On why some leaders transform after development and others don’t — how to design programs that are built for transfer from the start — and the question every evaluation framework eventually arrives at.
You designed the program carefully. You measured satisfaction. You tested knowledge. The feedback was positive. And then — in most organizations, for most leaders — the behavior back at work stayed largely the same. This is not a training problem. It is a systems problem. And there are four questions that get to the root of it.
The pattern is familiar enough to feel almost inevitable: a well-designed leadership development program produces enthusiastic participants, high satisfaction scores, and very little observable change in how leaders actually show up at work three months later. The investment was real. The intention was genuine. The results were not.
Understanding why this happens — and how to design against it — requires moving past the program itself and looking at the conditions surrounding it. Two frameworks address the design and selection decision directly. Two more address the measurement question. Together, they point toward a shared requirement that most organizations have not yet built.
Question 1
Are you designing programs — or environments?
Brinkerhoff’s Success Case Method
Robert Brinkerhoff’s most durable contribution to the field is a deceptively simple observation: when you measure the average outcome of a training program, you obscure the most useful information available. The average hides both the leaders for whom the program worked exceptionally well and those for whom it produced almost nothing.
His method — the Success Case Method — deliberately studies both extremes. Identify the leaders who applied the learning most effectively after the program ended. Identify those who applied it least. Then investigate, through structured inquiry, what actually made the difference between those two groups.
The findings are consistent enough to be treated as a principle: the gap between high-transfer and low-transfer leaders is almost never explained by the content or quality of the training. It is explained by what surrounds the learner back at work.
The practical power of Brinkerhoff’s approach is that it produces actionable intelligence rather than just evaluation scores. Rather than asking whether a program was effective on average, it asks effective for whom and why — which generates far more useful insight for every subsequent investment decision.
Question 2
Are you designing forwards — or backwards from the outcome?
The New World Kirkpatrick Model
Introduced by James and Wendy Kirkpatrick as a contemporary evolution of the original four-level model, the New World approach contains a central design principle that most L&D planning processes still violate: always begin with the end in mind — specifically, with the Level 4 business result the initiative needs to serve.
The conventional sequence is forward-looking: identify learning objectives, design content, deliver the program, evaluate results. The New World model reverses this entirely. You begin by defining the Level 4 business result, then move backwards to the Level 3 behaviors that would produce that result, then identify the required drivers that will sustain those behaviors, and only then design the program itself.
The concept of required drivers is the New World model’s most practically significant contribution. Required drivers are the organizational mechanisms — systems, conversations, feedback structures, recognition practices — that reinforce, encourage, reward, and monitor the application of new behaviors after the program ends. Without them, even the most thoughtfully designed program produces short-lived change.
Question 3
Who is actually watching whether behavior changed?
Kirkpatrick Level 3 · Learning Transfer Evaluation Model
Kirkpatrick’s Level 3 — behavior change on the job — is where almost every organization runs out of road. Levels 1 and 2 are measurable in the training room, in an environment you control. Level 3 happens back at the desk, in real meetings, under real conditions. And unlike the room you designed, you are not there to observe it.
This is not a data collection problem. It is a structural one. The evidence that Level 3 requires — behavioral change in the actual work environment — can only be gathered by people who are present in that environment. Not by the leader themselves. Not by a retrospective survey months after the program. By the colleagues, team members, and managers who experience that leader’s behavior every day.
The Learning Transfer Evaluation Model (LTEM), developed by Dr. Will Thalheimer, sharpens this point across eight tiers of evidence. Tiers 1 through 6 measure things that can be observed inside or immediately after a learning experience — attendance, activity, satisfaction, knowledge acquisition, decision competence, task competence. Tiers 7 and 8 — transfer and effects of transfer — require evidence from the actual work environment. They cannot be self-reported. They cannot be inferred. They must be externally verified.
Question 4
If behavior changed — how would anyone know?
The question every framework in this series eventually arrives at
This is not a rhetorical question. It is the most operationally concrete question an L&D function can ask — and its answer, or the absence of one, reveals more about an organization’s development culture than any program evaluation ever could.
Trace the logic of each framework in this series: the 70-20-10 model says the highest-leverage development comes from the 20% — learning from others. Phillips ROI requires pre-and-post behavioral data to make return measurable. Brinkerhoff’s high-transfer leaders have feedback-rich environments. The New World Kirkpatrick model lists feedback from the people around the leader as an explicit required driver. LTEM Tier 7 demands external verification of transfer.
Every framework, from a different direction, requires the same thing: a mechanism that gathers honest, structured input from the people who experience a leader’s behavior directly — before development and again after.
Without that mechanism, Level 3 remains invisible. ROI is an estimate. Required drivers have no feedback loop. The high-transfer environment that Brinkerhoff describes never fully forms. The 20% in 70-20-10 stays underfunded and unmeasured. And the investment made in programs — however well designed — produces change that no one can see, verify, or build on.
A final thought on where to start
None of these frameworks require an organization to overhaul its L&D function overnight. They each suggest a more modest starting question — one that, asked honestly, tends to clarify investment decisions considerably.
For Brinkerhoff: before your next program, identify two or three leaders for whom the conditions for transfer already exist — and two or three for whom they don’t. Design accordingly.
For New World Kirkpatrick: before your next design process, write down the Level 3 behaviors you need at the end of the program. If you cannot be specific, the program is not yet ready to be designed.
For LTEM and Kirkpatrick Level 3: ask what mechanism you have for observing behavior change from the outside — from the people who are actually present in the work environment every day. If the honest answer is “none,” that is the gap worth closing first.
The question is not whether your leaders are developing. It is whether you have built the conditions to see it — and to know.
Curious how other organizations are connecting L&D investment to business outcomes?
We’d be glad to share what we’re seeing — no pitch, just a conversation.
