Session 2: Demonstration, Practice & Assessment

A demonstration is not showing what you can do; it is showing what they should do, at a speed they can follow. The difference is significant. An expert demonstrating fluently is impressive and useless, because the learner cannot separate the essential actions from your personal shortcuts, and cannot see the decisions you are making without thinking about them.

So demonstrate slowly and deliberately, doing only what you would tell them to do. Remove the flourishes, the keyboard shortcuts you acquired over years, and the steps you do automatically. If a shortcut is genuinely worth teaching, teach it separately after the basic method works — introducing it during the demonstration means the learner cannot tell which actions matter.

Then show the finished result first. Seeing where they are going makes every subsequent step make sense, because the learner has a destination to attach the information to. Demonstrating the process first and revealing the result at the end means the learner holds a sequence of unexplained actions, most of which they will have forgotten by the time the point arrives.

The most valuable part of a demonstration is not what you do — it is why. 'I am clicking here' teaches an action. 'I am clicking here because this is the only place that applies the change to the whole column rather than one cell' teaches a decision, and decisions are what let someone work without you.

Narrate the choices you did not make as well as the ones you did. 'You could do this with a filter, but I am using a formula because the data will change and a filter would need reapplying.' That single sentence transfers judgement rather than procedure, and it is the difference between a learner who can follow instructions and one who can solve the next problem you have not shown them.

Then show a mistake on purpose. Do something wrong, let it produce the wrong result, and explain how you recognise it and fix it. Learners will make that mistake, and the person who has seen it made and corrected recovers in seconds while the one who has not sits stuck for an hour. It also gives them permission to make mistakes, which matters more than it sounds.

Practice is where learning actually happens, and it needs a gradient. Guided practice first: they do the task with you watching, able to intervene immediately. This is where errors are cheap, because they are caught in seconds rather than practised into habit. Then supported practice: they do a variation with the steps visible but you not intervening unless asked. Then independent practice: they do it alone, without the steps, which is the only real test of whether it stuck.

The critical rule is that practice must be slightly harder than the demonstration. Repeating exactly what you showed produces the illusion of learning — they succeed, everyone feels good, and nothing was learned, because they were following rather than solving. Change one thing: different data, a different goal, a small complication. That change is what forces the understanding rather than the memory.

Then the discipline of not rescuing too quickly. When a learner is stuck, the instinct is to take over and fix it, and it is the most counter-productive thing a teacher can do. Ask where they think it went wrong, ask what they expected to happen, ask what they would try next. Thirty seconds of productive struggle teaches more than thirty seconds of watching you fix it, and rescuing teaches them that the answer to difficulty is waiting for you.

Feedback is information about the gap between what was produced and what was wanted. Most feedback fails because it is evaluative rather than descriptive: 'good job' and 'this is wrong' both tell the learner where they stand and neither tells them what to do differently. Only the second kind changes the next attempt.

The structure that works has three parts. What specifically was right — 'your formula referenced the correct range'. What specifically was wrong — 'it used an absolute reference, so it broke when you copied it down'. What to do next time — 'use a relative reference when the formula needs to travel'. Specific, actionable, and about the work rather than the person.

Then the timing. Feedback given immediately after the attempt is worth several times more than feedback given a week later, because the learner still remembers what they were thinking. This is an argument for reviewing work in class rather than collecting it, and for commenting on the process rather than only the result. And keep it proportionate: three specific points land, twelve buries the learner and none of them get acted on.

Assessment answers one question: can they do the thing? Not can they describe it, not did they attend, not do they seem to have understood. The most common assessment mistake is testing recall of what was taught rather than the ability the objective stated — a quiz about formulas when the objective was writing them.

So assess by asking for the thing. If the objective was 'write a formula that totals a column and handles blanks', the assessment is a column with blanks in it and a request to total it. Not a question about what SUM does. Performance assessment is harder to mark and far more informative, and for a practical subject it is the only kind that means anything.

Then make it low-stakes and frequent. A large test at the end tells you too late to help anyone. Small checks during the lesson — a quick task everyone does, a question with a specific answer, a screenshot of their screen — tell you in the moment whether to move on or go back, which is the only time that information is useful. And state the standard in advance, so the learner knows what success looks like before attempting rather than guessing.