Forty-six. That's how many novel drugs the FDA's Center for Drug Evaluation and Research approved in 2025, after 50 the year before. Globally, IQVIA estimates that 73 novel active substances were launched in 2025, above the average of the past decade.
Those are two different counts, and neither is your launch plan. The US figure is approvals by one regulator, and on its own a dip from 50 to 46 says nothing about rising volume. The global figure counts new molecules reaching any market. What a given company's field force faces is something else again: its own new molecules, plus every new indication, label update and country roll-out, which neither number captures.
Now put that next to a second number: around $200 billion of revenue that IQVIA's Launch Excellence IX puts at risk from loss of exclusivity by 2030. IQVIA's conclusion is that companies need an unprecedented number of launches, each with consistently strong performance. Our argument is that the way most companies prepare their field forces was built for a slower rhythm, one big launch every few years with a kickoff, a set of modules and a lot of goodwill. That model doesn't scale. Readiness has to become a repeatable system, run the same way for every launch, indication and congress, and measured the same way too.
The pipeline is full, and the clock is running
It's worth looking at the shape of recent approvals, not just the count. Of the 46 novel drugs CDER approved in 2025, 23 had orphan designation, and 85% were approved on their first review cycle.
Each of those facts has a consequence for commercial and medical teams.
Orphan designation means small patient populations, small specialist audiences and, usually, small field teams with very deep knowledge requirements. Oncology means the same molecule often gains new indications, new lines of therapy and new combinations over a few years, each one a mini-launch in its own right. Expedited pathways and first-cycle approvals mean the runway between approval and launch can be shorter, and sometimes less predictable, than the launch plan assumed.
IQVIA's Early Bird review adds context that anyone in the field will recognise: healthcare systems under capacity strain and changing patterns of stakeholder engagement. Specialists are busier, access is harder and the conversations that do happen are shorter.
And the stakes of getting it right haven't changed. McKinsey's analysis of 210 launches between 2003 and 2009, published in 2014, found that about two thirds of new drugs missed pre-launch consensus sales expectations in their first year, and that 78% of year-one underperformers were still underperforming in year two. A slow start is hard to recover from. Multiply that risk by the number of launches, indications and updates in a typical company portfolio over the next five years, and readiness stops being a training question and becomes a portfolio question.
A launch every few years can survive on heroics. A launch every few months needs a system.
What that means for the people in the field
From the rep's or MSL's point of view, all of this lands as one experience: more to learn, more often, with less time between the moments that matter.
More to learn, more often
A field team today may face, within a single year, a new molecule, a new indication for an existing product, an updated treatment guideline, a competitor's new data at a major congress and a label change. Each one shifts what a specialist expects the rep to know. The knowledge doesn't arrive in one annual block; it arrives in waves, and the waves overlap.
Smaller, more specialised audiences
In rare disease, a team might be a handful of people covering a whole country, each speaking with a short list of specialists who know the condition better than almost anyone. There's no room for a weak rep in that setting. Every gap in every person matters, because every specialist relationship matters.
Global from day one
Launches increasingly roll out across many countries and languages close together, rather than market by market over several years. That means translated, locally adapted and approved materials in many languages at once, and a need to see readiness across countries on a single view rather than in a dozen separate reports.
The old model was built for a different rhythm
Most launch readiness processes we've seen follow the same pattern. Months before launch, an agency is briefed to build modules. The content goes through MLR, often slowly, because it's new and long. A kickoff is planned. Reps complete e-learning, attend the event, pass an end-of-module quiz and go to the field. Completion is reported. Then the process starts again, largely from scratch, for the next launch.
There's nothing wrong with any single step. The problem is that the whole thing is a project, rebuilt for each launch, with its own vendors, formats and metrics. It works tolerably when launches are rare. It breaks when they're frequent, for three reasons.
First, it's slow. Building long modules and reviewing them takes weeks or months, which is exactly the time a shortened runway no longer allows.
Second, it's front-loaded. Almost all the effort goes into the period before launch, and very little into the weeks after, which is when knowledge fades and field behaviour settles.
Third, it doesn't learn. Because each launch is measured on completion, with its own tools, there's no way to compare readiness between launches, teams or countries, and no way to know whether the process is getting better.
What a repeatable readiness system looks like
When we say "system", we mean something concrete: the same small set of mechanics, run the same way for every launch, indication, congress and cycle meeting, with the same metric at the end. In our experience it has four parts.
- Content from approved documents, in days. AI can draft questions, short content and translations from the approved label, publications and brand materials, as long as it works only from those documents, experts sign everything off and conflicts between sources are flagged for a human. In our experience that shortens production considerably, in every language the launch needs, and gives MLR a consistent format to review. Measure it in your own portfolio rather than taking our word for it.
- Short measurement waves. Instead of one big assessment, run a short, synchronised wave for each business moment: the launch, a new indication, a congress readout, a cycle meeting. Five questions in sixty seconds, repeated over weeks, show what each person knows on the topics that matter.
- Individual gap fixing. After each wave, every person receives the fix for their own mistakes, then gets re-tested on those concepts until they've answered them correctly enough times to count. No trainer needs to be in the room, and nobody wastes time on what they already know.
- Always-on assistant. Between waves, the field has questions. An assistant that answers from approved content only covers those moments, and every question it receives shows where the next wave should focus.
None of these is exotic on its own. What makes it a system is running them together, repeatedly, with a shared data model. The questions from one launch become the base for the next indication. The gaps still open at the end of one wave become the brief for the next.
A fair objection is that every launch is different, and a standard system will produce standard, generic readiness. We'd put it the other way round. What should be standard is the mechanics and the measurement; what should be specific is the content, the critical facts and the moments. Standardising the "how" is precisely what frees the team to spend its time on the "what", which is where launches really differ.
Consistency is what scale looks like
Two programs from our case library show what this looks like in practice.
One MedTech field force in Asia Pacific, in the vascular space, has run seven congress challenges over 26 months, each in a three-day window. Across roughly 190,000 answers they identified 24,183 knowledge gaps and closed 49%. The striking part isn't the headline figure. It's that closure landed between 46% and 52% in each of the five main content areas. The same mechanics, run the same way, gave results within a narrow band, which makes planning easier, though each wave had its own questions, so the band shows consistency of process rather than identical difficulty.
A global immunology kickoff shows the other side of scale: breadth. The program ran four franchise games at the same time for 1,330 players, with 404 questions over six days and about 301,000 answers. It closed 53% of 15,238 gaps, with franchises ranging from 47 to 58%. Four parallel launch-level readiness efforts, in one window, on one data model.
Across the whole platform, more than 275 million answers have now been recorded, in more than 500 organisations and more than 50 languages. One number from that data is worth holding onto as launch volume rises: in programs with gap fixing, each rep accumulates about 16 critical gaps a year on average. That's the steady flow of things they need to know and don't yet, across every launch, update and congress. It's not a one-off problem to be solved at a kickoff. It's a recurring one that needs a recurring answer.
Measure the system, not just the launch
If readiness is going to be a system, it needs one metric that means the same thing everywhere. Completion won't do it. We'd argue for the share of knowledge gaps verified closed: a gap being one person wrong on one question at first attempt, and closed only when that same person later answers it correctly.
Measured that way across our case library of 100 programs, closure ranges from 18% to 80%; across the full dataset, about 43% of identified gaps were verified closed. That spread is useful, because it shows how much design matters. One pattern is consistent: the 16 programs tied to a clear business moment (a launch, kickoff, certification, congress or cycle meeting) closed 55% of gaps, against 41% for the other 84. That's an observed association, not proof that the business moment caused the difference, but it's consistent enough that we plan around it.
A shared metric also lets leadership ask questions that are impossible today. Which launch prepared its field best? Which countries consistently close more gaps, and what do they do differently? Is our readiness improving from one launch to the next? A consistent system makes those questions answerable, with one caution. The same metric name doesn't make two numbers comparable by itself. Two launches with different question banks, different difficulty and different exposure (how many rounds, how many weeks) can produce closure rates that look alike and mean different things. Compare like with like: shared core questions, similar windows, and a note of how many times people were retested.
The next launch is the one to build the system on
The tempting response to rising launch volume is to add capacity: more agency briefs, more modules, a bigger kickoff. It helps for a while, and then the next launch arrives before the last one has settled.
The better response is to decide, at the next launch planning meeting, that this launch will be run as the first instance of a system rather than as a one-off project. Pick the mechanics. Agree the metric. Plan the waves around the business moments. And commit to reusing everything, from the question bank to the review process, for the indication or launch that follows.
That's the approach we take at Atrivity: AI-assisted content from approved documents, short competitive waves, Gap Fixing for each person, Atrivity Assistant between waves and readiness data by team, country and topic that stays comparable from one launch to the next. The launches will keep coming. The question is whether each one starts from zero, or from everything the last one taught you.