A medical affairs insight definition depends entirely on who you ask. So does "KPI", and so does "talking point". Three words, three functions, three different meanings — and measurement projects fail on the ambiguity more often than on the mathematics.
Cross-functional meetings in pharmaceutical companies have a particular failure mode. Everyone agrees, the actions are noted, and three weeks later it becomes clear that two teams committed to different things using identical words.
Medical Affairs is unusually exposed to this because it borrows vocabulary from clinical, commercial and regulatory functions simultaneously. Three words cause most of the damage. All three appear in the full glossary, but they are worth taking on their own, because these are the ones that break projects.
Insight
To a market researcher, an insight is a synthesised finding from structured research — a pattern across a sample, arrived at deliberately.
To a commercial team, an insight is often a competitive or market observation. Something learned about positioning, access or a rival's activity.
To a field medical team, an insight is something a single HCP said that the organisation did not know and can act on. An unmet need described from practice. An evidence gap. A barrier to guideline-concordant care.
These are not variations on a theme. They differ in sample size, in how they are collected, in what they justify, and in who is allowed to act on them. A market research insight might support a strategic decision; a single HCP's observation usually should not, and treating it as though it could is a governance problem as well as an analytical one.
The practical consequence appears when someone asks how many insights the field team generated last quarter. The number is meaningless without knowing which definition produced it — and if the answer is being compared against a market research programme's output, the comparison is nonsense.
There is a further distinction worth holding onto: a record that says "positive discussion, receptive to the data" is not an insight under any of the three definitions. It is an impression. The test is whether the organisation learned something it did not already know, which is also the basis for treating insight yield as a measure rather than a count.
KPI
Everywhere in business, a KPI is a key performance indicator — a metric tracked against a target to assess how something is going. That is what a finance director, a commercial lead and a board member all understand by the term. It is also what most teams mean by medical affairs KPIs in the conventional sense.
Inside some medical affairs teams, and inside several software products built for them, "KPI" has come to mean something entirely different: a planned scientific talking point. A specific piece of science intended for discussion with a particular HCP.
The two meanings are not related. One is a measure of performance; the other is an item of content. And because the term is used confidently in both senses, the confusion is rarely detected in the room — it surfaces weeks later, when a medical team reports that an MSL "covered eight KPIs" and a commercial colleague hears a performance claim.
The fix is to abandon the internal usage in any cross-functional setting. Planned scientific talking point is longer, unambiguous, and means the same thing to everybody. RocketMSL uses the longer form in all external material for exactly this reason, and this article is the only place the internal meaning is discussed at all.
Talking point
This one is less contested but more consequential, because the ambiguity is about provenance rather than definition.
Everyone agrees a talking point is a scientific point intended for discussion. What varies is where it came from — and that determines whether it is defensible.
A talking point derived from MLR-approved content, mapped to a communication topic, is a legitimate basis for both discussion and measurement. A talking point written by a brand team and passed to field medical is a promotional message wearing a scientific label, and using it as a measurement benchmark exposes the function to exactly the criticism it is structured to avoid.
The distinction rarely appears in the terminology. Both are called talking points. Both arrive in a slide. Only one survives a compliance review, and only one is safe to score against.
So the useful question, when anyone proposes measuring talking-point coverage, is not what the talking points are. It is who approved them, and against what.
Why this is worth fixing before a measurement project, not during one
None of this is pedantry, and none of it is expensive to fix. It costs one page of definitions agreed before the work starts.
The reason it matters is that measurement frameworks are built on shared terms. If "insight" means three things, an insight count means nothing. If "KPI" means two things, a KPI report will be misread by half its audience. If "talking point" hides a provenance question, the entire benchmark may be resting on material that was never approved for the purpose.
These failures do not announce themselves. They surface at the end, when the numbers are challenged and it turns out the definitions were never agreed. By then, the framework has to be rebuilt rather than corrected.
Write the definitions down first. Put them in the project charter, not the appendix. And where a term genuinely means different things to different teams, say so explicitly rather than picking one and hoping.
The full glossary is here — twenty-four terms, with the disputed ones marked.
