There is a familiar ending to an analytics project. The data has been scrubbed. The model has been defended. The deck is thick enough to stop a door. Everyone admires the rear-view mirror, then walks into a meeting to decide the future by habit, politics and whoever has the last available slide. John Busbice has spent much of his career objecting to that ending.
His objection is not theatrical. Busbice is a decision scientist, the sort of person who takes uncertainty as a working condition rather than a personal affront. At Keen Decision Systems, where he is founder and Chief Decision Science Officer, he leads the development of the company’s analytical methods. Keen calls him the “father” of its algorithms. The family resemblance is easy to spot: they are built to forecast, compare and revise, not simply to award medals after the race.
The question beneath the machinery is disarmingly practical: where should the next marketing dollar go? Not where did the previous dollar go. Not which channel can claim it with the most confident chart. The next dollar, with its still-unwritten consequences, awkward alternatives and impatient owners.
A better question than “What happened?”
Busbice’s education gave him a useful double exposure. He earned a bachelor’s degree in business administration from Appalachian State University, then an MBA from Washington University in St. Louis. By 2007 he had been promoted to principal at IMS Consulting. The work taught him how serious organizations consume analysis, and how readily a technically impressive study can fail the test of use.
Greg Dolan, who would become Busbice’s co-founder, later described meeting a data scientist conducting exhaustive studies with little utility for decision-making. Dolan came from the other side of the table. He had managed brands and large budgets, yet lacked tools that could connect spending choices to financial outcomes. One had models that did not travel far enough. The other had decisions that the available models did not properly serve. Their frustrations were compatible.
They helped start Keen in North Carolina in 2010. Busbice also developed Marketing Investment Decision Analytics, or MIDA, a forward-looking application built around investment choices. When Keen acquired MIDA in 2013, he joined as a managing partner and brought the decision-support software with him. The arrangement made the company’s thesis concrete: analytics would no longer be delivered as a handsome autopsy. It would sit closer to the planning table.
“Marketing mix and investment decisions are among the most challenging for marketers.”John Busbice, on the launch of MIDA at Keen
That sentence is revealing because it does not pretend the difficulty can be engineered away. Marketing decisions contain rival theories, partial data, changing conditions and human incentives. Busbice’s answer has been to make those complications visible and manageable. MIDA incorporated both organizational data and managerial guidance. It treated judgment as information to be examined, rather than a stain to be washed out of the spreadsheet.
The algorithm must survive the meeting
Busbice is associated with Bayesian methods, which begin with what is already known or reasonably believed, then update those beliefs as evidence arrives. In the abstract, that sounds like statistics. In a company, it is also manners. The model enters the room without claiming omniscience. It says: here is our starting view, here is what the new signal changes, and here is how uncertain we remain.
That humility has a hard commercial purpose. A marketing team needs to compare channels that behave differently over time, account for outside forces and decide within constraints. Finance wants the choice expressed in terms it can recognize. Leadership wants to know the risk. A useful system has to connect those conversations without turning the organization into an advanced-statistics seminar.
Busbice’s public comments repeatedly return to this bridge. “You need to have good information and understand what to value and what kind of business outcomes you’re after,” he has said. Then comes the less comfortable half: “You need to understand the risk and all the alternatives.” Numbers do not remove the trade-off. They improve the quality of the argument about it.
The seduction of tiny boxes
In 2023, Busbice and Dolan recorded a conversation about granularity. Marketers naturally want sharper detail: one more audience slice, one more market, one more box in the table. Precision looks responsible. Busbice’s framework asks the detail to pay rent.
He names three constraints: cost, complexity and conflict. More granular data costs money to gather and maintain. It makes a model harder to operate and explain. It can also multiply disputes, as teams defend a product line, geography or channel at a level the evidence cannot reliably support. A fine-grained answer to the wrong organizational question is merely a more expensive way to be distracted.
This is where Busbice’s work becomes less about mathematical horsepower and more about product judgment. The proper level of detail depends on the decision hierarchy. A weekly channel move does not require the same lens as an annual portfolio choice. The system should be precise enough to guide the decision, and no more ornate than the decision can bear.
“The power of MMM only accelerates when it supports an embedded decision process and is democratized across teams.”John Busbice, 2025
A company drawn around a belief
Before the platform acquired modules and the team acquired departments, Busbice and Dolan spent hours defining the values by which they wanted to run the company. That exercise can look suspiciously soft beside the algebra of optimization. It was, however, another form of model building. They were choosing the rules that would guide decisions once the founders could no longer be present for every one of them.
The pairing itself had a useful tension. Dolan understood the accountability of the budget owner. Busbice understood the analytical machinery and its limits. Keen’s founding proposition required both perspectives to remain in conversation. A product that delighted analysts but baffled operators would fail. So would a friendly planning tool whose recommendations could not withstand technical scrutiny. The company grew in the narrow country between those failures.
Busbice’s formulation also changed the noun at the center of the work. Marketing “spend” sounds like money disappearing through a trapdoor. Marketing “investment” invites a question about future value. He has argued for treating a brand as an asset that can be built in financial terms. This does not make creativity a spreadsheet cell. It asks the organization to state what it expects an expenditure to produce, over what period and with what uncertainty.
The distinction matters because language sets the terms of a budget debate. If marketing arrives as a cost, finance naturally looks for a reduction. If it arrives as an investment with modeled outcomes and visible risk, the conversation can become one of allocation. It may still be contentious. Busbice has never promised a conference room without disagreement. His work aims for a disagreement with better furniture.
From a model to a habit
Keen’s 2015 turn toward software as a service changed the delivery mechanism. A consulting study tends to arrive, impress and age. Software can remain in the workflow. It can absorb new data, rerun scenarios and compare forecasts with outcomes. That shift suited Busbice’s premise: a decision system should learn because the people using it are also learning.
It also demanded patience. Busbice has worked with some of Keen’s software collaborators for more than a decade. In describing one development partner, he praised engineers who operated as part of the organization and pushed for improvements to both product and process. The compliment is characteristic. He noticed not a dramatic rescue but durable integration, the unglamorous arrangement by which complicated systems actually improve.
Keen raised an $11 million Series B round in 2023 to accelerate product development and expansion. Funding is a milestone, but Busbice’s own public emphasis remains adoption. In a 2025 industry guide, he argued that marketing-mix modeling becomes more powerful when it is embedded across teams. Action generates feedback. Accurate forecasts earn confidence. Confidence makes the next action easier. The virtuous circle is behavioral as much as computational.
There is a modest radicalism in that view. It removes the analyst from the role of oracle and gives the model a job. The job is not to be admired. It is to help a group of people choose, observe and revise without losing the thread between marketing activity and financial consequence.
More than twenty years into his analytics career, Busbice is still working on the same stubborn hinge between knowing and doing. The vocabulary has evolved: predictive analytics, marketing mix modeling, decision optimization, AI. The essential test has not. At the end of the meeting, did the work clarify the choice? And after reality answered back, did the system learn?
A rear-view mirror remains useful. It tells you where you have been, and occasionally what you hit. Busbice simply refuses to confuse it with the steering wheel.