Rolling out a selling model is the easy part. You write the playbook, record the videos, run the workshops, and a few months later everyone in the organisation can name the steps. Whether they can use them in front of a pharmacist or a GP on a Tuesday afternoon is a different question, and one that no quiz alone can answer. This case is from a program that ran in February 2021, and it measured the step before that: whether people knew the model well enough to apply it.
A global consumer health company had just rolled out an insight selling model to its commercial teams worldwide. It wanted to know how well 1,531 people knew the model's principles: how the method reads an HCP persona, what makes a question courageous, what an action-based close asks for. Six days and 416,000 answers later, the answer was "mostly, but not where you would expect", and the weak spots were not the ones leadership had worried about.
One boundary before the numbers. A right answer in this program shows that a person agrees with the taught model: that, according to the method, a given statement signals a given persona. It does not show that a real HCP belongs to that persona, or that the rep can diagnose one in a live call. Observed skill needs role play or call observation. What a knowledge game can do is show where the model itself has not landed, which is the precondition.
The situation: a new selling method, a global field force, and no way to see adoption
The company had invested in a single, global way of selling. The model combined several well-known building blocks: a structured call opening and feature-to-benefit framing, reading HCP personas (the familiar analytical, amiable, directive and expressive styles), courageous questions that bring a disruptive insight into the conversation, pre-call and post-call planning with SMART objectives, an advocate ladder for moving HCPs from awareness to recommendation, and a closing technique built around asking for a specific action.
Training had reached everyone. Completion was not the issue. The issue was that nobody could see, across dozens of markets and several roles, which parts of the method had been absorbed and which had merely been heard. Sales reps, key account managers and marketing people all took part, in English, in the same cohort. That matters for reading the results: for many players English was not their first language, and a marketing manager and a field rep use the method in very different ways, so differences between groups may reflect language and role as much as understanding.
The stakes were real even without a product launch attached. A selling model only pays back if it changes what happens in calls. If a large share of the field misreads HCP personas, every later step of the method is built on the wrong foundation: the wrong questions, the wrong framing, the wrong close.
What we designed, and why method content behaves differently from product content
We have run many programmes on product knowledge. Method is harder. A product question has a fact behind it: an endpoint, a dose, a contraindication. A method question asks for judgement: given this statement from an HCP, which persona are you probably facing? Given this situation, what kind of close fits? The wrong answers are rarely random. They are plausible, and people pick them for reasons.
So the design choices followed from that:
- One global game, with all 1,531 players in the same battle window, so every market played the same questions on the same days and results were directly comparable.
- 111 questions, all on the selling method itself and not on any product, covering every block of the model.
- A six-day window, long enough for people in every time zone to play many times, short enough to keep the competitive energy of an event.
- One-minute challenges against real colleagues, with notifications to pull people back in.
- A strict definition of success: a gap is one person wrong on one question at first attempt, and it only counts as closed when that same person answers it correctly later.
That last point was the one we spent most time explaining to the sponsors. It is tempting to report the final score and call it readiness. We wanted to see who changed their mind and who did not, because in a method programme the people who stay wrong are telling you something about the method, or the way it was taught.
What happened: 43% of gaps closed, and personas were the stubborn part
Participation was intense. Players averaged 272 answers each in six days, for a total of about 416,000 answers.
Across the programme we identified 29,613 knowledge gaps at first attempt. Of these, 12,608 were verified closed on a later attempt, which is 43%. The remaining 17,005 were flagged for follow-up coaching.
The breakdown by block is where it gets interesting. Closure rates, verified on later attempts:
- Call structure and feature-to-benefit framing: 7,699 gaps, 36% closed.
- HCP personas and handling reactions: 7,216 gaps, 42% closed.
- Courageous questions and disruptive insight: 7,219 gaps, 44% closed.
- Planning, follow-up and the advocate ladder: 5,523 gaps, 49% closed.
- Other content blocks: 1,269 gaps, 49% closed.
- Action-based close: 687 gaps, 51% closed.
Two things stand out. First, the foundational structure of the call, the part leadership assumed everyone had mastered, produced the most gaps and the lowest closure. Second, the close, which many sponsors had expected to be the hardest part, had the fewest gaps and closed best.
Three questions that tell the story
Aggregates hide mechanisms, so we looked at individual questions. Three of them sum up the programme, and they fall roughly into the thirds we tend to see: one closed, one half closed, one still open.
The close: fixed by repetition. One question asked how an action-based close should relate to what the HCP would normally do. The method says it should be different: you ask for a change, not for more of the same. At first attempt, 381 people chose an option that tied the close to existing behaviour, 100 picked "similar" and a handful chose other options. Of the 492 people who were wrong at first, 315 got it right later, about 64%. This is a misunderstanding that one clear feedback moment corrects.
A cost and safety persona: half closed. Another question gave an HCP statement centred on cost, compliance and product safety and asked which persona it signalled. The correct answer was amiable. At first attempt 428 people picked directive and 178 expressive, with a few spread elsewhere. Of 619 people wrong at first, 276 corrected themselves, about 45%, and 343 were still wrong at the end. People could tell the statement was about practical concerns, but they mapped "cost" onto a results-driven style.
An analytical persona: still open. The hardest case was a statement about compliance, effectiveness and convenience, which the method associates with an analytical HCP. Most people got it right at first attempt. But of the 459 who did not, only 84 corrected themselves, roughly 18%, and 375 were still wrong at the end. At the end, 245 people still mapped that analytical statement to amiable. If the same reading happened in a real call, it would send the conversation down the wrong route: warm rapport-building for someone who wants data.
When fewer than one in five wrong answers corrects itself with repetition, the problem is probably not memory. It may be a belief, or it may be the material, and either way it needs a conversation.
Five lessons we took from it
1. The basics are not basic
Everyone assumed the call structure was the easy part. It was the block with the most gaps and the lowest closure. Foundational steps get taught once, early, and then rarely revisited, because trainers move on to the more exciting parts of the method. If you are rolling out a selling model, test the foundations as hard as the advanced material.
2. Judgement gaps concentrate, and that is useful
In product programmes, gaps are often spread across many facts. In this method programme they clustered around the steps that require judgement, especially persona reading. That is good news for managers: it tells them precisely where to spend coaching time on ride-alongs, rather than re-running the whole workshop.
3. Some gaps need further assessment and coaching
The close question corrected itself, which suggests people had simply got it wrong once. The analytical persona question did not, which is consistent with people holding a mental model that differs from the method's, though it doesn't prove it: some of those people may never have met the question again. On the other side, for those who did correct it, one later right answer does not rule out a lucky guess. The number to watch is not the final score but the share of wrong answers that correct themselves. When it is very low, stop sending more quiz questions and start a coaching conversation.
4. Personas are partly a teaching problem
A fair objection is that persona models are fuzzy by nature, and an HCP who talks about cost might really be directive. That is true in the real world. But inside a method, the mapping should be consistent, and when hundreds of people make the same "wrong" mapping, it is worth asking whether the training material itself is ambiguous. We shared the persona results with the method owners for exactly that reason. Not every persona error is a belief problem. If the playbook describes cost concerns under two personas, or the question uses a word the playbook never uses, the construct is unclear and the fix is in the material. If the playbook is clear and people still map it the same wrong way after feedback, that is a learner misunderstanding and the fix is coaching.
5. A single global window makes differences visible
Because every market played the same questions in the same six days, the comparison between teams was fairer than most. Nobody could argue that their market had easier questions or a different time frame. It was not perfectly fair: English-only content and different roles still weigh on the numbers. Read the differences as places to look, not as rankings.
What we would do next
If we ran this programme again, three things would change.
First, we would add a short baseline before the rollout workshops, not only after. It would show whether the persona confusion existed before the training or was introduced by it, which matters a great deal for whoever maintains the playbook.
Second, the 17,005 open gaps should feed a structured follow-up: targeted fixes for each person's own mistakes over the following weeks, with critical gaps assessed against the program's agreed repeat-answer criterion, and a manager guide for the persona misconceptions that repetition barely moved.
Third, we would connect the method to the products. A selling model tested in the abstract is a start. The real test is whether a rep applies the persona reading and the action-based close when talking about a specific brand. A second wave with method questions embedded in product scenarios would show whether the knowledge transfers to scenarios; observed role plays would show whether it transfers to behaviour.
Why this matters beyond consumer health
Pharma and MedTech companies roll out selling models, customer engagement frameworks and KAM methodologies all the time. They are usually measured by workshop attendance and a certification quiz. This programme showed that knowledge of a method can be measured with the same rigour as product knowledge, and that doing so tells you where the model has not landed. Whether it is applied in calls is a second question, for role play and observation.
It also showed that the most valuable output is not the average. It is the short list of questions where hundreds of people are wrong in the same way and do not change their mind. That list is a coaching agenda, a feedback loop for the method owners, and an early warning of where calls in the field may be going down the wrong route.
At Atrivity, this is what we build for commercial teams: short challenges against colleagues, Gap Fixing for each person's own mistakes, and readiness views that show where to coach. If you are rolling out a selling model this year, it is worth asking how you will know, three weeks later, who understands the model and who needs observed practice.