Business Context

A performance marketing team at Meta is planning next quarter's acquisition budget for a new advertiser using Meta Ads across Facebook Feed, Instagram Reels, Stories, and Audience Network. Last quarter, spend increased 18%, but finance is questioning whether budget was over-allocated to high-volume placements that drove cheap clicks but weaker downstream conversion quality.

Metric Scenario

You are given channel-level performance for the last 90 days. Facebook Feed spent $2.4M, delivered 120M impressions, 3.6M clicks, 144k purchases, and $7.2M in attributed 30-day revenue. Instagram Reels spent $1.8M, delivered 210M impressions, 4.2M clicks, 84k purchases, and $4.0M revenue. Stories spent $1.1M, delivered 95M impressions, 1.9M clicks, 76k purchases, and $3.8M revenue. Audience Network spent $0.7M, delivered 160M impressions, 2.8M clicks, 28k purchases, and $1.0M revenue.

The VP of Marketing wants a segmentation strategy for reallocating next quarter's $7M budget. They do not want a simple last-click ROAS ranking. They want to know how you would segment channels and audiences to balance scale, efficiency, and conversion quality, especially for new vs returning customers and high-LTV cohorts.

Requirements

1. Define the primary metric or metric set you would use to allocate budget across Meta channels.
2. Propose a segmentation framework that goes beyond channel-level averages.
3. Decompose performance into funnel and value components to identify where each channel is strong or weak.
4. Explain how you would decide whether to shift budget from Instagram Reels or Audience Network into Facebook Feed or Stories.
5. Identify guardrails so budget optimization does not improve CAC while hurting revenue quality or retention.

Data Available

• meta_ads_delivery: campaign_id, adset_id, placement, impressions, clicks, spend, reach, frequency, date
• web_conversion_events: user_id, campaign_id, placement, add_to_cart, checkout, purchase, purchase_value, event_time
• crm_customers: user_id, acquisition_date, new_vs_returning, 90d_revenue, 180d_revenue, refund_flag
• geo_device_breakdown: country, device_os, placement, spend, clicks, purchases, revenue