Shining a Light on the Dark Funnel: Overcoming Attribution Challenges with MMM

How can marketers track what cannot be tracked? The ‘dark funnel’ is one of the main challenges we all face when it comes to attribution. Over the past decade, marketers have relied on digital tracking and following consumers across their entire buying journey. However, there were “blind spots” where you’d lose visibility and they’d become untraceable. 

With digital tracking becoming more and more broken because of privacy changes and the death of the cookie, those blind spots are continuously expanding. Understanding and tracking the dark funnel is crucial for marketers to efficiently allocate resources, maximize ROI, and gain insights into consumer behavior that might otherwise remain hidden.

In this article, we will cover what is within the scope of the dark funnel and the main strategies you can use to solve for its attribution. 

The Dark Funnel: Understanding the Unknown

The dark funnel contains those channels that we are not able to digitally track whether through pixels, cookies, or referral sources. The three main causes of dark funnel traffic are:

  • Private browsing and cookie tracking issues: Privacy-focused browsers and cookie blockers impede the tracking of user activity, resulting in dark funnel traffic. 47% of internet users worldwide use ad blockers and cookies are blocked for approximately 64% of website visits – it’s very significant.
  • Social media platforms and messaging apps: Dark social sharing through private messaging apps and untraceable social media interactions contribute to it as well. 70% of millennials share links via private messaging apps – often to make product recommendations.
  • Untraceable referral sources: Some referral sources, like offline word-of-mouth or untagged links, cannot be traced either.

Now, what’s the impact of the dark funnel? Why is it so important for marketers to shine a light on it and see what is going on behind the scenes of their dashboard?

Simply put, if the dark funnel is a black hole and you operate blindly, you can’t accurately attribute your marketing efforts. The dark funnel obscures the true sources of consumer engagement. That means you then can’t measure the effectiveness of your campaigns either. 

Will you attribute to brand search the people that heard you on a podcast and Googled you? Will you think organic social isn’t performing because they don’t click the link on your bio with a referral? What’s working? What’s not? You can’t possibly know.

And if you don’t have a clear understanding of the effectiveness of your channels, it becomes very challenging to optimize your budget allocation. As far as you know, brand search, for example, could be extremely ROI-positive but actually might not be so incremental. You don’t want to waste hundreds of thousands or millions of dollars because of the dark funnel.

Marketing Mix Modeling (MMM): A Solution for the Dark Funnel

The idea that you can actually trace and track people and see what they purchase has only really been around for 10 or 15 years. Before that, everything was the dark funnel, right? 

You showed someone a TV ad and then they drove to the 7-11 and purchased a Coke. Did the TV ad cause that? No one really knew. The idea of the dark funnel being so problematic is funny to us because it is truly a new thing that was fairly short-lived. 

If we want to solve old problems, we need to look at modern evolutions of traditional attribution methods – mainly Marketing Mix Modeling. 

Unlike digital tracking, MMM doesn’t track people across the Internet. Instead, it analyzes historical data and runs statistical analysis to find whether each channel is incremental or not. It is the  pre-digital tracking attribution method, just more sophisticated and evolved to use all the technology available today.

Finding Incrementality in the Dark Funnel

Incrementality might be the most important concept to a marketer – if you increase/decrease your budget on a specific channel, what will it bring you in return? 

To measure incrementality – even in “dark funnel” channels –, MMM considers all available data, including untraceable touchpoints, providing a more comprehensive view of marketing effectiveness.

Additionally, it also incorporates offline and external factors. It accounts for economic indicators and seasonality, offering a more accurate representation of marketing performance. Then, through statistical analysis, it helps uncover hidden relationships and trends within the data, shedding light on the dark funnel.

You can then establish a feedback loop that incorporates learnings from MMM models back into your marketing strategy to get continuous improvement and optimization.

Practical Solutions

Let’s say you run a podcast and repurpose it into organic social content and want to measure its impact beyond surveys and customer/sales interviews. Most marketers would consider it part of the dark funnel. 

If you’re not using paid ads to distribute it, geo-testing and A/B tests would not be the most valuable and appropriate experiments. However, you could run a deprivation test where you turn off podcasting and see if there’s an impact on your pipeline – just make sure to account for the lag that branding deprivation tests come with. For more on deprivation experiments, we wrote an article covering them.

Find which experiment works best for each channel and use MMM platforms like Recast to run the statistical model that shares how incremental each channel really is. 


The dark funnel isn’t a problem. It’s the truth. And it’s only getting bigger. We all need to embrace that and start to understand that most of the funnel is dark.

If eight years ago you were able to successfully track 70% of people, you’re now probably under 50%, maybe under 40%, estimates Mike Kaminsky, co-founder at Recast. You’re seeing a smaller and smaller portion of the people who are actually making purchases. And it will only get worse.

We’re in a new world where marketers have much more uncertainty around consumer journeys and there’s only a very small window that they actually have insight into.

In this new world, we need to go back to (modern versions of) traditional attribution methods that don’t rely on digital tracking and can give you information through statistical analysis. That’s what we’re doing for eCom brands at Recast, and how we will be able to support them through this transition.

About The Author