Video MetricsMeasuring the job, not the views

The problem with views#

‘How many views did it get?’ is the first question asked about most videos and the least useful. Each platform defines a view differently, using different durations and conditions, and the definitions change. A view can mean someone watched the whole film carefully or that it autoplayed briefly while they scrolled past.

Views are not useless. They are a delivery and attention signal. But they rarely tell you whether the video did its job, and comparing view counts across platforms is close to meaningless. A measurement framework begins by asking what the video was for.

Four levels of video measurement#

Every video should be judged at the highest level its job allows. A brand film may only be measurable up to response and, with a brand-lift study, outcome. A product demo can often be measured right up to sales. Reporting only the bottom of the pyramid is a sign that nobody agreed what the video was for.

Match the metric to the job#

Reading a retention curve#

The audience retention graph, available on YouTube and in different forms on other platforms, is the most useful diagnostic for the craft of a video. It shows the share of viewers still watching at each moment. Its shape tells you where the video lost people, which tells you what to fix.

• A steep drop in the opening moments usually means the opening did not match the promise of the thumbnail, title or ad placement.
• A gradual decline is normal; the question is whether it is steeper than your other videos of similar length and type.
• Sharp dips mid-video point to a slow section, a confusing moment or an unwanted tangent.
• Bumps upward show rewatching: a moment people found valuable, interesting or confusing. Check which.
• A drop before the end card is normal; if your call to action sits there, move it earlier.

Vanity metrics versus decision metrics#

Measuring outcomes honestly#

Outcome measurement is hard because video rarely acts alone. Someone sees an ad, later searches for the brand, reads a page and buys. Attribution models will credit different touchpoints depending on how they are built. Our guides to marketing attribution and incrementality testing explain the options.

• Brand-lift studies offered by major platforms compare exposed and unexposed groups on recall or consideration. They are the cleanest awareness measure available to most advertisers.
• Holdout and geo tests compare regions or audiences that did and did not see the campaign.
• Viewer versus non-viewer comparisons on your site are useful but biased; keener buyers watch more. Use them as directional evidence.
• Before-and-after comparisons for support and onboarding videos, such as tickets on a topic before and after a help video went live.

Build your own benchmarks#

Published ‘average’ view rates and completion rates are tempting and mostly useless. They mix industries, formats, lengths, placements and view definitions, and they are rarely transparent about method. Your own history is a far better benchmark, because it holds most of those variables constant.

Keep a simple log of every video: its job, format, length, placement, spend, and the metrics agreed for it. After a dozen or so entries, patterns appear. You learn what a reasonable retention curve looks like for your explainers, what view rate your ads typically achieve on each platform, and which openings tend to hold attention. Those internal norms make every new result interpretable.

Reporting video to leadership#

Leadership does not need every metric. It needs to know whether video is doing the jobs it was funded for, at what cost, and what will change next. A good video report fits on one page.

• Start with the job and the outcome: what each major video or campaign was for, and whether the agreed measure moved.
• Show cost alongside result: cost per qualified lead, per completed view or per point of brand lift where measured.
• Explain one lesson: what the retention data or test results taught the team.
• State the decision: what will be scaled, stopped or changed.
• Keep attention metrics as context, not headlines.

The wider principles are in board reporting. One caution: resist the temptation to add every platform’s native metrics to a shared dashboard. Each platform defines views, engagement and reach differently, and putting them in one column invites false comparisons. Normalise where you can, for example by using reach among the target audience and cost per completed view, and label clearly where definitions differ.

Measuring video you do not pay to distribute

Organic, sales and support video is often left unmeasured because there is no media report to read. It can still be measured. Track plays and completion through your website analytics or video host, tag links in sales emails, ask sales teams which videos prospects mention, and compare support tickets on a topic before and after a help video goes live. The evidence is less tidy than a campaign dashboard, but it often shows video doing its most valuable work.

Setting up a video measurement plan#

Write the measurement plan into the brief. For each video, state its job, its primary metric, the supporting metrics, how the data will be collected, and when it will be reviewed. Make sure tracking exists before launch: UTM tags on links, events for video plays on your site, and conversion tracking on the landing page. See UTM parameters and GA4 events.

Published by Fabulous.Media, a network of specialist marketing agencies. Updated 9 October 2026. Platform features change often; check current official documentation before acting on platform-specific detail.