ValeriiaVoshta

The problem

A camera-shop assistant knows the specs better than the customer, and that is exactly why the conversation fails. He answers a question about the sensor with numbers, while the person came in wanting good photos of her child. The course had to retrain something other than product knowledge – the translation from what the spec sheet says to what the camera is bought for.

The objective: the assistant matches a camera and a lens to what the customer asks for, and explains the choice in one sentence without retreating into specifications.

Why it works this way

The model behind it

Gagné's nine events, laid out in the order of the module: the shop scene gains attention, the buyer archetypes activate what the assistant already knows from his own shifts, the sensor and model screens present the content, the anchor lines serve as worked examples, and only then comes practice. The order of pages follows those events but is not locked: navigation is free, and the comparison table can be reached from any step. The order of pages stays a default route.

How it was tested

No grade is reported to the LMS: eleven pages, completion by progress. The single scored task is matching a camera to four requests, a 70 % threshold, unlimited attempts, feedback after every answer. It is practice rather than an exam: what matters is that the assistant lands on the right model in the end, not that he lands on it first time. What it measures, though, is recognition rather than transfer: the same four archetypes return and a model has to be picked from three. The question is no longer «what sensor does the 2000D have» but «what do you offer a mother with a forty-thousand budget» – and picking from ready options is still easier than saying it out loud. The transfer is tested on the next page, in the simulation, where the line has to be spoken to a live interlocutor. That is why the course does not end with a test. The link is stitched into the material itself: the archetype card for the young mother says outright that the assistant will meet a customer like her in the final simulation.

Mistakes cluster on the student and the blogger – the two requests where the right answer sits at the ends of the range: the cheapest body for one, the most expensive for the other. For the mother and the traveller the choice converges in the middle, so the habit of offering the popular model does no harm there and fails the answer here. What the task really catches is not missing knowledge but the reflex to sell whatever sells best.

The problem

Unpacking communication mistakes usually looks like a list of prohibitions and triggers a defensive reaction: the listener recognises his own phrase and shuts down. The module had to move the mistake out of the personal and into the shared – to make it an exhibit people look at with curiosity rather than guilt.

The objective: the employee judges a line from a real exchange – spotting what in it breaks the conversation and offering a replacement.

Why it works this way

The model behind it

The 5E cycle: Engage – the night hall and the promise that mistakes are preserved here rather than judged; Explore – walking the cases with no prompts; Explain – the analysis card, arriving after the visitor has formed an opinion; Elaborate – the greenhouse and the task of placing the exhibits; Evaluate – a check on the visitor's own formulations. The order is the point: a rule spoken before the walk would turn the museum into the same list of prohibitions.

How it was tested

The Evaluate step is closed by the museum's final check: twelve tasks drawn from across the halls, an 80 % threshold, unlimited attempts, shuffled answers. The feedback keeps the metaphor – a correct answer is «exhibit identified», a wrong one «closer to an error of display». No grade is deliberately reported to the LMS: the test is there for the visitor as a self-check, and if the client needs the actual scores, an administrator pulls them from the platform's statistics.

The 80 % is not an arbitrary figure: it is the baseline threshold approved company-wide for soft-skills courses, so the module was built to the shared standard rather than to its own.

This is where the module argues with itself. It is built to strip judgement out of the mistake, and it ends with a scored test – the judgement removed at the entrance comes back at the exit. Removing it was not an option: without a threshold the course would not have fitted the corporate standard. So it is held in check by tone instead: feedback says «exhibit identified» rather than «correct», a failed attempt sends the visitor back to the halls rather than announcing a failure, and the score is never reported to the LMS as a grade. The threshold could stay without bringing judgement back into the module.

The problem

A recruiter without an engineering background reads a résumé as a list of technologies and screens against that list. The mistake that gets corrected is the visible one: a weak candidate reaches the technical team and burns an hour. The expensive one leaves no trace anywhere. The course had to retrain what the recruiter listens for, not what they know.

The objective: the recruiter runs a screening call end to end and hands over a decision an engineer can act on – or a clear reason not to.

Why it works this way

The model behind it

Merrill’s first principles. The problem comes before any theory: the first lesson is the asymmetry of the two errors, not a definition. Activation runs through the recruiter’s own blind spots, demonstration through six questions that work, application through telling a real question from a gap – and integration through the last lesson, where the whole call is run end to end.

How it was checked

The quiz sets the bar at 75% with unlimited retries and answers revealed only after a wrong choice, so every second attempt leaves a trace. The first pass is a pilot with five to seven recruiters – too few for item statistics, enough to catch the questions everyone fails and the ones nobody does. The error the course is built around stays outside all of it: rejecting a strong engineer leaves no record anywhere, and the only way to see it is a second reader going back through the résumés that were turned down. The pilot brought back two things. The first question – the one that names the asymmetry the whole course is built on – nobody got wrong: it came minutes after the lesson that states it, so it measured recall rather than judgement. It now gives two recruiters' numbers for the week and asks whose week hides the more expensive problem: the principle has to be recognised in someone else's figures, where the visible loss looks the more costly of the two. The question about the candidate who would «have used a different logging library» split the group: half read the answer as cosmetic reflection, half as awareness of a constraint someone else imposed. The second reading is fair, and the wording still has to close it off.

The problem

HR specialists rarely fail to know what to say. They know – and then say it in a way that makes a decision look handed down from above and a refusal look personal. The programme was built around conversations that actually happen: turning down a request, delivering a decision you did not make, answering a question about a raise, and giving feedback.

The objective: the HR specialist words a refusal, someone else's decision and a piece of feedback so that both the clarity and the relationship survive.

Why it works this way

The model behind it

The gradual release of responsibility: «I do → we do → you do». The same model explains why the tasks compare two similar options rather than a right one and a wrong one: at the «we do» stage the learner is still leaning on a model answer, and the difference has to be visible in contrast, or the support does not hold.

The precision of that contrast follows from the model too. Push the two versions too far apart and the task becomes obvious and the support pointless; bring them too close and the difference disappears and the support falls short. In the audio task that calibration was the main work: both tracks were written for a single sentence, and the gap between the deliveries was adjusted until it sat in the right place. A synthesised voice gave what a live recording does not – the freedom to re-listen and re-cut the options as many times as it took.

How it was tested

Twenty-two tasks across sixty-five slides, with a certificate at the end. Feedback lands on each of them, and attempts are unlimited.

The tasks aim at what those conversations are assembled from: brevity of wording, preparing for a difficult meeting, the tone that keeps trust, active listening, self-control at the moment of irritation. The whole conversation is held outside the module, with a live counterpart – a deliberate split, because a formulation can be checked in a closed question and a conversation cannot.

The problem

In a training room Lencioni's pyramid turns into five words on a slide. Managers nod, and afterwards cannot say how avoidance of accountability differs from inattention to results. What was needed was a form memory can hold on to.

The objective: the participant tells the five dysfunctions apart and recognises their signs in his own team.

Why it works this way

The model behind it

Keller's ARCS model, its first two steps. Attention: a cocktail menu breaks the expectation set by slides about Lencioni's pyramid, which managers have already heard several times over. Relevance: a recipe with ingredients forces the dysfunction to be broken into signs, and the signs are what a participant recognises in his own team. Motivation had to be designed as deliberately as the content – a model everyone considers familiar gets no attention otherwise.

How it was tested

There is no assessment inside, and none was intended: eleven spreads, ten interactive actions, not a single scored task. This is material for a live training session rather than a standalone module, and the facilitator does the assessing in the room.

The material was used for internal training with regular employees rather than senior management, and it worked best somewhere other than where it was aimed. The departments work remotely, and the spreads were picked up for a team's first introductions: a cocktail recipe gives people a way to talk about trust and conflict without getting personal and without needing to be close first.

The problem

Knowing the script does not carry over into the conversation. The assistant remembers not to push, pushes anyway, and the customer finds it awkward to object and leaves without a word. The simulation had to create a place where that mistake costs something visible and costs no revenue.

The objective: the assistant holds a conversation with a first-time buyer – finding out the task before naming a model, and keeping the contact when the customer wavers.

Why it works this way

The model behind it

Miller's pyramid. The «Canon DSLRs» course takes the assistant to «knows how»: asked, he can name the right line. The simulation aims one step higher, at «shows how» – the line has to be spoken in a conversation where the other person does not answer from a script. Most communication training is lost between those two steps, which is why the simulation is the final task of the course: it can be taken on its own, but it was designed as the way out to «shows how» once «knows how» is in place.

How it was tested

Eight of the thirteen scenes contain an answer that ends the conversation on the spot: the customer walks out. The bad ending is not hidden at the finish – it is reachable from almost any step, and that skew is deliberate, because on a real shop floor conversations do not collapse in the final minute either. Feedback comes through emotion: calm, confused, unhappy and annoyed shift from line to line, long before the outcome is visible. The player can check himself immediately, without waiting for the closing screen. The entry screen says it plainly: this is not an exam, and if the conversation goes wrong, take it again with a different tactic. The cost of a mistake is visible and still not punitive – otherwise the assistant would start picking safe lines instead of honest ones.

The problem

On day one a new manager gets a stream rather than a task list: emails, meetings, people with their own agendas. Decisions are made in minutes and surface weeks later. A workshop cannot reproduce that: it teaches correct answers, and here there are none.

Why it works this way

The model behind it

Merrill’s principles: learning happens when someone works on a real problem, draws on prior experience, applies a solution and transfers it into their own practice. That is why there is no introductory theory here – the player lands inside the working day, and the debrief arrives after the decision rather than before it.

How it was tested

I watched where players stalled and which options nobody picked: an ignored branch means the choice reads as obvious, and the wording needs rewriting. Separately I checked whether the morning bet changes after the evening comparison, which is the sign that the reflective loop worked. The actual figures are covered by an NDA.