The Channel That Looks Great and Isn't
A direct-to-consumer brand spends across paid social, search, email, and affiliate. Last-click attribution shows paid social delivering a 4.2x return on ad spend and affiliate at 1.6x, so the team plans to shift 30% of budget from affiliate to paid social. A marketing analyst notices that affiliate partners run coupon and review sites that customers visit late in the purchase journey, and that email revenue spikes in the 48 hours after paid social campaigns run. Total blended return across all channels is 2.8x and has been flat for three quarters despite rising spend. Customer acquisition cost has increased 22% year over year while average order value is flat.
1. Why is the proposed 30% shift likely to be a mistake, and what is the analytical flaw?
Read the worked answer
The flaw is treating last-click attribution as a measure of incremental contribution. Last-click systematically over-credits channels that sit near the point of purchase and under-credits channels that create awareness or assist earlier.
Evidence in the setup:
- Affiliate partners are coupon and review sites, classic late-journey touchpoints. Their 1.6x reflects that they capture credit for demand created elsewhere, not that they create little value.
- Email spikes 48 hours after paid social campaigns, which suggests paid social is doing awareness work whose effects show up in other channels. Paid social's 4.2x is partly borrowed credit.
- Blended return is flat at 2.8x while spend rises, and CAC is up 22%. If paid social were truly a 4.2x engine at the margin, scaling it should improve blended efficiency, not leave it flat.
So the shift would likely move budget away from a channel whose true contribution is understated toward one whose contribution is overstated, and the flat blended return is the tell.
2. What should the team do to measure incremental contribution before reallocating?
Read the worked answer
Move from attribution to experimentation:
- Run geo holdout or incrementality tests: hold paid social spend flat or off in matched markets and measure total revenue change, not channel-attributed revenue.
- Run a similar test for affiliate, or reduce affiliate spend in a subset and observe whether total orders fall or simply shift to other channels.
- Compare marginal return, not average return: measure what the next dollar in each channel produces, since average ROAS hides diminishing returns at higher spend.
- Reconcile with the blended 2.8x and the 22% CAC increase as the ground truth the channel numbers must explain.
Decision rule: reallocate only where incremental return is demonstrably higher, and size the shift to what the test can support rather than to a round 30%.
3. What does the flat blended return alongside rising spend suggest about the overall plan?
Read the worked answer
It suggests the business is at or near diminishing returns on its current channel mix and creative, and that adding spend is buying volume at rising cost rather than improving efficiency. Flat AOV compounds this: the CAC increase isn't being offset by higher order value.
Implications:
- More budget into the same mix is unlikely to fix the trend; the constraint may be creative, audience saturation, or offer rather than channel allocation.
- The team should test new audiences, creative formats, or offers, and consider retention and repeat purchase as a lever on CAC payback, not just acquisition.
- Set an efficiency guardrail (e.g., blended return must hold above a threshold as spend scales) so the plan self-corrects rather than drifting.
The right question shifts from 'which channel gets the budget' to 'why is the whole machine getting less efficient,' which the channel reallocation debate was masking.