Ask a pharma commercial director for the market share of the lead brand in the northern region and you will have an answer before you finish the sentence. Ask what share of the field force in that region can correctly explain the primary endpoint of the trial behind it, and the room goes quiet.
That silence is the subject of this piece. Companies measure revenue, pipeline, call frequency, reach and cost per call down to the decimal. The thing that decides what happens inside each call, what the rep actually knows, is almost never measured at all. We think knowledge should be managed like any other commercial asset: counted, tracked over time, and reported to the people who make investment decisions. Here is why, what that looks like in practice, and where to begin.
The costliest gaps never show up on a report
Knowledge gaps do not appear in the P&L. Their effects do, usually months later and under someone else's name.
A rep who cannot explain the difference between the new product and the standard of care loses the argument with a specialist who has seen the competitor's congress data. A key account manager who misunderstands the access criteria sends a hospital pharmacist down the wrong route, and the formulary decision slips a quarter. An MSL who is unsure of the latest safety update hedges, and the KOL notices. A rep who confuses the indication with a neighbouring one creates an off-label risk that Compliance only hears about when it is already a problem.
In the sales data, each of these looks like something else: slower uptake, lower share of voice, a weak region, a stubborn account. Nobody writes "knowledge gap" in the monthly business review, because nobody can see it.
Think of it as an inventory nobody counts. A warehouse manager who could not say what was on the shelves would not keep the job for long. Yet most commercial organisations run launch after launch without a count of what their people know, and then act surprised when the shelves turn out to be empty in the one region that mattered.
The problem is not a lack of training. Most Life Sciences companies train a great deal: kickoffs, cycle meetings, launch meetings, certification, e-learning for every policy. The problem is a lack of visibility. We know who completed the course. We do not know who retained what mattered.
Training is booked as an expense. Knowledge behaves like an asset.
The way most organisations handle learning follows a simple production line. Create the content. Get it through medical-legal review. Deliver it. Track completion. Move on to the next launch.
That model treats knowledge as content, a thing you ship. But knowledge behaves differently from content. It shapes decisions, conversations and outcomes. It produces value when it is used. It loses value when it is not maintained, as anyone who has watched a field force drift away from the launch message by the second cycle meeting will recognise. And it can be counted, if you ask the right questions.
Those are the characteristics of an asset. Companies already manage financial assets, customer relationships and intellectual property with care. A patent portfolio gets a dedicated team. The knowledge that turns that patent into prescriptions usually gets a completion report.
A patent gets a dedicated team. The knowledge that turns it into prescriptions usually gets a completion report.
A fair objection: knowledge is not an asset in the accounting sense. You cannot put it on the balance sheet, it walks out of the door every evening, and it belongs to people, not the company. All true. We are not proposing to book it. We are proposing to manage it with the same discipline, which means knowing its current state, watching how it changes and deciding where to invest. You do not need an accountant's definition to do that. You need data.
Knowledge moves before sales. Treat it as a candidate leading indicator.
Commercial teams are very good at reading lagging indicators. Prescriptions, market share, sales against forecast. By the time those move, though, the cause is usually weeks or months old.
Consider the sequence at a launch. The field is trained at the launch meeting. Over the following weeks, part of it forgets or never fully understood a key message. Calls on that message become less convincing. Early prescribers are slower to try the product. Months later, uptake in some regions falls behind forecast, and the brand team starts a root cause analysis.
The timing matters more at launch than at any other moment. McKinsey's analysis of 210 launches between 2003 and 2009, published in 2014, found that about two thirds of new drugs miss pre-launch consensus sales expectations in their first year, and that of the underperformers, 78% were still behind in year two and 70% of those in year three. A launch that starts slowly rarely catches up. If the first sign of a knowledge problem is the sales data, you have lost the months when it could have been fixed.
Measured properly, knowledge can be the earliest signal you get. If you know by week four that one region is consistently getting the efficacy message wrong, you can correct it while the launch is still young, instead of waiting for a root cause analysis. That is what an early signal is for.
A caveat we take seriously: calling knowledge a leading indicator of sales is a hypothesis until you have validated it on your own data. Sales also move with access, pricing, competitor activity, supply and territory design, none of which a quiz measures. If you want to use knowledge as a performance predictor, track it alongside those factors for a few cycles and see whether it actually predicts anything. Until then, treat it as an early warning worth acting on, not a forecast.
What to measure instead of hours and completions
Most L&D reports still lead with activity: courses completed, hours delivered, attendance, certifications issued. These are fine as delivery indicators. They say nothing about the asset.
Knowledge metrics look different. The ones we find most useful:
- Gaps per person, per question. A gap is one person wrong on one question at first attempt. Counting gaps this way, instead of reporting an average score, shows exactly where the asset is thin.
- Gap closure. The share of those gaps for which the same person later answers the same question correctly. This is the knowledge return on the reinforcement you invest in. It is not a financial return.
- Coverage of critical knowledge. The proportion of the field that answers the handful of must-know questions correctly: the primary endpoint, the key safety message, the patient inside the label, the answer to the competitor's main objection.
- Readiness by team, country and topic. The same data, cut the way the commercial organisation is run, so a sales director can see which region needs attention.
- Change over time. Knowledge before and after an event, and again weeks later, to see what held.
A concrete example from the programs we have run. In 2021 a pharma field force in Latin America played a three-day battle on the principles of its virtual selling model. 725 reps generated 14,868 gaps at first attempt. By the end, 73% were verified closed, and readiness went from 72% on first attempts to 92% on last attempts. The 4,079 gaps still open went to managers with names attached. That is what knowledge looks like when it is measured like an asset: a starting position, a measured change and a list of outstanding items. A later correct answer is a verified correction within the assessment, not proof of better calls, and one later right answer does not rule out a lucky guess.
The same program showed something every asset manager knows: not every holding responds to the same treatment. On most questions, repetition closed the large majority of gaps. On one, about the effect of mood on a virtual call, barely more than half closed, because reps held a plausible but wrong belief. That gap needed a conversation with a manager, not another round of questions. Measurement tells you which is which.
How AI is changing the cost of measuring knowledge
Until recently, measuring knowledge continuously was expensive. Writing good questions took time. Getting them through review took longer. Translating and adapting them for every affiliate multiplied the effort. So most companies measured knowledge once a year, at certification, and assumed the rest.
Generative AI changes the cost side of that equation. Drafting questions and corrections from approved documents, in several languages, now takes a fraction of the time it did. That makes it realistic to run an ongoing loop: find gaps, send each person the fix for their own mistakes, check the gap closed, and flag the ones that need a manager.
The condition in a regulated industry is strict. The AI must work from approved sources only, and an expert has to sign off before anything reaches the field. A fast wrong answer about a contraindication is not efficiency. The value of AI here is that it makes the measurement loop affordable, not that it replaces the judgement of Medical or the first-line manager.
What changes when knowledge is visible
When a company can see its field's knowledge, several conversations change at once.
The launch team can see readiness by region before the launch date, not after the first sales figures. Sales managers can see which reps need coaching on which topic, so ride-alongs become targeted instead of general. Medical can see which parts of the scientific story the field consistently misunderstands, and fix the source material. Compliance can see where off-label risk is concentrated. L&D can show what reinforcement changed, in gaps closed, rather than in hours delivered.
If someone asks for a money figure, build it as a model, not a result. Write down the assumptions (how many calls touch the message, how often a gap plausibly changes the outcome of a call, the value at stake per account), give a low and a high case rather than one number, and name the commercial factors outside the field's knowledge that move the same sales line. A modelled range with its assumptions visible earns trust. A single euro figure presented as ROI does not.
And the executive team gets something it has never had: a regular, credible reading of the state of an asset it has been investing in for years without knowing what it owns.
One more objection deserves an answer. Will reps feel watched? Some will, at first, if the data is used to rank and punish. Programs that work treat the data as preparation, not surveillance: each person sees their own gaps and gets the fix, managers use it to coach, and the competitive element makes it closer to a game than an exam. In our experience the field's resistance fades quickly once people see that the questions are about what they genuinely need in their calls.
Start with one brand and one number
You do not need a company-wide knowledge strategy to begin. Pick the brand where the stakes are highest, usually the next launch or a product facing new competition. Agree the facts that matter. Measure them across the field, then report one number to the commercial leadership: the share of the field that can answer every critical question correctly, today.
That single number usually starts a better conversation than any completion report ever has.
At Atrivity, this is the work we do every day with pharma and MedTech field forces: short competitive challenges that measure what each person knows, gap-by-gap tracking that shows which gaps have a verified correction and which remain open, and readiness views the commercial team can act on. Whatever tools you use, bring one question to your next business review: what does our field actually know about our most important brand, and how would we find out?