Measuring owned video performance

Measure each video against the job it does on its website page. Check whether visitors had an opportunity to use it, whether playback began and reached a useful point, and whether the relevant task was completed. A play count alone cannot show that the video helped.

Set the purpose before choosing metrics

Record the page, the visitor's question and the outcome the video is meant to support. A product demonstration may help someone judge fit before an enquiry; a setup walkthrough may help an existing customer finish a step. They need different outcome measures.

Define each rate's denominator and counting unit. Starts per eligible page view differs from the share of starters who reach a viewing point. Replays can produce more start events than page views, so count or label them separately. A page view also does not prove that a player farther down the page was seen.

Check the events available

Record what each player and placement sends: the trigger, video identifier, page or placement, and any time or progress value. Inspect the events on the actual page before relying on a report. Keep internal or test activity out of production figures where the setup permits.

Google Analytics enhanced measurement can automatically collect certain interactions when enabled. For each player and embed, check which viewing interactions are collected and define a measure that matches the page's purpose.

Identify automatic playback and repeat plays. Use a stable video identifier and a page or placement field when a video appears in several places. Note when the video, page, player or tracking changes so later figures can be interpreted in context.

Google Analytics lets you view enhanced measurement event data in the Events report. For other events, Google's developer guidance identifies Realtime and DebugView as places to check arriving events and parameters. Use these checks to catch missing or unexpected data before interpreting performance.

Key Video Tracking Considerations

Validate collection and protect data

Google Analytics enhanced measurement options can be enabled or disabled for a web data stream, and individual options can be edited. Check the settings before treating an event as available: Google says a user needs Editor or higher access at the property level to change those options.

For a website event that does not fit an automatically collected, enhanced or recommended event, Google Analytics supports custom events. These require implementation, so confirm the event name and parameters in Realtime or DebugView before using them in reporting.

Enhanced measurement can send events as soon as it is enabled. Google requires that personally identifiable information is not collected, so review event data and parameters for information that could identify a person before relying on the measurement.

Keep viewing and outcomes separate

A compact report can show each video's purpose, eligible page views, starts, a declared useful-viewing measure, the task outcome, raw counts and date range. Define the useful point for that video before examining the results.

In a demonstration, the useful point may be the visible result; in a tutorial, the essential instruction may come earlier than the end. Reaching a playback point does not prove that someone paid attention or learnt the material.

Match the outcome to the page. An enquiry page might record a confirmed submission; a help page might record a completed setup step if the product can observe it. A click towards the next step is weaker evidence than confirmed completion. For an information task that analytics cannot confirm, observe people attempting it or ask a question that checks whether they found the right answer.

Combine analytics with task evidence

Digital analytics is only one way to assess whether a service is working. For a video-supported task, combine playback and outcome measures with relevant user feedback or usability testing; call centre information may also reveal difficulties that page events do not capture.

For a task that can be observed, completion and the time taken can add context to the event data. A change in starts or progress does not by itself show that the task became easier, so interpret those measures alongside evidence of the outcome.

Make a decision from the evidence

Review a video within its page context first. Few starts may warrant a check of placement, surrounding text and player availability. Many starts followed by little progress may warrant a review of the opening and where the answer appears. Strong viewing alongside task failure may point to unclear instructions or a difficult next step elsewhere on the page.

These patterns are prompts to investigate, not diagnoses. People who choose to watch may differ from those who skip the video. If the decision warrants a causal test, compare otherwise similar page experiences against the same task outcome.

Use task observation to learn where either experience fails. Then decide whether to retain the video, revise it, change its placement, improve the written answer or remove an asset that no longer helps.

Repeat outcome measurement periodically when you need to know whether a change helped. Comparing task completion and time taken across rounds can show whether the task is becoming easier, while playback measures help describe how visitors used the video during those periods. Keep the compared page experience and outcome definition consistent.