Key takeaway
A viewability rate uses its measurable population. Keep non-measurable delivery visible as unknown rather than treating it as observed non-viewability or attention.
Which impressions could the measuring system evaluate?
A publisher’s report can contain served impressions that the viewability technology could not evaluate. That is different from an impression the technology measured and found below its visibility conditions. If an export combines those buckets, a recipient may read missing measurement as observed failure. Identify the product, format and metric definition before describing what the report proves.
Google Ads defines Active View measurable and non-measurable impressions and distinguishes viewability rate from viewable-impression distribution. Its guidance says non-measurable delivery does not establish whether the impression was noticed. Preserve that uncertainty. A report about measurement coverage is useful without pretending to observe every delivered impression.
Reconcile a complete fictional reporting window
This hypothetical aggregate report contains 100 served impressions in one declared reporting window. Of those, 80 were measurable by the selected technology. Forty were viewable under the selected product/format definition and 40 were measured but non-viewable. Twenty were non-measurable. The buckets are invented and reconcile to the stated total; they are not an observed account result.
The viewability rate is 40 divided by 80, or 50%. Viewable-impression distribution is 40 divided by 100, or 40%. Those percentages answer different questions. Neither calculation requires assuming that the 20 non-measurable impressions were seen or unseen. Retain the count bridge and the metric name so a recipient can reproduce the denominator instead of guessing from a column called viewability.
| Bucket or metric | Completed toy value | Supported interpretation |
|---|---|---|
| Served | 100 | Declared total delivery population |
| Measurable and viewable | 40 | Meets the selected visibility rule |
| Measurable and non-viewable | 40 | Measured below that rule |
| Non-measurable | 20 | Viewability unknown to this technology |
| Measurable population | 40 + 40 = 80 | Denominator for this viewability rate |
| Viewability rate | 40 / 80 = 50% | Viewable share of measurable delivery |
| Viewable-impression distribution | 40 / 100 = 40% | Viewable share of total served delivery |
Keep the visibility rule and product with the result
Google Ad Manager describes viewability as an opportunity to be seen and documents pixel/time conditions that differ by format, including a separate large-display condition. Do not apply a display-ad threshold to every video or large-ad record. Nor should a label from one reporting product silently become the definition used for another. Preserve the actual format rule, measurement system and reporting scope.
Visibility is also a narrower observation than human attention, a unique person, a conversion or a valid commercial outcome. The fictional report establishes only its declared measurement buckets. A receiving analyst who needs attention labels cannot acquire them by renaming viewable to attended. That would add a conclusion the retained measurement did not observe.
Record whether the report contains aggregate counts, per-event evidence or only a final percentage. Check the product’s actual export availability before promising to reconstruct unavailable raw fields. Two rounded percentages alone may not recover the original counts or explain differences across reporting windows. A missing bucket should remain a documented reporting limitation, not an estimated value presented as observed.
Release a metric definition rather than an overstated label
The completed owner decision preserves the 20 unknown impressions, labels the 50% and 40% measures separately and withholds any attention claim. Its accompanying definition names the reporting product, observation window, formats, measured conditions, bucket counts and known gaps. If the archive stores only a percentage, state which denominator is known and which counts cannot be reconstructed.
The practical review can begin with one bounded report and one edge case where measurement is unavailable. Ask the reporting owner to reproduce the buckets and identify any format exclusions. The purpose is to decide whether the historical summary supports the proposed analysis, not to recommend advertising spend or declare that a higher percentage improves commercial results.
Inventory can describe aggregate reporting categories without exposing impression identifiers, cookies, devices or user behavior. Review source terms and confidentiality before any actual report is shared. A permissioned VOID introduction does not authorize customer-level records, certify the metric or establish buyer demand. A later sample, evaluation and license each require their own scope and approval.