Keep viewability, attention signals, and outcomes in separate columns

Viewability is an eligibility measure

A viewable impression applies pixel and time conditions to a measurable ad impression. The MRC display guidance uses a baseline in which at least half of a display ad's pixels are in view continuously for at least one second, with a different pixel provision for certain large units. Video and mobile guidance have their own details. A report should identify the medium, rule set, measurer, measurable count, and denominator rather than borrowing one headline threshold for every format.

This measure answers whether an ad had an opportunity to be seen under the stated rule. It does not establish that a person looked at it, understood it, remembered it, approved of it, or bought anything. A non-viewable result is useful diagnostic evidence; a viewable result is not a human-response certificate. Keep the wording disciplined in media kits and sponsor reports.

Source notes: Viewable Ad Impression Measurement Guidelines, version 2.0

Attention needs a declared model

The IAB and MRC attention guidelines group approaches around exposure, attentive response, cognitive and emotional response, and outcomes. Implementations may use inputs such as time in view, audibility, interaction, gaze or modeled data. Those inputs are not interchangeable. A hover is behavior; gaze measurement is a different observation; a modeled score is an estimate. The provider should disclose collection, training or calibration, missingness, coverage, privacy treatment, and validation.

Ask which environments are measurable and which are inferred. If a score covers 40% of impressions directly and models the rest, show both the coverage and the model status. Do not compare two attention scores merely because each runs from zero to one hundred. Their inputs, calibration populations, and intended decisions may differ. Vendor labels remain vendor claims until the method and evidence are reviewed.

Source notes: IAB and MRC Attention Measurement Guidelines, version 1.0

Use a three-column campaign read

Separate delivery eligibility, response signals, and business outcomes. Illustrative campaign: 120,000 rendered impressions; 90,000 measurable; 58,500 viewable; 1,800 qualified interaction signals; 420 landing sessions reported by the advertiser; 24 trial starts. Viewability is 65% of measurable impressions. The interaction rate is 2% of measurable impressions only if that is the declared denominator. Trial starts belong to the advertiser's measurement and should not be attributed to attention without an agreed design.

The useful questions differ by column. Low measurability prompts implementation checks. Low viewability prompts placement and page-behavior checks. A change in an attention signal prompts examination of creative, context, coverage, and model stability. A change in outcomes prompts landing-page, offer, attribution, and audience review. Collapsing all four into a single quality score hides the decision each metric can support.

Invented values; no performance benchmark is implied.

Source notes: Viewable Ad Impression Measurement Guidelines, version 2.0 · IAB and MRC Attention Measurement Guidelines, version 1.0

Specify the question before buying measurement

Write the decision first: compare two placements, diagnose a creative, or add an advertiser reporting field. Then ask whether ordinary viewability and outcome data already answer it. If an attention product remains useful, run a bounded test with coverage reporting, a predeclared comparison, and no promise that the score will predict sales. Preserve raw component metrics where the agreement allows; a composite score alone is hard to audit after the model changes.

Do not invent a fee estimate or assume a small publisher needs specialist attention measurement. Request current pricing and terms only for a real use case. The test report should state method version, directly observed share, modeled share, excluded environments, sample size, uncertainty where provided, and any outcome data supplied by the advertiser. End with what changed in an operating decision, not whether the dashboard looked sophisticated.

• Name the medium and viewability rule used.
• Report measurable and unmeasurable inventory separately.
• Treat modeled attention as an estimate with coverage limits.
• Keep conversions and causal claims outside the score unless the study supports them.

Source notes: IAB and MRC Attention Measurement Guidelines, version 1.0