Show first-use guidance at the moment of need, not as a dump up front

What it is

First-use guidance is about when teaching appears in the first session, before anyone has a working method. Just-in-time puts the sentence on the action at the instant that action becomes necessary. Front-loading puts a tour, welcome stack, or feature list at the door, as if there were a training interval before work. This entry is only about timing in that first session. It is not about later progressive reveals as features become relevant, and not about whether a tour can be skipped. Front-loading extracts teaching from the task; just-in-time embeds it.

Why it happens

People arrive with a job, not an appointment for a product class. Working memory is already holding “where do I click to make the first thing.” A front-loaded tour inserts a second goal: memorize five names. Encoding specificity says knowledge comes back in a similar situation; hearing “export lives here” on a welcome page tags the memory to the wrong episode, so the cue is missing when export is actually needed. Front-loading also delays time to first success. Without a success, the product is stored as an interrupted wait, with nothing for the lesson to attach to. Just-in-time teaching uses the current action as the retrieval cue—the control and the sentence share one fixation. Front-loading looks more complete because it follows the product’s table of contents, not the task.

Studying it

Compare “watch a full tour, then do the first task” with “no tour; one sentence at the step where the first task stalls.”

Independent variables: instruction before versus during the task, number of front-loaded screens, whether copy sits next to the control. Dependent variables: time to first success, wrong-target clicks after the tour, whether the same action can be done next day with no prompt.

Lab participants expect a test and force themselves through the front matter, which inflates retention. Live install funnels lose people on screen one; that measures the gate, not teaching. Higher retention among people who finish a tour is not evidence the tour worked—those people were already more motivated.

Where it stops holding

Safety, consent, and irreversible authorization must be seen before the action; that is notice, not teaching. A genuinely novel canvas grammar may need a very short demonstration, but it should still connect immediately to a first doable step rather than dump the whole syntax. Returning experts and people coming from a previous version already hold a task schema; a front-loaded class teaches what they can already do. Booths and training rooms are explicit classes, so front-loading is legitimate there—those completion rates do not justify a class at everyday install.

Applying it

• Name the one core action after first open; bind the sentence to that control; delete doorstep feature lists that do not serve that action.
• If a welcome page only restates what the UI will show, remove the page and put the sentence on the empty work area or beside the button.
• For a first step that needs permission or payment, speak at the trigger, not as a post-install inventory.
• Verify by walking someone new from install to first success and counting stretches where they only watch. Delete or relocate those stretches to the action, then compare time to first success and next-day redo.

Related

• Within the group: H2.01.2 Multi-screen intro tours are skipped as a set · H2.01.3 First-run teaching decays almost immediately
• Adjacent: H2.03 Progressive onboarding · H2.04 Skippable and replayable guidance · E4.14 Coach marks
• Search terms: just-in-time onboarding · front-loaded tour · time to first value