The 8-month arc · May 2025 engagement → CTV Aug 11 2025 → Apr 2026.
The feedback loop was disconnected. The way they measured success was wrong. Acquisition strategies were hitting diminishing returns. Spend climbed +64% MoM yet new-customer CAC blew +55% and one-time CAC blew +142% - Meta CAC held flat at $138 and that was the headline they were tracking. The composed view told the truth: the system needed a strategic pivot.
Operating model reframe delivered. “It was never a Meta problem. It was a mix problem.” Audit → align → bridge. Cash-view (CFO) and accrual-view (CMO) reconciled in one operating model. Vendor measurement became a guide, not a gospel. Triangulation replaced single-source decisioning.
Phase 1 hero · Aug 11 – Oct 5 2025. Halo-forward CAC compressed from a $42.69 pre-CTV baseline to $32.58 over 8 weeks (−24%). Phase 2 (Oct – Dec, 9 weeks of holiday brand-layer cuts) drifted to $41.70. Phase 3 (Dec – Mar, 16 weeks, brand-layer restored at 40% intensity) settled at $40.20. Every week of the 33-week arc held below the pre-CTV baseline. MER trajectory aligns with brand-converted demand landing in highest-retention cohorts.
Halo + marginal CAC. TV makes every other paid dollar cheaper.
Weekly read. Three MER reads in parallel: Standard MER · Halo MER (TV cut) · Halo MER (brand-layer cut). Marginal CAC is the next-dollar read per channel. Together they answer the CFO question: where do I put the next dollar, and what's it worth? Numbers below are this brand's · every brand recomputes from its own paid + CTV reach partnership data at onboarding. Methodology is universal · numbers are per-brand.
Halo CAC · 33-week trajectory.
What landed. Pre-CTV (May 1 – Aug 10, wks 1-2): halo-forward CAC anchored at $14.37. CTV launched Aug 11 (week 3). Over the next 8 weeks the orange actual line compressed — first 8wk averaged -24% to $11.13, peak week Sep 29 hit -25% at $10.78, and the Sep 22-Oct 12 3wk peak window averaged -14% at $12.41. The gray dashed counterfactual is what would have held without CTV — flat at $14.37 throughout. The gap IS the halo. Phase 2 (wks 11-19, holiday TV cuts) drifted toward baseline at $13.50 as Brand Layer Intensity ramped to 27%. Phase 3 (wks 20-35, TV restored higher, 40% brand-layer intensity) held $13.30 — still under baseline, halo durable at peak intensity. The marginal CAC ladder below shows where the next dollar goes inside this trajectory.
Halo Visibility Cut · hold TV (or full brand-layer) out of the denominator
Toggle three views: Standard MER (everything in denominator) · TV Withheld MER (only CTV held out · isolates TV’s halo) · Brand-layer held-out MER (full brand-layer bucket held out · TV + Meta brand-layer + programmatic display + audio · shows TV’s relative weight inside the brand layer). Same data. Same window. Different lens. Pre-CTV baseline = MER 1.22 / CAC $14.37 (case-study · master baseline · May 1-Aug 10 2025) (TV at 0% of total spend).
What does the next $1K buy you?
Marginal CAC is what acquiring the next customer costs on each channel. Not the average of customers you've already won. Different number, different decision.
How to use it. When marginal CAC sits below your target ceiling, you have headroom - scale spend. When it climbs above the ceiling, you're at saturation - optimize creative or audience before adding more budget.
YouTube + TV have headroom - scale them. Meta brand-layer and Meta performance are above the LTV-derived ceiling - optimize creative + audience before adding budget. The "expensive" channel (TV at $215) is what's making YouTube efficient at $195. Pull TV down and YouTube's marginal CAC drifts up.
MER under three TV spend ramps · the long-run decision
TV is an always-on signal. The decision is the long-run spend tier, not whether to flip the switch. Brands that turn TV on and off lose the compounding halo entirely — the ad-stock physics in L4 only work when the signal is sustained. Pre-TV baseline Halo MER is 1.22; sustained Brand Layer at ~25-27% of total spend lands at 1.28× standard during the Phase 1 hero window (+6% vs baseline), with week 8 compression hitting -25% on Halo CAC.
| Scenario | Brand-layer MER | Performance MER | Total MER | Weekly halo revenue |
|---|---|---|---|---|
Ramp down −25% sustained $37.5K/wk long-run (down from $50K) | 0.65x ↓ from 0.75x | 1.18x ↓ from 1.28× · halo wears off | 1.16x ↓ below 1.22 baseline | ~$8K ↓ from $15K · −47% |
Steady at today's tier $50K/wk · Brand Layer at ~25-27% of total spend · Phase 1 hero sustained | 0.75x | 1.28× | 1.28× +6% vs 1.22 pre-TV baseline · peak wk sustained -14% through week 8 on Halo CAC | ~$15K |
Ramp up +25% sustained $62.5K/wk long-run (up from $50K) | 0.85x ↑ diminishing returns | 1.36x ↑ halo expands performance | 1.89x ↑ vs 1.22 pre-TV baseline | ~$22K ↑ +47% |
Consistency drives results. TV doesn't work as on/off - it works as a sustained signal that compounds. Ramping the long-run down 25% costs ~$7K of weekly halo revenue and 0.4x of total MER. Ramping up +25% lifts both, but at compressing returns. Operator decision: dial the sustained tier until total MER stops climbing - not until brand-layer MER stops climbing. Different cliffs.
The killer move isn't TV's halo · it's TV's durability. Hold TV at ~15% of total spend, and Meta + Google CACs drop below their respective pre-CTV baselines ~10-14 days after launch (ad-stock fills) and stay there. The cheap-Meta + cheap-Google switch turns ON 10-14 days after TV launches and stays ON. The only way to lose it: shift TV spend by -3-7pp · the ad-stock decays · the warmed audiences are gone · CACs bounce back above baseline within 2 weeks. The brand that holds TV steady gets every other paid dollar working harder · forever. The brand that flips TV on/off pays the same CAC they always have.
| Channel | Pre-CTV baseline | T+10-14d fill | Sustained (TV at 15%) | If TV cut −5pp |
|---|---|---|---|---|
| Meta DTC · mid-tier/performance | $170 | $89 ↓ −48% halo | $89 durable · TV-active non-holiday | → $170+ ↑ reverts to pre-TV baseline |
| Google Nonbranded Search mid-tier · category capture | $32 | $21 ↓ −33% halo | $21 TV-driven category queries | → $32+ ↑ reverts |
| Snapchat mid-tier · paid social brand-layer | $282 | $125 ↓ −56% halo | $125 brand-warmed retargeting pool | → $282+ ↑ reverts |
| Google PMax performance · cannibal regime | $7 | $20 ↑ +177% · cannibalizing branded search | $20 workbook β=0.14 = heavy diminishing | $7 cannibal effect resets at TV-dark |
Pre-TV baseline is what the brand was paying per channel before TV ran. Numbers above are the brand's actual values from the Northbeam Actuals workbook (18 pre-TV months Feb 2024 - Jul 2025 vs 4 TV-active non-holiday months Aug/Sep/Oct 2025 + Jan 2026 · Q4 holiday excluded per workbook's own seasonality flag). Halo channels (Meta −48%, Google Nonbranded −33%, Snapchat −56%) settle at compressed CACs as long as TV holds the optimal range. Cannibal channels (Google PMax +177%, Branded Search not shown but +106%) get worse as TV scales because PMax aggressively bids on branded queries that the brand would otherwise win cheaply via direct branded search. TV-driven brand-search demand inflates PMax's captured spend without proportional incremental conversions. This is the read most operators don't see in vendor mix model: halo + cannibal at the same time, not just halo.
For every $50K/week of sustained TV spend, the rest of the portfolio earns roughly the headline figure of incremental revenue beyond direct TV-attributed sales - brand-search visits, direct site visits, retargeting conversions that would not have happened if TV were dark. Pull TV down and that halo revenue decays over 2-4 weeks (per L4 ad-stock half-life), on top of losing the direct TV-attributed revenue. Range $656K-$963K. 9 weeks of data · tightest fit · high confidence. Solid enough to plan against.
Show the mathHow marginal CAC + the $15K/wk halo revenue number are calculated
Marginal CAC calculation. Rolling 14-day least-squares regression of incremental conversions on incremental spend per channel, fit over a 60-90 day window. Slope of that regression at the current spend tier = marginal CAC.
Why marginal ≠ average. Marginal forward-looks (cost of next customer); average backward-looks (cost of customers won so far). Industry-standard for budget allocation; what every causal-mix-model vendor outputs as the actionable number.
Halo revenue calculation. Regressed weekly revenue from non-TV channels against weekly TV spend over 43 weeks, controlling for non-TV spend levels and seasonality. Each $1K of TV spend produced roughly $1.83 of incremental non-TV revenue. At $50K/week of sustained TV → ~$15.4K/week of incremental revenue beyond direct TV attribution. Range $8.2K-$22.6K · high confidence.
Per-brand calibration. Each brand's baseline, halo magnitude, ad-stock decay, and per-channel durability profile recomputes from that brand's own CTV reach partnership data + first 90 days of paid history. Methodology is universal · numbers are per-brand.
The match-investment offer for any UA brand partner. "Match $50K of weekly TV spend. We'll show you the +$15K/week of incremental halo revenue across your other paid channels - with a 95% confidence interval you can run your own numbers against." That's the underwriting math, not a deck claim.