Teaching technical skills

Why it matters

Technical skills are learned by doing, but beginners cannot do much until they have seen how an expert approaches the task. A live demo can be the clearest part of a session or the most confusing: too fast, too small to read, full of shortcuts the trainer does not even notice. Combining a well-run demo with structured, gradually released practice is the core method of technical training.

Make expert thinking visible

Experts work largely on autopilot. Allan Collins, John Seely Brown and Susan Newman (1989) proposed cognitive apprenticeshipAn approach where an expert models a task while thinking aloud, then coaches and scaffolds learners and gradually fades support. to address this: the expert models the task while thinking aloud, then coaches and scaffolds the learner's attempts, and gradually fades support. Learners articulate their reasoning and reflect on it.

The key move is the think-aloudNarrating your reasoning while doing a task, explaining why you make each decision and not just what you click.: say why, not just what.

• Silent demo: opens incident, picks SEV2, assigns team.
• Think-aloud demo: "First I check if customers are affected. The alert says checkout errors at 8%, so yes. Is there a workaround? No. Our matrix says customer-facing with no workaround is SEV2 unless revenue is fully down, so SEV2. Now I page the on-call for payments, because that's the service in the alert."

Gradual release: I do, we do, you do

P. David Pearson and Margaret Gallagher (1983) described the gradual release of responsibilityMoving responsibility from trainer to learner in stages: the trainer demonstrates, the group practises together, then learners work alone., often summarized as:

1. I do: the trainer demonstrates with think-aloud.
2. We do: the group does a similar task together, the trainer guiding and asking learners to predict each step.
3. You do together: pairs do a new task while the trainer circulates.
4. You do alone: each learner does a task independently, ideally with a check at the end.

Rosenshine's Principles of Instruction (2012) adds a useful target for guided practice: aim for a high success rate (he suggests around 80%) before moving to independent work. If most people are failing in "we do", go back to "I do" with a smaller step.

Running the live demo

Prepare:

• Use a clean practice environment (a training tenant, sandbox or test account), never production or real customer data.
• Increase font size and zoom; close notifications; use a high-contrast theme.
• Script the demo steps, and have a recorded or screenshot fallback in case the tool or network fails.
• Reset the environment so it starts the same every time.

During:

• Say what you are about to do before you do it, then do it slowly.
• Narrate where you are clicking ("the Severity dropdown, top right").
• Avoid keyboard shortcuts, or say them out loud.
• Pause after each step and ask a prediction question: "What do you think happens when I set SEV1?"
• If something goes wrong, use it: diagnose out loud. Learners will hit the same errors.

Practice structures that help

PRIMMA practice sequence for code or tools: Predict what it does, Run it, Investigate it, Modify it, then Make something new.. For programming and scripting, Sue Sentance and colleagues developed PRIMM: Predict what code will do, Run it, Investigate it with questions, Modify it, and then Make something new. The same idea works for tools: predict what a configuration change will do, try it in the sandbox, investigate the result, change it, then build your own.

Faded worked examples. Give a fully worked incident write-up, then one with the timeline missing, then one with only the headings.

Parsons problemsExercises that give the correct steps or lines of code in scrambled order for learners to arrange, focusing them on structure rather than recall.. For procedures and code, give learners the correct steps in a scrambled order to arrange. Dale Parsons and Patricia Haden (2006) introduced them for programming; they reduce load because learners focus on structure rather than recall.

A 30-minute technical segment, end to end

Common mistakes

• Demoing too fast with tiny text.
• Showing the happy path only, so learners are lost at the first error.
• Long demo, then "now you try everything".
• Demoing in production or with real data.
• Moving to independent practice when most learners are still failing.

Quick recap

• Model with a think-aloud so expert decisions are visible.
• I do, we do, you do together, you do alone.
• Prepare a clean, resettable environment and a fallback.
• Use predictions, PRIMM, faded examples and Parsons problems for practice.

Try it

On the Do tab, run a live demo with guided practice for one or two learners.