Are you making critical budget decisions based on data you don’t fully trust? It’s a common and costly problem in modern marketing. Your ad platforms may claim success, but the numbers often don’t add up, leaving you to question every dollar you spend. The solution isn’t another dashboard, but rather a new way of thinking about the relationship between day-to-day execution and long-term strategy.
In this episode, host Brad Keefer, CRO of Keen, sat down with Ryan Koonce, CEO of Attribution App. They detailed a groundbreaking approach that combines Multi-Touch Attribution (MTA) and Marketing Mix Modeling (MMM) into a single, cohesive system that provides clarity, confidence, and a true measure of marketing ROI.
Key Takeaways
- Most attribution solutions can’t connect cost data to actual customer journeys, limiting their business value
- Platform attribution models contain structural bias because they’re designed to justify increased spending
- Real multi-touch attribution requires identity resolution to track customers across devices and touchpoints
- The combination of deterministic MTA data and probabilistic MMM creates more reliable forecasting capabilities
- Modern implementation approaches have reduced setup complexity through one-click platform integrations
Can You Trust Your Marketing Data?
Every marketer knows the story. You spend money on Google and Meta, and both platforms take credit for the same conversion. As Ryan Koonce puts it:
“Let’s say you spent five dollars for a click on Google and five dollars for a click on Meta, and you made eight dollars. Well, Google’s gonna say you made three, and Meta’s gonna say you made three, but you lost two. And so how do you reconcile that where you can audit the data and really have the confidence that what you’re looking at is true?”
Ryan Koonce, CEO, Attribution App
This problem goes beyond a simple reporting glitch to become a fundamental breakdown of trust. These “walled gardens” have an inherent bias, as Brad Keefer notes. Their attribution models are incentivized to prove their own value, creating a dangerous cycle of spending more money based on flawed data.
What Best-in-Class MTA Actually Looks Like
In a market flooded with so-called “attribution” tools, Ryan defines the core principles that differentiate a true MTA platform. He warns marketers to look for two major red flags in any potential vendor:
- They cannot measure true ROAS. Many tools only attribute a percentage of conversions to a source but fail to incorporate the actual cost. “If they can’t tell you the real ROAS,” Ryan warns, “that should be a big red flag.”
- The data is not auditable. Many solutions are “just session based or click based,” with no ability to stitch together a user’s journey across devices and time. A real MTA solution provides full identity resolution, allowing you to drill down to the user level and verify the data for yourself.
Without these capabilities, you are likely looking at what Ryan calls “a prettied up version of Google Analytics, which quite frankly just isn’t gonna do it if you’re really trying to optimize your ad spend.”
Connecting Deterministic Execution and Probabilistic Strategy
For years, marketers have treated MTA and MMM as separate disciplines. MTA provides deterministic data, which is a granular record of what happened at the user level. MMM provides a probabilistic model, which is a high-level, strategic view of what could happen.
The partnership between Attribution App and Keen finally unites them. “I think the MTA is the foundation in which you build the house,” Ryan says.
This creates a powerful, closed-loop system for decision-making in three stages:
- First, the system creates an informed strategic model by feeding Attribution App’s clean, user-level data into Keen’s Bayesian MMM. This provides a foundation built on ground truth rather than assumptions.
- Second, it generates an actionable, forward-looking plan by using the data to create an optimized media plan and a probabilistic revenue forecast. This moves beyond simple measurement to an accountable prediction of future results.
- Finally, the system offers diagnosis and validation by comparing the forecast to actual results at the end of a quarter. This closes the loop and allows marketers to determine if a variance was caused by the media plan or by a shift in consumer behavior, an insight only available by analyzing the underlying MTA data.
A New Standard for Accountable Forecasting
For Keen, measurement for the sake of measurement is a dead end. “We always say… ‘so what?'” Brad remarks. “At the end of the day, if you can’t turn that into the so what therefore, what does it matter?”
This philosophy drives Keen’s focus on putting its “money where its mouth is” with an actualized forecast. It’s a commitment to predicting an outcome, then analyzing the results to understand every driver of performance, from the media plan to macro-economics to shifts in consumer behavior.
Demolishing the Implementation Barrier
The most common objection to adopting a sophisticated analytics stack is the fear of a long and painful implementation. “I just don’t know if we’re ready… I don’t have the resources to set it up,” is a constant refrain.
This fear may be outdated. Ryan explains that for platforms like Salesforce, HubSpot, and Shopify, it’s a “one-click install.” He notes, “Getting it set up is usually a day of work. It’s not that big a deal.”
The partnership streamlines this even further. Once Attribution is live, the clean, structured data is ready for Keen’s MMM, dramatically reducing the time and cost to get both systems running and delivering value.
AI, Data, and the Future of Marketing Analytics
Looking ahead, Ryan believes we are in a “golden era of MTA,” precisely because of the rise of Artificial Intelligence.
“As some of the new reasoning models come out with AI, you have to have accurate, reliable, consistent, accessible data in order to use them, and we’re the only company that can provide that data.”
Ryan Koonce, CEO, Attribution App
However, Brad offers a crucial dose of reality about the state of AI today. LLMs are trained to find the “fastest and most logical answer,” which “doesn’t make it the most right or the more right answer.”
The true, lasting value isn’t in a flashy AI interface. It’s in the one thing that makes trustworthy AI possible: a pristine, reliable data infrastructure.
Learn More & Connect
- Ryan Koonce & Attribution App: To get a complete, auditable picture of your marketing performance and build your data foundation, visit AttributionApp.com.
- Brad Keefer & Keen Decision Systems: To transform your marketing data into forward-looking, optimized financial plans that drive growth, visit KeenDS.com.
