The Action Loop
Information has no strategic value if it never changes a decision.
That proposition concerns the relationship between research and organizational judgment. An organization can collect substantial information without changing its understanding, its decisions or its subsequent behavior. The Action Loop makes those different activities visible rather than treating a dashboard as evidence that learning has occurred.
Detect → Decide → Deploy → Reallocate
The exact loop was originally developed in a media-planning context in Most Media Plans Are Structurally Wrong. Applying it more broadly to organizational learning is an editorial extension, informed by other essays about judgment, incentives and execution. It is not a demonstrated formula for better outcomes in every setting.
Detect
Identify a meaningful change, risk, opportunity or unanswered question. Detection is more than noticing activity: an observation needs an object, a context and an account of uncertainty. A striking result can be a genuine change, a measurement problem or a question that requires further investigation.
Decide
Apply judgment. Determine whether the signal warrants action. This is where evidence meets responsibility, competing explanations and the consequences of being wrong. An interpretation can be useful without being certain; uncertainty also cannot be made to disappear simply by producing another chart.
Deploy
Put the decision into practice. The distinction between deciding and doing matters because organizations can agree on an interpretation without carrying out the decision that follows. Implementation is a separate event whose actual form may differ from what was intended.
Reallocate
Use subsequent evidence to determine whether resources, attention, message or strategy should change. In the original essay, the clearest meaning concerns media resources. The wider organizational interpretation on this page extends that language beyond its original setting; it should not be mistaken for a separately validated general model.
A loop is not an instruction to keep changing
Faster change is not always better. Frequent adjustment can reflect noise, inconsistent measurement or uncertainty about the original question. Activity alone does not show learning, and a rapid response does not demonstrate that the response was appropriate.
Action can include holding course when new evidence does not warrant a change. In that case, the observation may still matter: it has been considered and has informed a decision to continue. The distinction is between an accountable judgment and simply ignoring evidence that is uncomfortable.
Subsequent observation also needs interpretation. A changed result after an action does not, by itself, establish that the action caused it. Conditions may have changed, the measure may be different, or the original account may have omitted something important. The loop describes an organizational process, not a substitute for evaluating explanations.
The human problem
The hardest step is often Decide. Useful intelligence can coexist with incentives, hierarchy, overconfidence or prior commitments that make an organization reluctant to reconsider. This is a recurring professional observation in my writing, rather than a universal empirical law about how every organization behaves.
The Most Dangerous Person in Politics concerns the difficulty of updating beliefs when experience becomes a source of certainty. Intelligence Doesn’t Win Campaigns distinguishes possessing information from the organization’s capacity to use it. The Oklahoma retrospective examines the gap between research and judgment, including the limitations of hindsight.
Those essays matter because they resist a purely technological explanation. More information does not automatically resolve conflicting responsibilities or make an unwelcome interpretation easier to accept. Human judgment remains part of the process, including judgment about what the available evidence cannot establish.
AI and accountability
AI can improve observation and synthesis, but it does not eliminate accountable human judgment. A generated summary or pattern still depends on source quality and interpretation. Responsibility for a consequential decision cannot be transferred to a tool merely because the tool contributed to the analysis.
The broader lesson here is modest: information, judgment, implementation and subsequent learning are different activities. The Action Loop supplies a vocabulary for their relationship. Its usefulness should be assessed through the quality of that relationship, not through the volume of reports produced or the frequency of changes announced.
Further reading
Most Media Plans Are Structurally Wrong — The published loop and its original media-planning context.
Intelligence Doesn’t Win Campaigns — Why information and organizational capacity are different.
Campaigns Are Becoming Media Companies. That’s Not Enough to Win. — An earlier discussion connecting information, decisions and execution.
The Most Dangerous Person in Politics — Experience, overconfidence and the difficulty of revising beliefs.
Lessons From Wins, Losses and a 2,045-Vote Oklahoma Runoff — A retrospective on research, judgment and the limits of hindsight.
Watch
WPA CEO Chris Wilson discusses the run for the White House on Glenn Beck
Earlier discussion of a related idea, around 3:35: observing response before interpreting a subsequent decision. This is a narrow historical precursor, not the complete loop or a controlled case study. Uploaded August 6, 2025; the discussion concerns the pre-2008 primary period, with the original broadcast date unverified.
