A leadership program worked if, six months later, participants lead differently and their teams perform better because of it. Everything else is a proxy.
Most organizations never check. They hand out a feedback form on the last afternoon, collect an average of 4.6 out of 5, and file the program as a success. That score tells you the room was comfortable, the facilitator was engaging and lunch was good. It says almost nothing about whether anyone changed how they run a one-on-one, handle conflict or make decisions under pressure.
The gap is well known. Donald Kirkpatrick’s four-level model, first published in the 1950s, separates reaction (did they like it?) from learning, behavior and results. Decades of evaluation practice since have shown that reaction scores correlate weakly with the levels that matter. People can love a program that changes nothing, and resent one that changes a great deal.
Six months is the right checkpoint for a simple reason. It is long enough for the post-program glow to fade and for new habits to either stick or collapse under normal workload. It is short enough that you can still reasonably connect what you see back to the program. This article sets out how to build that six-month view, starting before the program begins.
Measurement starts before day one
You cannot show change at six months without a clear picture of month zero. Most failed evaluations fail here, not at the end.
Start with the business problem the program is meant to solve. “Develop our leaders” is not a problem. “First-line managers in operations lose a quarter of new hires within a year” is. So is “Cross-functional projects stall because senior managers escalate rather than resolve disagreements.” A named problem tells you what to measure.
Then translate the problem into a short list of observable behaviors. If the issue is new-hire attrition, the behaviors might be holding weekly one-on-ones, giving specific feedback within days rather than at review time, and having a career conversation in the first 90 days. Keep it to three to five behaviors. A program that targets fifteen targets none.
Finally, capture a baseline for each behavior and each outcome before the program begins:
- A pre-program 360 or pulse survey on the specific behaviors, from direct reports and peers.
- Current team metrics: engagement scores, regretted attrition, absenteeism, internal mobility, and any operational numbers tied to the problem.
- A comparison group where possible: managers who will attend a later cohort, or similar teams not yet enrolled.
Agree with the program sponsor, in writing, what success at six months would look like. That conversation is uncomfortable, and it is also the single most useful thing you can do for the evaluation.
Behavior: are they leading differently?
Behavior change is the hinge of the whole evaluation. If behavior has not changed, no business result can be credited to the program. If it has, you have a plausible path to results.
The strongest evidence comes from the people on the receiving end. Ask direct reports and peers, not participants, whether specific behaviors have changed. Self-ratings drift upward after training because people feel more capable, even when others notice no difference.
Use a targeted mini-360. Rerun the same short survey you used at baseline, limited to the three to five target behaviors. Frequency questions work better than quality judgments: “In the past month, how often did your manager ask for your view before making a decision affecting your work?” is easier to answer honestly than “How inclusive is your manager?”
Ask about change directly. A useful companion item is: “Compared with six months ago, how has this person’s approach to [behavior] changed?” with a scale from “much less effective” to “much more effective.” It captures perceived change even where absolute scores were already high.
Look for behavioral traces. Many leadership behaviors leave records. Calendars show whether one-on-ones happen. Performance systems show whether feedback is logged during the year or only at review. Meeting notes show whether decisions are delegated. These traces are imperfect, but they are not subject to rater goodwill.
Check for decay, not just change. If you can, pulse at three months and six months. A behavior that rose at three months and fell back by six tells you the program created intent but the environment did not sustain it. That is a different problem from a program that never shifted behavior at all.
Team signals: is anyone better off?
Leadership is measured in the people being led. Six months is often enough to see early movement in team-level indicators, even if hard business results take longer.
The most useful signals are those that managers influence directly:
- Engagement items tied to the manager. Pull the specific questions about recognition, growth, clarity of expectations and trust in one’s manager, rather than the overall engagement index, which moves slowly and reflects many forces.
- Regretted attrition and transfer requests. People leave managers. A drop in voluntary exits or internal transfer requests from participants’ teams is meaningful, especially against a comparison group.
- Absenteeism and sick days. These often track team climate and can shift within months.
- Internal mobility and promotions from the team. Leaders who develop people export talent. Rising promotions out of a participant’s team is a positive sign, even though it looks like loss to the manager.
- Psychological safety indicators. Short validated scales, such as the items developed by Amy Edmondson, can be added to a pulse survey and compared to baseline.
Read these numbers with care. Small teams produce noisy data: one resignation on a team of six swings attrition dramatically. Aggregate across the cohort, look at direction rather than precise values, and always compare against teams whose managers did not attend.
Business results: separating the program from everything else
The hardest question is not whether results improved but whether the program caused the improvement. In six months a reorganization, a new product, a hiring freeze or a market shift can swamp any training effect.
Three techniques make the attribution case credible.
Comparison groups. The cleanest design compares participants with similar managers who have not yet attended. Staggered cohorts make this easy: the people waiting for cohort two are a natural control for cohort one. Compare the change in each group from baseline, not just end-point values.
Trend analysis. Where no comparison group exists, compare post-program results with the trend that was already in place. If attrition was falling 2% a quarter before the program and keeps falling at the same rate, the program has not shown an effect.
Participant and stakeholder estimates. Jack Phillips’s ROI Methodology adds a structured fallback. Participants, and ideally their managers, estimate what share of an improvement came from the program and how confident they are in that estimate. Multiplying the two gives a deliberately conservative figure. It is not proof, but it is far more defensible than assuming the program caused everything.
If you do convert results into money, be conservative at every step: include full program costs, including participants’ time away from work; count only benefits within the first year; and report the assumptions alongside the figure. A modest, well-supported return persuades executives more than a dramatic one nobody believes.
Stories with evidence: the Success Case Method
Numbers show whether something changed. Interviews show how, for whom and why. Robert Brinkerhoff’s Success Case Method is built for exactly this six-month question.
The method works in two steps. First, send a brief survey to all participants asking how much they have applied the program and with what result. Second, identify the most successful and least successful cases and interview a handful of each in depth.
The success interviews document what people actually did differently and what came of it, with enough specifics to verify. “I started weekly skip-level meetings and caught a safety issue two months before it became an incident” can be checked. “I feel more confident” cannot.
The non-success interviews are often more valuable. They reveal why the program did not take: a manager who discouraged the new approach, workloads that left no room to practice, or content that did not fit the participant’s role. These are the findings that improve the next cohort.
Good interview questions at six months include:
- What is one thing you do now as a leader that you did not do before the program?
- Can you describe a specific situation where you used it, and what happened?
- Who else would notice this change, and would they agree?
- What got in the way of applying what you learned?
- If the program had never happened, what would be different today?
Report the proportion of participants in each group alongside the stories. Saying “roughly a fifth of participants show strong, verified change, and the main barrier for the rest is X” is a more honest and more useful finding than an average.
Measure the environment, not just the participants
When a program appears to fail, the cause is often the workplace it sent people back into. Research on training transfer has repeatedly found that what happens after a program, especially the participant’s own manager’s support, shapes whether learning is applied.
So measure the transfer climate alongside outcomes. Ask participants at three and six months:
- Did your manager discuss the program with you before and after it?
- Have you had the opportunity to use what you learned in real situations?
- Are the new behaviors recognized, ignored or discouraged?
- Do company systems, such as performance reviews, promotion criteria and meeting norms, reward the behaviors the program taught?
These answers turn a vague verdict into a diagnosis. A program with strong learning scores, weak behavior change and poor climate scores does not need redesigning. The organization around it does. That is a finding worth taking to the executive team, because it is a problem training alone cannot solve.
A practical six-month measurement plan
A credible evaluation does not need a research budget. It needs a few measures, chosen early and repeated on schedule.
| When | What to measure | Source | What it tells you |
| Before the program | Target behaviors, team metrics, business baseline | Mini-360, HR and operational data | Where everyone started |
| End of program | Reaction and learning, plus a written action plan | Participant survey, knowledge check | Whether the content landed |
| Month 1 | Manager support conversation held | Short pulse to participants | Whether transfer has a chance |
| Month 3 | Early behavior change, transfer climate | Pulse to reports and participants | Whether habits are forming |
| Month 6 | Behavior change vs baseline and comparison group | Repeat mini-360, behavioral traces | Whether leading has changed |
| Month 6 | Team indicators vs baseline and comparison group | Engagement items, attrition, mobility | Whether teams are better off |
| Month 6 | Success and non-success cases | Brief survey plus 8–12 interviews | How change happened and what blocked it |
| Month 6–12 | Business outcomes tied to the original problem | Operational data, trend or control comparison | Whether the problem moved |
A few principles keep the plan honest. Measure the same things the same way at each point. Keep surveys short enough that people answer them. Report what did not work as clearly as what did. And share the results with participants, which itself reinforces the behaviors you are hoping to see.
The question worth asking
The satisfaction survey asks, “Did you enjoy it?” The six-month evaluation asks, “What is different now, who noticed, and what did it change?” Those questions are harder to answer, and they are the only ones that justify the investment.
Organizations that answer them well tend to share three habits. They define the problem before they buy the program. They measure behavior through the eyes of the people being led. And they treat a disappointing result as information about the system, not just the course.
Do that, and the next budget conversation changes. Instead of defending a satisfaction score, you can show which behaviors moved, which teams benefited, what got in the way, and what you will do differently next time.
