Media mix: assigning a role and a horizon to every spend.
Key distinctions
Three conditions before calculating.
Role before channel
The same channel can create demand in one campaign and capture it in another. The role depends on targeting, message, format, sequence and audience, not on the platform's name.
Differentiated windows
Direct measurement, persistence and brand effect must be observed over windows consistent with their mechanism. A short window does not refute a long effect; it simply does not allow it to be measured.
Interaction without double credit
When two media together produce an effect greater than the sum of their separate effects, the interaction must be isolated. Re-crediting it in full to both channels inflates the portfolio and distorts the allocation.
Method
Moving from a channel ranking to a portfolio of roles
- 01Formulate the mechanisms
For each activity, state the audience, the intermediate behaviour, the expected time lag and the outcome metric; an activity without an explicit mechanism remains a spend to diagnose.
- 02Choose the evidence
Match individual experiments, geo tests, MMM, surveys or journey data to the role and to the available granularity, and keep incompatible estimands separate.
- 03Build the scenarios
Project direct, assisted and persistent effects without adding them twice, and apply margin, media cost and an uncertainty range to each allocation scenario.
- 04Revise by marginal increment
Compare the next budget increment, not the historical average return, and reserve a learning budget for the roles whose uncertainty prevents a robust decision.
The role × horizon × evidence matrix
The media mix is shown as a time chain: creation, capture, conversion and retention, with the evidence suited to each role.
A balanced portfolio creates €31,350 of additional contribution
Two plans spend €80,000. The capture plan allocates €60,000 to search, €10,000 to video and €10,000 to CRM. The role plan allocates €30,000, €40,000 and €10,000 respectively. The margin per incremental sale is €55.
Capture plan: 1,100 search sales + 150 video + 50 assisted + 230 CRM = 1,530 sales; net contribution 1,530 × 55 − 80,000 = €4,150. Role plan: 900 + 600 + 350 + 250 = 2,100 sales; net contribution €35,500.
The role plan is adopted if the 350 assisted sales are confirmed by a causal design. Without that evidence, they become a sensitivity assumption and not a certain credit.
Acceptance conditions
What must be true to act.
- 01
Each activity has a role, an audience, a time lag and expected evidence.
- 02
Direct, persistent and interactive effects are not counted twice.
- 03
The scenarios apply the same margin and the same economic horizon.
- 04
The allocation is revised on marginal returns and on uncertainty.
Limits
What this analysis does not prove.
- Roles change with targeting, creative and spend level.
- Brand effects are hard to convert immediately into contribution.
- An interaction observed in one sequence is not automatically transportable to another.
References
Works cited.
- Naik, P. A. & Raman, K. (2003), “Understanding the Impact of Synergy in Multimedia Communications”, Journal of Marketing Research, 40(4), 375–388.
- Stephen, A. T. & Galak, J. (2012), “The Effects of Traditional and Social Earned Media on Sales”, Journal of Marketing Research, 49(5), 624–639.
- Lewis, R. A. & Rao, J. M. (2015), “The Unfavorable Economics of Measuring the Returns to Advertising”, Quarterly Journal of Economics, 130(4), 1941–1973.