Every Medical Affairs organization can point to the same evidence of Medical Science Liaison (MSL) training success: certification scores, completion rates, and therapeutic-area assessments passed on the first attempt. By those measures, most MSL teams are exceptionally well trained. And yet the field tells a different story of scientific exchanges that stall at data recitation, key opinion leaders (KOLs) who remain politely unmoved, and insights that never quite materialize. The issue arises when knowledge meets a live, skeptical, expert audience.
KOLs are the most evidence-saturated professionals in healthcare. They have read the pivotal trial. They may have enrolled patients in it. Presenting them with more data, such as another slide, another forest plot, or another subgroup analysis, is rarely what changes their thinking, because information scarcity is not a constraint.
The research on physician behavior change is unambiguous on this point. A landmark systematic review of randomized controlled trials published in JAMA found that didactic sessions, the one-way presentation of information, do not appear to be effective in changing physician performance, while interactive sessions that enhance participant activity and provide the opportunity to practice skills can effect change in professional practice.[1] If lectures don’t change physician behavior, an MSL training model built on data delivery shouldn’t be expected to either.
The stakes of getting this right extend well beyond any single product. Research published in the Journal of the Royal Society of Medicine estimates a 17-year lag between the generation of clinical evidence and its adoption into practice.[2] The MSL exists, in large part, to compress that lag. An MSL who can only transmit data takes part in the delay. An MSL who can engage a KOL in genuine scientific dialogue, contextualize evidence against the KOL’s own clinical experience, and engage skepticism rather than deflecting it, accelerates adoption of legitimate science.
Here is the assumption embedded in most MSL training: if we build enough scientific knowledge, communication competence will follow. The expertise literature says otherwise.
Research on expert performance demonstrates only a weak relationship between traditional indicators of expertise, such as experience, credentials, or perceived mastery of knowledge, and actual observed performance. What separates high performers is deliberate practice: training focused on specific tasks, with immediate feedback, structured problem-solving, and repeated opportunities to refine behavior.[3]
Consider what an MSL does in a high-stakes exchange. They read the room. They sequence evidence in response to what the KOL has just said, not in the order the slide deck dictates. They absorb a methodological challenge they didn’t anticipate, concede what deserves concession, and hold the line where the evidence holds. They recognize the moment a KOL’s question reveals an insight the organization couldn’t obtain any other way and they capture it. None of this is knowledge. All of it is skill. And skill, the research is clear, is built through structured rehearsal under realistic conditions with expert feedback, not through modules, and not through certification exams.
This has a direct design implication: MSL development must be built around the conversation itself. Scenario-based simulation against a genuinely skeptical counterpart. Practice navigating the specific objections the therapeutic area will generate. Rehearsal of evidence framing within regulatory boundaries, because an MSL who improvises under pressure without that fluency creates compliance risk, not scientific credibility. Training that stops at “what the data shows” prepares MSLs for an exam. Training built on deliberate practice prepares them for the field.
The second assumption worth abandoning is that training completion is evidence of training effectiveness. The industry’s own data is sobering. Research from ATD (then ASTD) found that while 92 percent of organizations evaluate learner reaction, use of evaluation drops off dramatically at each subsequent level, with very few organizations ever measuring behavior change or business results (ATD).[4]
The Kirkpatrick Model provides a useful framework for understanding this gap. Level 1 measures participants’ reaction to the training, Level 2 assesses what they learned, Level 3 examines whether they apply that learning on the job, and Level 4 evaluates whether the training contributes to meaningful organizational results. In Kirkpatrick’s terms, we measure Level 1 and Level 2 for reaction and knowledge and rarely Level 3 and Level 4, where field behavior and organizational impact live. The data that matters most is the data least often collected.
For Medical Affairs, this is a strategic vulnerability. The function is under sustained pressure to demonstrate its value in terms leadership recognizes, such as insight generation, launch readiness, and the quality of its scientific relationships. A Medical Affairs leader who reports 100 percent training completion is reporting an activity metric. A leader who reports a measurable improvement in the depth of scientific exchange, the volume and quality of field insights, or KOL engagement over time is reporting an outcome. Only one of those earns the function credibility.
The practical shift: define what a successful scientific exchange looks like before designing the training, then measure against it.
Observed performance in simulation
Structured field coaching assessments
Insight quality and actionability, not insight counts
KOL relationship depth over time
These are harder metrics to build than a completion dashboard, which is precisely why they are worth more.
All of this points to an uncomfortable conclusion about how MSL training gets sourced. A content vendor can build scientifically accurate modules. What they typically cannot do is design deliberate practice at the level of genuine scientific exchange because that requires publication-depth fluency in the science, command of the regulatory boundaries that govern the conversation, learning design grounded in how experts actually build performance skill, and a shared commitment to measuring field outcomes rather than deliverables shipped.
This type of training becomes a strategic decision with much broader implications for the organization. Medical Affairs leaders who treat MSL development as a content purchase will get content. Those who engage a strategic learning and development partner, one accountable for what MSLs can do, not what they’ve completed, are building something different: a field force capable of the conversations that move the most demanding audience in medicine.
Your MSLs already know the data. The question worth asking is whether your training, and your metrics, were ever designed for anything more.
What makes MSL training effective?
Effective MSL training builds scientific exchange skill, not just knowledge. Research shows interactive, practice-based learning changes professional behavior while didactic content does not. Programs built on scenario-based simulation, expert feedback, and repeated rehearsal prepare MSLs for live KOL conversations and are measured by field outcomes, not completions.
How do you measure MSL training outcomes?
Move beyond completion rates and certification scores to behavior and results: observed performance in simulation, structured field coaching assessments, the quality and actionability of field insights, and KOL relationship depth over time. Define what a successful scientific exchange looks like before designing the training, then measure against it.
What is scientific exchange?
Scientific exchange is the compliant, nonpromotional, two-way dialogue between MSLs and healthcare professionals about clinical evidence. It is designed to contextualize data against clinical experience, engage skepticism credibly, and generate field insights for the organization, all within the regulatory boundaries that govern Medical Affairs communication.
[1] Davis D, et al. Impact of formal continuing medical education. JAMA. 1999;282(9):867–874. https://pubmed.ncbi.nlm.nih.gov/10478694/
[2] Morris ZS, Wooding S, Grant J. The answer is 17 years, what is the question: Understanding time lags in health research. J R Soc Med. 2011;104(12):510–520. https://pmc.ncbi.nlm.nih.gov/articles/PMC3241518/
[3] Ericsson KA. Deliberate practice and expert performance in medicine. Acad Med. 2004;79(10 Suppl):S70–S81. https://doi.org/10.1097/00001888-200410001-00022
[4] ATD. New study shows training evaluation efforts need help. https://www.td.org/content/press-release/astd-new-study-shows-training-evaluation-efforts-need-help
