A cat owner walks into a veterinary clinic in Spain and mentions, almost in passing, that her two cats have started hissing at each other near the food. What happens next depends less on the product on the shelf than on what the person at the counter understands about cats. If they know why shared resources create tension, they can recommend the right environmental change and, where it fits, the right pheromone product. If they don't, the owner leaves with the wrong advice and comes back three months later with the same problem. The program we describe ran in 2022.
That was the premise of this program. An animal health company wanted to know whether veterinary clinic teams could read feline behaviour well enough to make a recommendation land. Over a four-day battle, 1,173 clinic staff answered 67,000 questions. We identified 12,339 knowledge gaps at first attempt and verified 19% of them as closed. That is a low closure rate by the standards of the programs we run, and it is the most instructive part of the story. It shows exactly what a short, single battle can do, and what it can't.
A recommended product lives or dies on behavioural knowledge
Prescription medicines have a clear decision-maker: the vet writes the prescription. Products that are recommended rather than prescribed work differently. The recommendation can come from the vet, a veterinary nurse, a receptionist or a clinic assistant, often in a short conversation at the counter, and it only works if the person making it understands the problem the owner is describing.
For a feline pheromone range, that problem is almost always behavioural. Stress in a multi-cat home, scratching, marking, tension around a move or a new arrival, a cat that suddenly stops playing. The product is part of the answer. The larger part is the advice around it: where to put the food bowls, how to play with a kitten, what low activity might signal.
So the client's question was not "do clinic teams know our product?". It was closer to "do clinic teams know cats well enough that our product makes sense to them, and to the owner?". That is a harder question, and a more honest one. It also meant the program had to be mostly about the animal, not the brand.
The audience added its own constraints. Clinic teams are not the client's employees. Nobody can make them attend a course. Their time is fragmented, between consultations, surgery, phone calls and the queue at the desk. Anything we asked of them had to be short, voluntary and worth their while professionally.
What we designed: one game, 211 questions, four days
We built a single competitive game in Spanish, open to veterinary clinic teams across Spain, with notifications to bring people back into the window.
- 211 questions in play, in one game.
- A four-day battle window, from a Thursday to a Sunday in March 2022.
- One-minute challenges against real colleagues from other clinics, with a live ranking.
- About four fifths of the questions on the cat itself, one fifth on the product range.
The content split was the key design decision. The biology block covered vision and pupil shape, whiskers, skeleton, gestation, hunting, grooming and the socialisation window in kittens. The behaviour block covered tension between cats, aggression in play and signs of stress. A pheromone science block explained how appeasing pheromones signal and where cats produce them. The product block was deliberately small and practical: which format suits which situation, and when to reach for each.
We chose a short battle rather than a longer program for reasons that made sense at the time. A four-day window creates intensity. It fits into a normal working week without asking clinics to commit to something open-ended. And for an audience that owes the brand nothing, a short, sharp competition was an easier ask than two months of reinforcement.
That choice turned out to shape the results more than anything else.
What happened: huge reach, a measurement, and only a small fix
Participation was strong for a voluntary, external audience. 1,173 players took part and generated about 67,000 answers, an average of 57 answers per player.
The platform's readiness index was high from the start and barely moved: 95% at first attempt, 96% at last. Treat that figure with care, because it raises a definition question. With 12,339 wrong first attempts inside about 67,000 answers, the index cannot be a simple share of all answers that were correct; on a shared denominator that would sit far lower. The index is calculated differently from raw answer accuracy, and the program records we are working from do not set out its formula, so we don't lean on it. That is a reason to explain the denominator before quoting any readiness score, not proof that the index is wrong. The gap counts below are defined plainly, and they are what we use.
At first attempt we recorded 12,339 knowledge gaps, each one a person wrong on one question. By the end of the four days, 2,382 had been verified as closed, meaning the same person had answered the same question correctly later. That is 19%. The remaining 9,957 gaps were flagged for follow-up.
By content area, closure was remarkably flat. The five areas account for 12,175 of the 12,339 gaps; the program records do not assign the remaining 164 to any of these areas:
- Feline biology and senses: 9,534 gaps, 19% closed.
- The product range: 1,757 gaps, 20% closed.
- Behaviour and stress: 427 gaps, 16% closed.
- Pheromone science: 404 gaps, 19% closed.
- Pet ownership context: 53 gaps, 2% closed.
When closure is that even across very different topics, the explanation is rarely the content. It is the structure. Fifty-seven answers per player in four days is enough to measure, but not enough for most questions to come back to the same person a second or third time. You can't verify that a gap closed if the question never returns.
It is worth being clear about what the 19% does not mean. It does not mean 81% of clinic staff stayed wrong about everything. Many of the open gaps belong to people who simply never met that question again before the window closed. They are unverified rather than failed. The problem is that, from a commercial point of view, an unverified gap and a stubborn one look the same until someone asks again.
Three misconceptions the counter conversation depends on
The aggregate numbers explain the design lesson. The individual questions explain why it matters for the brand.
Food bowls in a multi-cat home
Around 200 people, at first attempt, advised a single shared feeding bowl for several cats rather than one per cat. By the last attempt, 190 still did. Of 214 people who were wrong at first, only 9 corrected themselves through repetition. This is the single piece of advice that keeps multi-cat stress cases coming back, and it barely moved.
Low activity blamed on parasites
Asked about possible causes of low activity, 169 people pointed to fleas rather than a lack of stimulation in the cat's environment. That fell to 109 by the end, with 65 of 183 people correcting themselves. Better, but it still means a behavioural cause can walk out of the clinic treated as a parasite problem.
How to play with an aggressive kitten
188 people, at first attempt, recommended playing with hands and feet so the cat gets used to them, which is the habit that produces the biting owners later complain about. 152 still gave that answer at the end. Only 37 of 201 fixed it through repetition.
These three have something in common. They are not obscure facts. They are intuitive, plausible pieces of advice that people have probably given owners for years. That is exactly the profile of a misconception: confidently held, reinforced by habit, and resistant to a correction seen once on a phone screen.
In a recommendation channel the gap is rarely the product. It is the animal. Teams that can read the behaviour place the product without having to sell it.
Five lessons from a program that measured more than it fixed
1. A short battle is a diagnostic, not a cure
Four days produced a detailed map of who was wrong about what across more than a thousand people. That map is valuable in itself. But closure depends on the question coming back to the same person, spaced over days or weeks, and a four-day window doesn't allow much of that. If the goal is to change what people say at the counter, the battle should open a reinforcement period, not end the program.
2. A high index can hide the gaps that matter
A 95% readiness index looks like a job already done, especially when nobody has checked how it is calculated. Underneath it sat thousands of individual gaps, several of them on exactly the advice that decides whether a behavioural case is solved. Averages flatter. Question-level data doesn't.
3. In recommendation channels, teach the animal first
The decision to put four fifths of the questions on feline biology and behaviour was right. The product block had roughly the same closure rate as the rest, but the misconceptions with the biggest commercial consequence were behavioural. A clinic team that understands multi-cat tension will recommend a pheromone product for the right reason. One that doesn't will recommend it, or not, at random.
4. Plausible wrong answers are the most useful questions
The questions that revealed the most were the ones where the wrong option was common sense to a non-specialist: one bowl for everyone, fleas as the obvious cause, getting a kitten used to hands. Writing distractors like these takes more effort with the expert team, and it is where most of the insight comes from.
5. External audiences need a reason to come back
Clinic staff joined because the competition was short and fun. Keeping them engaged for longer would need a different offer: professional value, recognition for the clinic, content they can use with owners the same afternoon. That design work has to happen before launch, not after.
What we would do next
With the benefit of the data, we would keep the four-day battle as the opening event, because it created reach and energy that a slower start would not. Then we would follow it with several weeks of light reinforcement: short challenges that bring back each person's own wrong answers, spaced out so that closure can actually be verified.
We would send individual fixes for the critical questions, with a short explanation written for the counter conversation, not for a textbook. And we would take the three stubborn misconceptions to the client's field team, so that the reps who visit clinics could raise them in person, with an explanation and perhaps a simple owner leaflet. Repetition barely moved them. A conversation might.
Finally, we would plan a second battle a few months later with the same questions mixed in. The comparison between the two would show which gaps had truly closed and which had simply been forgotten again.
Nineteen per cent is not a headline number. It is an honest one, and it came with something more useful than a high score: a clear picture of what a clinic team believes about the advice it gives cat owners, and which pieces of that advice need more than a quiz to change. That is the kind of result we would rather give a client in Atrivity than a flattering average.