Video quartiles: find the drop before calling the ad ineffective

A video report lists impressions, starts, 25% progress, 50%, 75%, and completions. The numbers look like a simple funnel, but that interpretation needs a few checks.

A quartile marks progress through the asset. Depending on the reporting system, these fields may represent event counts with their own rules for retries, playback, and eligibility. Read those definitions before treating every row as a unique person.

Establish what completion rate divides by

Some reports use video impressions as the denominator; another analysis might use starts. Google’s video reporting guidance and metric definitions describe the fields used in Display & Video 360.

Suppose an illustrative report contains 10,000 impressions, 8,000 starts, and 6,000 completions. Completions divided by impressions is 60%. Completions divided by starts is 75%.

Both calculations are arithmetically valid. They answer different questions and should not carry the same unlabeled heading.

Keep the raw counts alongside the rate. If two partners disagree, align their event definitions before investigating missing playback.

Compare videos that offer the same choice

A six-second non-skippable ad and a thirty-second skippable ad do not present the same opportunity to finish. Duration, skip behavior, placement, device, and whether the viewer initiated playback all affect the comparison.

Break the report into meaningful groups before ranking creative. If one version receives more short-form, non-skippable inventory, its higher completion rate may reflect the delivery mix.

Likewise, a completed video does not establish that the sound was on or the message was noticed. Those require separate evidence where measurement supports it.

Use the pattern to choose a test

A loss between impressions and starts points toward a different investigation from a decline near the end of playback.

For the first case, inspect rendering, loading, player compatibility, and definitions. For later progress losses, review the viewing environment, skip behavior, asset length, and message sequence.

The VAST troubleshooting guide covers technical failures. Do not request a creative rewrite before ruling out a playback problem concentrated on one device.

Watch for impossible-looking sequences, such as more completions than starts. Check date boundaries, filtering, and duplicate-event handling before drawing a conclusion.

Turn the diagnosis into a creative decision

If the evidence suggests viewers leave before the key message, test a version that presents that message earlier while keeping the objective and comparison method clear.

Record the asset version, duration, start date, and eligible inventory. Evaluate the agreed outcome as well as completion. A shorter asset can finish more often without producing a better business result.

A useful video analysis identifies where the change occurs, which environments contribute, and what test would distinguish the likely explanations. The headline completion percentage is only the starting point.