The theory: boosting organic social posts (engagement-optimized, not conversion-optimized) trains Meta's pixel and audiences on low-intent viewers — quietly raising CAC, CPMs, and dragging volume. Did it? Every campaign here is classified by Meta's actual objective field, not its name.
Bars = share of Meta spend going to genuinely non-conversion campaigns (awareness / engagement / reach / traffic / video-views, per Meta's objective field). Lines = blended D2C CAC (Daily Spend Tracker: spend ÷ Shopify new customers) and Meta CPM. If the hypothesis were right, the bars and the CAC/CPM lines should move together. They don't — the bars vanish while the problems continue.
Genuine engagement spend ran only in the second half of March (peaking ~$11k/week, 4.8% of Meta spend) and was switched off entirely from early April. The CPM spike to ~$48 (late April / early May) and the worst CAC weeks (mid-May → June, $98–$113) both happened with zero engagement spend running.
First pass, I classified campaigns by name and the number looked like 2.7% of spend. That was wrong — it swept in MUD_CPP_D0_TopAds ($119k), which Meta's objective field shows is a conversion (sales) campaign, not a brand boost. Cross-checking every campaign against Meta's real objective field, the truth is smaller and clearer:
Across the 12 weeks, here's how engagement's share of spend correlates with the metrics it was supposed to be hurting. (+1 = move together, 0 = unrelated, −1 = move opposite.)
Steelman — does it hit with a delay? Pixel/audience pollution could lag. Engagement in week t vs CAC two weeks later is +0.17; vs CPM two weeks later is −0.34 — small, inconsistent, and with engagement off after early April there's simply nothing left to lag. No delayed-damage pattern exists in the data.
Engagement spend did ramp Jan → Feb → March (peaking ~$46k for the month, 4.8% of Meta spend in its top week), and March was a worrying month. This is the one genuine piece of support for the hunch — and why it was worth checking.
The core fear is engagement → higher CPMs. But CPM was at its lowest ($24) during the March engagement peak, and only spiked ($44–48) after engagement was switched off. That's the seasonal summer ramp, not pollution. Correlation −0.74.
The "we scaled engagement in early May" memory doesn't survive the objective field — that spend was MUD_CPP_D0_TopAds, a conversion campaign. Real engagement spend in May was $0. The early-May CAC slide had nothing to do with brand boosting.
At its peak, real engagement was ~1% of total paid spend, and zero for half the window. Even if every engagement dollar were pure waste, it mathematically cannot move blended CAC the way the lines moved.
Same story on volume. Weekly new customers (Shopify) don't rise with engagement spend or fall when it's cut — the relationship is flat (+0.05). Volume softened into June while engagement had been off for two months.
The CAC pressure lines up with things we've already documented — none of which is engagement spend, which was off entirely during the worst stretch:
Using Meta's objective field removes the classification guesswork, but a couple of limits remain:
Given engagement is ~1% of spend and already off, the practical answer is "this isn't your CAC problem." If you still want to close it out:
| Week of | Engagement $ | Eng % of Meta | Blended CAC | Meta CPM | New customers |
|---|
CAC & new customers: MUDWTR Daily Spend Tracker (D2C spend ÷ Shopify new customers), Pacific dates. Engagement spend: Northbeam campaign spend, filtered to campaigns whose Meta objective is non-conversion (awareness / engagement / reach / traffic / video-views). CPM: Meta Ads, weekly. Monthly verified engagement: $7k (Jan) → $23k (Feb) → $46k (Mar) → $1.6k (Apr) → $0 (May) → $0 (Jun). Six-month total $77,790 ≈ 1% of paid spend.