A mother sits across from a paediatrician and asks a simple question: should she be keeping fat out of her baby's diet? The paediatrician may never ask the company's nutrition advisor that question directly. But the advisor who visits that clinic next week had better know the answer, and the reasoning behind it, because what they say about complementary feeding ends up, one way or another, in that conversation. The program in this case ran twice, in May 2025 and May 2026.
That was the logic behind this program. An infant nutrition business in Central America asked its zone medical and nutrition teams to show, every year, that they had command of complementary feeding science before they spoke to health professionals. Not a competition to win, a certification to pass. Across two annual editions, 932 player places recorded 381,000 answers, and 73% of the 11,434 knowledge gaps we found at first attempt were verified closed within the assessment, with the same rounded rate in both editions. That consistency, more than the headline number, is what made the program useful, although, as we explain below, it proves less than it might seem.
In infant nutrition, the field has to be right before it is persuasive
Infant nutrition is an unusual commercial environment. The people the field team meets are paediatricians, nutritionists and other health professionals who advise families at a sensitive moment. The subject is science-heavy: when to introduce solid foods, in what order, how to read a baby's readiness signs, which micronutrients matter most, how infant cereals are made and prepared. The communication rules are strict, and rightly so. A company that gets the science wrong in front of a health professional loses more than a sale. It loses the right to be in the room.
The client's framing reflected this. The goal was not to rank people or reward the fastest. It was to verify command of the subject in the assessment, zone by zone, and to do it again twelve months later so that new joiners were covered and experienced advisors did not drift.
The audience was a Spanish-speaking team of medical reps and nutrition advisors spread across several countries. The content was deliberately narrow: one subject, examined deeply. We grouped the questions into the blocks the exam actually covered:
- complementary feeding practice (when and how to introduce foods)
- infant cereals (composition, preparation, range)
- childhood nutrition and cognition
- micronutrients (vitamins and minerals)
- general nutrition fundamentals
The client's own taxonomy used brand and occasion coding. For reporting, we simplified it into those five blocks so that a medical lead could read the results in terms of science, not product lines.
We designed a certification that people would take seriously, twice
A certification has a credibility problem. If it is too easy, people pass it and learn nothing. If it is a single sitting, it measures what someone crammed the night before. And if it happens once, it says nothing about next year.
We built each edition around a short, intense window of three to five days. Inside it, players worked through a bank of 286 questions in one-minute challenges (five questions in 60 seconds). Questions came back to the same player so that we could see whether an early mistake was corrected later.
The volume is worth pausing on. Players answered 409 questions each on average across the editions. That is not a light-touch exercise. The average was about 409 answers per participant place, including repeat attempts, against a bank of 286 questions, with feedback after every mistake, in less than a week. That cuts both ways. With so many repeats, some later correct answers will reflect memory of the specific item rather than understanding of the concept, and on a true or false question one later right answer does not rule out a lucky guess. We can't separate those in the data, so we read closure as a verified correction within the assessment, nothing more.
Three design choices mattered most.
First, the window was short on purpose. A certification that stays open for a month becomes background noise. Three to five days made it an occasion that managers could plan around and that people talked about.
Second, the definition of success was strict. A gap is one player wrong on one question at first attempt. It only counts as closed when that same player gets the same question right later. The idea was that certification should mean closing your own gaps, not just scoring well on the ones you already knew. The pass mark itself was the client's decision, and it is not part of the records we used for this case, so we report gap closure rather than pass rates.
Third, the program repeated a year later on the same subject. That turned a one-off exam into a way of tracking the team's knowledge over time, and gave newcomers a clear standard to reach.
A fair objection is that a certification built on multiple-choice questions cannot capture the nuance of counselling. A nutrition advisor needs to explain, not just select. We agree, and that is why the questions were written around counselling situations as much as facts: what would you recommend when a mother asks about fat, which food group would you moderate first, what does a particular reflex tell you. A good distractor in that setting is the answer a well-meaning but under-trained person would actually give. Choosing the right option among plausible ones shows that someone knows the right advice, and it is measurable. It does not show that they can explain it well to a health professional; that takes role play or observation.
A closure rate that repeats is reassuring. It tells you the process behaved the same way twice, not that the same people kept what they learned.
73% closed, and the same 73% the following year
Across both editions, we identified 11,434 knowledge gaps at first attempt and verified 8,388 of them as closed. That is 73%. The remaining 3,046 were flagged for follow-up coaching.
The closure rate was strikingly even across the five blocks the exam covered. These add up to 11,406; a further 28 gaps sat in smaller topics outside them.
- Complementary feeding practice: 3,815 gaps, 73% closed.
- Infant cereals: 3,348 gaps, 74% closed.
- Childhood nutrition and cognition: 2,747 gaps, 73% closed.
- Micronutrients: 1,086 gaps, 73% closed.
- General nutrition: 410 gaps, 77% closed.
And it held from one year to the next: 73% in 2025, 73% in 2026, with 473 players in the second edition and 459 in the first. The 932 counts each person once per edition, and the records we used do not say how many people took both, so the two editions are not a matched cohort. The same rounded rate tells us the design behaved consistently with two partly different groups. It does not prove that people who certified in 2025 still held the science in 2026, and it does not by itself validate the certification as a measure of competence.
Three questions show what the averages hide
Averages are comfortable. Individual questions are where the coaching agenda comes from. Here are three from the 2026 edition, described rather than quoted.
Reading readiness signs was the biggest single correction. One statement asked whether a specific infant reflex indicates readiness for solid food. It does not; if anything, it suggests the opposite. At first attempt, 267 of the 473 players called the statement true; counting everyone who did not choose the correct option, 272 started with a gap. That is more than half the team holding a belief that would lead them to encourage solids too early. By the end, 177 of those 272 had corrected it. Ninety-five still had not.
Which food group to moderate first still split the team. Asked which food group should be moderated first for digestive reasons, 77 people were wrong at first attempt, most of them choosing cereals. Repetition fixed 47, which is 61%. Thirty were still wrong at the end. It is a practical question, and one that families raise early.
The counselling answer on fat mostly landed. Asked what to recommend when a mother asks about fat in her baby's diet, 70 people initially chose to avoid fat entirely. The question assessed the recommendation in the client's approved nutrition material. Repetition fixed 43 of the 70, and 27 still got it wrong at close.
These three questions show the range of what a certification finds. The readiness reflex was a widespread misconception that responded well to repetition, perhaps because once people saw the correct explanation it made sense, although on a true or false item some of that correction may be guessing. The food group and fat questions were smaller but more stubborn. They look like habits of advice, probably absorbed from general parenting culture, and they are exactly the ones worth a conversation with a medical lead.
Four lessons from running the same certification twice
- Narrow and deep beats broad and shallow. One scientific subject, studied thoroughly, gave people a clear standard and, we think, a reason to come back until they met it. When a certification tries to cover everything, the results become hard to act on.
- Repeatability is a useful check, not a validation. The same closure rate two years running tells the medical team the process behaves consistently and lets them compare zones and years. To show retention, track the same people across editions; to validate the certification, check it against observed counselling.
- The most common error is not always the hardest to fix. The readiness reflex fooled more than half the team at first and still closed well. The smaller counselling errors were stickier. Don't rank gaps only by how many people had them; rank them by how many are still open at the end.
- Open gaps belong to the medical lead, not the trainer. The 3,046 gaps still open were concentrated in a handful of counselling questions. Those are best closed by a medical or nutrition lead explaining the reasoning in a team call, where people can ask why, soon after the window closes rather than at the next edition.
What we would do next
With two editions behind it, the program has a baseline. The natural next step is to use it.
We would run a short refresher round at the midpoint between editions, focused only on the questions that stayed open. Six months is a long gap, and a few one-minute challenges on the stubborn counselling questions would show whether the corrections from May had held.
We would also give the field a way to check the science between certifications. An assistant that provides answers based on approved scientific material, and that passes the questions it receives back to the medical team, would show which topics advisors are unsure about in real conversations, not just in the exam. That list would be the best possible input to the next edition.
And we would share the per-question results with the people who write the training content. When more than half a team holds the same wrong belief about a readiness sign, the question is not only how to fix it in the field. It is where that belief came from, and whether the existing material is accidentally reinforcing it.
A question for your next medical and commercial planning meeting
Many teams in nutrition and Medical Affairs certify their field once, at onboarding, and assume the knowledge holds. Ask yourselves a direct question: if we re-examined the team today on the science they talk about every week, what share of them would give the right answer to the first thing a parent asks?
If you do not know, that is the reason to run a certification. If you think you know, a second edition is how you find out whether you were right. At Atrivity we built this program with a nutrition team that wanted a verified standard, not a leaderboard, and the result was a standard they could hold themselves to year after year.