Final decision log shows a coherent campaign trajectory, not contradictory choices week to week