Measuring whether a video helps users complete a task

Define successful task completion without referring to the video. Measure how often eligible attempts reach that end point and observe where people struggle. Comparing people who watched with those who skipped the video can reveal an association, but it cannot by itself show the video caused the difference.

Define an attempt and its finish

Describe something a visitor can do: submit an enquiry, find the correct service condition or complete a setup step. Decide when a genuine attempt starts, what counts as success and what evidence will confirm it. A button click may show intent; a confirmed submission, completed product step or correct answer is stronger evidence of completion.

Set eligibility before calculating a rate. A visitor reading opening hours is not a failed setup attempt. Keep incomplete forms, error responses and confirmed submissions distinct. For an information task, analytics may not reveal whether someone found the right answer, so a task-based question or observation may be needed.

Record the unit of analysis, start, successful end point, time window and evidence source. If a person can make several attempts, decide whether to count attempts or people. An apparent exit may also reflect a tracking gap or a move to another device; avoid labelling every missing end event a failure.

Connect viewing to the journey

Map eligible attempt, opportunity to view, playback if chosen, task action and confirmed outcome. Instrument the steps that answer the decision. GA4 can collect custom events and parameters.

If you use a funnel report, check what it counts and whether its rules fit the task before treating it as an attempt-level completion rate.

Report overall completion for eligible attempts. Show watchers and non-watchers as separate descriptive groups, with their counts and the definition of “watched”. Leave viewing optional in the task path: someone may complete the task using the written instructions.

Observe what the numbers miss

Ask people who match the intended audience to attempt a realistic task using the page normally. Let them decide whether to play the video. Record whether they reach the correct end point, where they hesitate and what they expected. Someone who succeeds without playing the video may show that the written page already serves that task.

On a setup page, a person might watch the demonstration and still choose the wrong setting because the final label is unclear. In that case, a clearer label or written step may help more than another video. This is an example of how to interpret a finding, not a reported result.

Assess the video’s contribution

If the decision warrants a test, compare eligible visitors assigned to otherwise similar page versions, with and without the proposed video treatment. Set the outcome, allocation method, time window and stopping rule before launch.

Keep essential written instructions usable in both versions, and record page or product changes during the comparison. Check the quality of outcomes as well as their count: a larger number of invalid enquiries would not establish a better task result.

Where a controlled comparison is impractical, describe the observed association and task research with their limits. The decision is whether the page helps people complete the right task and what evidence shows the video contributes.