Marketing-mix audit: investigating the gaps, not handing out marks.
Key distinctions
Three conditions before calculating.
Traceable finding
An observation must refer to a source, a period, a definition and a calculation rule. "The price is too high" is not auditable; "the conversion rate falls by 18% above €79 across three comparable cohorts" is.
Graded evidence
Stated, observed, compared and causal are not interchangeable. The level of evidence limits the authorised decisions: a correlation can trigger a test, not always a rollout.
Non-compensable risk
Compliance, ability to deliver, unit contribution and minimum data quality must be treated as gates, not as points in an average.
Method
Conducting the audit from finding to decision
- 01List the active decisions
Document the choices genuinely in force — assortment, net price, promotional rules, coverage, messages, budgets and measurement methods.
- 02Test the interfaces
Audit first product-price, price-channel, channel-service, message-evidence and budget-capacity. The large losses often sit between two areas of responsibility.
- 03Qualify the evidence
Assign to each finding a source, a freshness, a comparability, a level of identification and a range of uncertainty.
- 04Prioritise by regret
Combine potential impact, probability, cost of verification and reversibility. The most urgent action is not always the one with the highest average gain.
The non-compensatory grid
Each dimension receives an evidence status and a threshold. A single critical blocker can stop the recommendation, even if the overall average looks favourable.
Secondary information
Growth is not authorised
| Interface | Evidence | Threshold | Observation | Decision |
|---|---|---|---|---|
| Product-target | Comparative usage | ≥ 70 | 82 | pass |
| Product-price | Choice + margin | ≥ 65 | 68 | pass |
| Price-channel | Net contribution | ≥ €0 | −€3.20 | block |
| Channel-service | Availability | ≥ 95% | 96.4% | pass |
| Message-evidence | Comprehension | ≥ 75 | 90 | pass |
| Measurement-decision | Comparator | required | partial | investigate |
An average of 76 out of 100 that must lead to a stop
An offer scores 90 on attractiveness, 82 on communication, 78 on distribution, 74 on data quality and 56 on contribution. The weighted average reaches 76.
The contribution threshold was set at 65 and declared non-compensable. The margin after discounts and cost to serve is −€3.20 on the channel that represents 38% of sales. The right priority is therefore not to improve the overall score but to correct or close that combination.
Red status on price-channel, freeze of acquisition on the loss-making perimeter and analysis of the discounts before any growth recommendation.
Acceptance conditions
What must be true to act.
- 01
No critical blocker is diluted in an average.
- 02
Every finding states source, period, unit and level of evidence.
- 03
The proposed action states impact, cost of verification and reversibility.
- 04
The owner of the correction and the counter-audit date are named.
Limits
What this analysis does not prove.
- An audit locates and prioritises; it does not automatically prove the cause of a gap.
- The thresholds must be adapted to the economics and to the risk of the organisation.
- A grid that is too long creates an illusion of exhaustiveness and reduces the depth of the critical checks.
References
Works cited.
- Kaplan, R. S. & Norton, D. P. (1992), “The Balanced Scorecard”, Harvard Business Review.
- Keeney, R. L. & Raiffa, H. (1976), Decisions with Multiple Objectives.
- Shadish, W. R., Cook, T. D. & Campbell, D. T. (2002), Experimental and Quasi-Experimental Designs.