MMM Readiness - some hurdles to overcome (and their height)

My last 2023 #WednesdayWhizz was concerned with MMM Readiness. In other words: What are some prerequisites for implementing Marketing Mix Modelling successfully in a complex organization:

People liked it. But wanted more (details).

Beyond guidance on "what is the right moment/maturity to invest into MMM", readers wanted to know more specifically, which criteria should be met for MMM to work or provide meaningful results.

So here is Part 2 with regards to MMM Readiness. My answers to these questions are normally much more nuanced and industry / client specific. So please take it in the spirit is was written: guidelines and rules of thumb, not exact science...

Minimum Annual Net Media Spend (per Brand)

• High-Level Answer: Your overall net spend should be significant enough to impact sales measurably. For small brands this means tens of thousands, while larger brands may need to invest millions to move the needle detectable by aggregate models such as MMMs.
• Specific Answer: A general benchmark would be at least 1% of a single brand's annual revenue. For example, with an annual revenue of $10 million, the minimum (net) media spend for an MMM to measure impact would range need to be >=$100,000. Again, this is a rough rule of thumb.

Please keep in mind that especially for young, innovative brands (even within big company brand portfolios), media is not a COST but an INVESTMENT. That sounds like a truism but what I mean is, you need to be prepared for MMMs to show you negative ROIs in such cases. This is not a result of your media spend being too low for an MMM to detect its impact but the nature of the beast: When media spends are very high relative to value sales, ROIs are negative.

• Read more in defense of ROI as a metric here.
• Read more on how much media spend is too little here.

Equally important for the question if an MMM makes sense is actually the question if it makes ECONOMIC sense, i.e. does it pay off. We covered this in the previous blogpost on MMM Readiness:

Minimum Spend for a Single Media Channel

• High-Level Answer: Each channel must have enough spend to generate measurable outcomes, varying based on the channel's cost and reach efficiency.
• Specific Version: Depending on the channel, digital channels (due to their typically more targeted nature) may require at least $10,000 to $20,000 p.a. and per brand, while traditional channels might need at least $50,000 for MMM to measure impact.

More important for an MMM to give you meaningful channel level outcomes than annual sum of channel spend is definitely VARIANCE. See further below...

Minimal Amount of Weeks per Year That a Channel is Used

• High-Level Answer: Consistency and duration are key; a channel should be used for enough weeks to gather meaningful data and observe trends.
• Specific Answer: (Smaller spend) digital channels should be active for 20-30 weeks p.a., while traditional channels should remain active for at least 10-15 weeks.

Variance in Channel Spend

• High-Level Answer: Assess if the spend and activity vary enough over time to isolate their impact on sales.
• Specific Answer: Ensure that spending in each channel varies by at least 30-50% p.a. across different time periods (weeks).

Other Criteria

• High-Level Answer: Include enough historical data, competitive spends and include market conditions (media inflation, market growth) in your MMM exercise - and take care of data consistency over time.
• Specific Version: Aim for at least 3 years of historical data (assuming weekly granularity), get competitive gross spend per week from Nielsen (often via your media agency) and maintain consistent data structures AND media taxonomy over time.

If both your media spends and your sales (target variable) are available on a daily level (as is the case with many of our ecommerce clients), you can reduce the history needed to 1,5-2 years. If you sit on monthly data only, 5 years are typically required and granularity of answers is most likely reduced.

The upshot

Remember, these guidelines are starting points, and adjustments may be necessary based on specific circumstances such as

• industry sector
• client / brand maturity
• data availability and consistency breaks in the data that you should avoid and therefore potentially limit the data available for modeling.

The synergy of business acumen (yes, that includes gut feel!) and attention to detail and consistency in implementation is key to effective media spend optimization.