Marketing cases decide where the next dollar of spend goes when last-click lies. They are not a full retailer P&L and not a TMT ads-inventory case, though they share a few metrics.
Last full pass: 24 August 2026.
MBB and boutique interviewers use marketing prompts to test whether you know that attribution is not causation. A 4.2 last-click ROAS can be a 1.6 incremental ROAS. Brand and performance are not two religions; they are two measurement problems. If the prompt is shelf, comps, and four-wall, use the retail case interview. If it is CPG trade spend, use consumer goods. If it is a streamer’s ad tier, use TMT. This page is budget allocation, pricing communication, and acquisition funnels.
What you are deciding
- Reallocate spend across channels given incrementality, not last-click.
- CAC vs LTV (and payback) for a subscription or marketplace.
- Promo vs brand: a 20% off event that pulls demand forward is not new demand.
- Targeting / mix: who is incremental vs who would have bought anyway.
Clarify the KPI: incremental contribution, not clicks. Clarify the time lag (brand). Clarify whether CRM / existing customers are in the “acquisition” number (they should not be). In the first minute, also ask whether the client is acquiring new customers or stealing from a retail partner’s traffic — a DTC CAC that looks cheap can be channel conflict with the account that still sells 80% of volume.
A clean opening structure is: (1) what “return” means (contribution, not revenue), (2) causal vs last-click evidence, (3) brand vs performance time horizon, (4) creative and conversion (the non-media lever), (5) organizational constraint (agency, brand safety, retailer MDF). If you skip (2), you will reallocate into the channel with the best dashboard, which is usually branded search harvesting demand that already existed.
Exhibits you should expect
| Exhibit | Trap |
|---|---|
| Last-click ROAS by channel | Search harvests demand that TV created. |
| Geo or holdout lift | The gold standard in a case. Use it even if the sample is ugly. |
| CAC, LTV, payback | LTV needs contribution margin and retention, not revenue. |
| Funnel (impr → click → convert) | A CRO fix can beat a media increase. |
| Promo calendar vs baseline | Seasonality masquerading as campaign lift. |
Units that trip people
ROAS vs incremental ROAS vs contribution. Revenue / spend can be high while incremental contribution is negative after COGS.
CAC on the wrong denominator. Cost / new customers, not cost / all orders.
LTV as revenue. Use gross profit after variable costs and an honest churn. For paid social, also tax the LTV with refunds.
Brand vs performance time scales. You cannot kill brand in a 6-week test and call it a full answer; you can still refuse to increase unmeasured brand without a proxy (search lift, branded search, holdout).
Practice a marketing incrementality case
Run a scored case where last-click and incrementality disagree, and you have to pick a budget.
Worked mini-case: cut search that “works” on last-click
Prompt. Northline CPG has $12.0m annual media. Paid search $7.0m at 4.2 last-click ROAS → $29.4m attributed revenue. A geo holdout says incremental ROAS on search is 1.6. Contribution margin 40%. Brand TV is “unmeasurable” in the last-click model. Can we cut $3.0m of search?
Math. Incremental revenue from $7.0m search ≈ 7.0 × 1.6 = $11.2m. Incremental contribution ≈ 11.2 × 0.40 = $4.48m. Spend $7.0m to get $4.48m contribution → −$2.52m vs not spending (plus whatever brand effects search click-steals).
Cut $3.0m search: lost incremental revenue ≈ 3.0 × 1.6 = $4.8m; lost contribution ≈ $1.92m; save $3.0m spend → +$1.08m contribution. That is the so-what.
Do not dump the $3.0m into unmeasured TV by default. Put it into (1) a brand holdout in a few DMAs, or (2) retail trade if sell-through data shows a real gap (then you are in CPG/retail), or (3) hold as profit if the brand is already salient. Recommendation: cut the $3.0m from generic search (not branded navigational terms, which are harvest), keep branded search, and fund a 90-day TV incrementality test with $1.0m of the savings.
What a generic profitability tree misses here
- Spend is not “variable COGS.” It is an investment with a causal return.
- Volume from last-click is not incremental volume.
- Price and promo steal from future weeks (pull-forward).
- Channel conflict (DTC ads vs retailer) is a mix issue, not a 4P slogan.
- Creative and landing page can dominate media. A tree that only has “more spend / less spend” is incomplete.
If you answer Northline like a retail comps case, you will talk same-store sales. If you answer it like TMT, you will talk ARPU. This case is causal return on the next dollar.
See where you stand on a marketing case
Practice incrementality math and the last-click trap.
Related guides
CoachNed is independent and unaffiliated with McKinsey, BCG, Bain, or other firms named for interview-style practice. Cases on CoachNed are AI-simulated.