You spent millions on a brand awareness campaign – great! But now you’re left with a tough question: how do you measure if it worked?
Unlike performance marketing, where you can measure ROI in days or weeks, brand campaigns often deliver results over months or even years. And, like every other measurement method out there right now, media mix modeling (MMM), also struggles to provide great clarity for brand-focused initiatives.
In this article, we’ll explore why brand campaigns are so difficult to measure with MMM, what MMM can and can’t do well, and what marketers can do to get a clearer picture of their impact.
The Challenges of Measuring Brand Campaigns with MMM
Measuring brand campaigns with MMM is one of the hardest challenges in statistical modeling, and there are a few key reasons why.
1. Noisy, low-resolution data.
Unlike sales data, which is often precise and available daily, brand metrics like awareness and consideration come from surveys. These surveys are typically conducted weekly or monthly, and even the best surveys have a lot of errors. Small changes in awareness caused by advertising can be buried in the noise, which makes it hard for MMM to reliably detect them.
2. Long-term effects.
Brand campaigns work over long time horizons, building mental availability and loyalty that might not pay off for months or even years. Most MMMs have a measurement horizon of about 90 to 120 days. That means when you spend a dollar today, we can confidently track its effects for about 120 days.
Beyond that, it becomes increasingly difficult to draw a straight line between a brand campaign and sales outcomes. Connecting today’s investment to future results is one of the toughest challenges for any measurement method.
3. Statistical complexities.
Even with good data, it’s hard to separate faint brand signals from the noise created by macroeconomic trends, competitive activity, and other external factors. The more distant the brand effects are in time, the more difficult this becomes.
4. Proxies that can mislead.
Some companies use branded search volume or organic search volume as a proxy for brand awareness. While these proxies can be helpful, they often mix unrelated behaviors.
For example, someone searching for “Honda” might be looking to buy a car, but they could also be trying to troubleshoot an issue with their existing Honda. Without filtering for intent, these proxies aren’t actually telling you much – and they might just be biasing you.
What MMM Can Do Well for Brand Measurement
MMM isn’t perfect for brand campaigns, but there are specific areas where it can be valuable.
1. Measuring the impact of contextual variables.
Recast uses what we call contextual variables—factors like brand awareness, consideration, and price—that influence overall marketing performance. For example:
- A 5% increase in brand awareness might lead to a 5% lift in organic demand and improve the ROI of your paid media efforts.
- Price changes might reduce new customer acquisition rates but increase long-term profitability.
These variables act as multipliers in the model, lifting or suppressing the performance of all advertising at once. While MMM can’t measure how specific channels affect awareness, it can estimate how changes in awareness impact outcomes like sales.
Short-Term vs. Long-Term Brand Effects
Brand campaigns deliver two types of effects: short-term and long-term, and each requires a different measurement approach.
Short-term effects are immediate. For example, a radio ad that drives immediate purchases during a holiday weekend is easier to measure because the effects occur within the MMM’s typical time horizon experiments like geo-holdouts. The goal here is to connect ad spend to a lift in sales over a baseline so you can attribute additional sales to the campaign.
Long-term effects are more elusive.
These are the cumulative results of building mental availability over time. For example, a person exposed to your campaign might choose your brand six months or even a year later. Measuring this with MMM is difficult because effects attenuate after the 120-day horizon most models handle.
But we’ve got a workaround. As we mentioned above, surveys on brand awareness can help fill this gap and give you a way to understand how your campaign is building mental availability without relying solely on purchases. Statistical models can then estimate the relationship between brand awareness and downstream performance.
Even if we can’t directly measure every single outcome over a year or more, this helps us capture the essence of long-term brand effects.
Conclusion
While some claim to have solved this problem, the reality is that it’s an area that needs a lot more research and development. We’d rather be transparent on what MMM can and can’t do so you can actually trust the model.
If you’re interested in this topic, we’ve discussed it with measurement experts in previous Recast’s Coffee Breaks sessions. Check these ones out – you might find them valuable: