Measuring repeated exposure within a household ID

Choose the identity unit and campaign date range before counting. Treat each distinct household ID as one household unit; a device count and a modelled person count answer different questions and should not be combined with household counts.

Viant’s cookieless Household ID™ is a named CTV identity example. It connects a CTV impression with smartphones, laptops and tablets operating under the same roof, so exposure can be assigned to a household rather than counted separately for each linked device.

Household IP matching, device graphs and deterministic tracking through authenticated streaming platforms are other named identity approaches. Household IP matching is described as faster to implement but less precise; person-level deterministic tracking is described as more accurate and requiring first-party integrations. Keep the resulting identity unit clear in every report.

Use the same campaign dates, inventory, household identity definition and impression-counting rule for each metric. Keep delivered impressions separate from viewable or completed impressions where those measures are reported; a frequency cap may use a different counting rule from a frequency report.

For household reach, count the distinct household IDs with at least one counted impression in the window. Calculate average frequency by dividing counted impressions by that reached-household count.

Show the exposure distribution as cumulative household counts at thresholds such as 1+, 2+ and 3+ impressions. For each threshold, count the distinct household IDs that meet it, then divide by the number of households reached (the 1+ count) to show its share.

With ten impressions and five household IDs, average frequency is two. If each ID receives two impressions, the 1+, 2+ and 3+ counts are 5/5, 5/5 and 0/5. If one ID receives six and four receive one each, the counts are 5/5, 1/5 and 1/5; the same average conceals a different repeat pattern.

Count an impression on a shared television against its household ID, not once for every person who may be watching. Household frequency is not person-level frequency; report a modelled person count separately and label it as such.

To combine sellers, join their impression records using a common, compatible household ID. Deduplicate IDs to calculate reach but retain the counted impressions to calculate frequency; if sellers cannot be joined this way, limit the result to the joined inventory and state the gap.

Review the distribution over settled, like-for-like weekly windows. Flag the heavily exposed tail and inspect contributing publishers or placements; do not treat a live dashboard as a complete household census.

Key Metrics for Measuring Repeated CTV Exposure Within a Household

Identity Approaches for Household-Level Tracking

Steps to Measure Repeated CTV Exposure at Household Level