All session materials
IIT Kanpur ·
Session 4 — Intercept → Interrupt · Prompts & Activities
Decide the channel → match the format → feed the event
Decide the channel → match the format → feed the event
How to use this sheet
- Every prompt follows CRISP. Replace everything in
[brackets]with your real numbers. - Search captures demand; YouTube and Meta create it. Both prompts force the channel-affordability call first — don't run a channel your runway can't wait out or your pixel can't feed.
- The output is a draft to interrogate. The AI picks a format and an event in seconds; your skill is judging it on the right metric, not the cheapest one.
The CRISP framework — how we prompt, every time
| C | Context | Your offer, budget/mo, AOV, the funnel goal. |
| R | Role | The expert hat the AI wears. |
| I | Intent | The task, sequenced. One job at a time. |
| S | Specifics | The metric to judge on, the minimum budget, the misleading event. |
| P | Parameters | The guardrails and the self-check. |
A CRISP prompt is a brief, not a wish. The brief is where the quality is decided.
Part 1
Prompt 1 — YouTube media planner (does it earn a seat?)
Act as a YouTube media planner for an Indian D2C/EdTech brand at ₹[budget]/mo. Goal [awareness / consideration / action]. Recommend format, targeting, video length, the ONE metric to judge it on at this funnel stage, and the minimum monthly budget below which YouTube isn't worth running.
Watch-outs
- Format follows funnel: bumpers (6s) = reach/recall · skippable in-stream = consideration · in-feed = intent-adjacent. Wrong format + wrong metric = wasted spend judged unfairly.
- The affordability ladder (run the math, don't take a verdict): under ₹1L/mo → Meta conversion + Search brand-defence only · ₹1–3L/mo → add retargeting depth · ₹3L+/mo → YouTube top-funnel earns a seat.
- Don't judge a reach format on last-click conversions. YouTube awareness won't show in last-click GA4 — judge on reach/frequency/view-rate plus the remarketing-driven downstream conversions you wired.
- SEO is a decision, not a module: this program teaches whether/when to invest, not how to do SEO. It's a 6–12 month CAC bet — the test is whether your runway can wait. Your Week-2 intent map becomes the content-priority list.
Coach's note — Search waits for the question. YouTube plants it. Don't grade the seed by next week's harvest — but do wire the harvest so you can find it.
Part 2
Prompt 2 — Meta media buyer (right objective, right event)
Act as a Meta media buyer. Goal [sales / leads], offer [X], AOV ₹[Y]. Tell me the campaign objective AND the exact optimisation event to choose (and which cheaper event will mislead me and why). What does success look like in week 1, and what's the vanity-action trap for my goal?
Watch-outs
- The objective is the steering wheel; the optimisation event is the accelerator. Meta floods you with whatever event you pick — choose "Add to Cart" and you buy cart-adders, not buyers.
- The learning-phase trap (small budgets): an adset needs ~50 optimisation events/week to exit learning. If you fire ~100–160 Purchases/month total, optimising on Purchase starves the algorithm. Fix: (1) consolidate to ONE adset, and/or (2) optimise on a higher-volume proxy (Initiate Checkout / Add-to-Cart) until Purchase clears ~50/week, then graduate down.
- A "cheap" cost on the wrong event is the most expensive number in marketing. India indicative: D2C conversion CPMs ₹150–400; lead-gen CPLs ₹40–150 (EdTech) to ₹200–600 (high-ticket).
- Check the pixel first: can it pass the Purchase event reliably enough to optimise on it? If not, fix tracking before you spend. (Sets up the Week-5 Pixel + CAPI work.)
Coach's note — Meta gives you exactly what you optimise for. The skill is wanting the right event — not the cheapest one.
Decide the channel. Match the format to the funnel. Feed the event you optimise for.
A 'cheap' cost on the wrong event is the most expensive number in marketing.