Performance marketing — overview

Ten use cases covering the weekly NRoAS question, ad-action decisions, channel scenarios, creator trends, efficiency targeting, and MBR/QBR assembly.

This is the perf-marketing team's working catalogue: the questions actually asked in the weekly marketing RCA and the Friday finance RCA, mapped to platform recipes.

The list comes from real discovery work with the perf-marketing leadership and analytics teams — see docs/research/transcripts/perf-marketing-2026-05-15.md for the source. Each use case below is grounded in a verbatim question from that conversation.

The 10 use cases

Daily / weekly

NRoAS drop RCA

The #1 weekly question. "Why is my acquisition NRoAS not holding up?"

Ad action decisions

Which ads to kill, pause, scale, or reduce spend on — with reasoning.

Audience overlap diagnosis

Deep dive: are two ad sets eating each other's new-customer reach?

Product-lag RCA

"Shilajit is -30% behind target. Why?" The product-specific weekly RCA.

Target vs landing

Heuristic month-end forecast accounting for sale days and weekday mix.

Monthly

Creator pattern trends

What creator pattern is working this month, and has it changed?

Creator size mix

Mega vs Micro vs Nano — which size is delivering for which product?

MBR / QBR assembly

Compose the monthly business review or quarterly review automatically.

Strategic / scenario

Channel budget scenarios

"What if I move 10% from Meta to Google?" Forecast revenue and CM3 impact.

Marketing efficiency

Reaching the 40% marketing-percent target by closing executional + inferential gaps.

How the team works today

Context for users new to the perf-marketing function:

Team: Five people. ~₹2.5–3 Cr per person per month in spend.
Channel mix: Heavy Meta dependence, Google as secondary. Goal: become 30% less Meta-dependent over time.
Efficiency target: Moving marketing % from 55% → 40% while growing 40% YoY.
Cadences: Daily monitoring; weekly RCA with marketing; Friday RCA with finance; monthly MBR; quarterly QBR.
Data home: BigQuery — spend, revenue, new-vs-repeat segregation. Ad-set-level targeting / Meta interest configs are not in the BigQuery export today (a known gap; see Sources).

A note on logic trees

The team uses logic trees (RCA decision trees) to navigate from "metric down" to "root cause." Each use case page below maps to a logic tree. These start at ~70% coverage and self-extend as edge cases are added — the platform learns the team's tree over time so the next "why?" question routes itself.