Incrementality
CTV attribution and incrementality
Understand what CTV attribution credits, what an incrementality test estimates, and how to use each result within its measurement limits.
CTV attribution credits a recorded action to an eligible ad exposure under specified matching, timing and credit rules. Incrementality estimates what changed because the advertising ran. Because a credited sale might have happened anyway, choose the method to match the decision you need to make.
Start with the decision
If you need to understand which recorded actions received CTV credit, request an attribution report. If you need to decide whether CTV added sales or enquiries, plan a credible control or comparison. Define the business outcome, Australian locations, campaign dates and decision threshold before launch. A purchase, submitted enquiry and page visit are different outcomes.
| Question | Evidence | Supported conclusion |
|---|---|---|
| Did the buy deliver? | Impressions, dates, inventory and device reporting | Delivery within the reported scope |
| Which actions received CTV credit? | Recorded events with match, window and credit rules | Association under those rules |
| Did the CTV change affect the outcome? | A planned control or credible market comparison using the same business outcome | An estimated effect for the tested population and period, with uncertainty |
An attribution system needs an eligible exposure, a recorded action and a supported way to associate them. A CTV ad can run on a shared television while a purchase occurs later on a phone. A household association cannot establish which resident watched or bought. Credit also depends on the post-view window and how other ad touchpoints are treated.
Display & Video 360 provides documentation titled “Connected TV reporting and forecasting”. Confirm which cross-device conversions, inventory and events are supported for the proposed buy.
CTV Attribution vs Incrementality: Key Differences and Use Cases
- Purpose
- Identify which recorded actions were credited to CTV ads under system rules
- Method
- Attribution reporting with match, timing and credit rules
- Output
- Number of credited conversions (e.g. purchases, enquiries)
- Use Case
- Assessing campaign delivery and performance within defined scope
- Purpose
- Estimate the actual change in business outcome caused by CTV advertising
- Method
- Controlled geographic test or credible market comparison
- Output
- Estimated incremental impact (e.g. additional sales or enquiries)
- Use Case
- Deciding whether CTV advertising drove measurable change in outcomes
Use complementary evidence for different questions
For a budget decision, use attribution reporting to see which recorded actions received credit under the system’s rules. Pair it with a controlled test of the same business outcome to estimate whether CTV added sales or enquiries.
The IAB Australia Video Measurement Framework identifies attribution and incrementality testing as distinct methodologies that can be used together. If the credited count and test estimate differ, use the test to assess the effect and do not treat attributed actions as incremental outcomes.
Pros and Cons of Using Attribution vs Incrementality for CTV Decisions
- Attribution Reporting – Pros
- Fast, scalable, provides insight into which actions were linked to ads
- Attribution Reporting – Cons
- May overstate impact – credited actions may have occurred anyway
- Incrementality Testing – Pros
- Provides causal evidence of advertising impact; supports investment decisions
- Incrementality Testing – Cons
- Time-consuming (weeks), requires careful design and sufficient data
Plan an effect estimate separately
A before-and-after sales chart can reflect promotions, prices, stock, competitors or other media. Where feasible, assign comparable geographic areas to a CTV treatment and control before launch, then analyse the same business outcome in both.
If assignment is not random, explain why the selected markets form a credible comparison and what differences remain. Check whether the buying path can implement the geographic contrast and whether travel or targeting imprecision could blur it.
Use raw business outcomes for a sales test where possible. Attributed conversions can change when matching or credit rules change, even if underlying sales do not.
Assess whether the available markets and outcome volume can distinguish an effect large enough to matter. If the estimate remains too uncertain for the spending decision, call it inconclusive.
Planning a CTV Incrementality Test: Key Steps Before Launch
- Define business outcomee.g. purchase, enquiry, page visit – must be consistent across markets
- Select comparable geographic areasAssign treatment and control regions before launch; ensure similarity in demographics and market size
- Confirm buying path feasibilityEnsure platforms can implement geographic targeting and budget allocation differences
- Set primary KPI and measurement levelConfigure at campaign or advertising-account level (e.g. transaction, revenue)
- Assess statistical powerCheck if available markets and outcome volume can detect a meaningful effect
Allow for the practical test schedule
Google’s Meridian GeoX implementation guidance says a geographic study normally takes 7 to 14 weeks from pre-test planning through analysis. The planning phase can take up to two weeks, the active test can run for 3 to 10 weeks, and the waiting and analysis phase can take more than two weeks. Treat this as a planning reference, not a guaranteed duration for every CTV test.
Before launch, the GeoX guidance calls for defining geographic units, preparing pre-test data, establishing geographic assignment and determining the experiment’s budget changes. It also recommends pre-configuring geographic targeting at campaign level across participating buying platforms, so the intended treatment and comparison conditions can be implemented.
During the active test, run the strategy in designated treatment regions and maintain standard operations in control regions. If campaign effects may be delayed by the time between exposure and conversion, a waiting period is recommended; during it, all regions remain under business-as-usual conditions.
GeoX allows one primary KPI variable, with transaction, sales and revenue given as examples. Its guidance says the measurement level is configured at campaign or advertising-account level. Set the primary outcome and measurement level before launch so the test’s result answers a defined business question.
Typical Timeline for a GeoX-Based CTV Incrementality Test
- 2weeksPre-test Planning
- 3Active Test Period
- 2weeksWaiting & Analysis Phase
- 7Total Duration
Report each layer within its boundary
Show delivery, credited actions and estimated effect separately. Give each its dates, outcome definition, inventory or market coverage and denominator. An attribution count must not be relabelled incremental sales. A market-test estimate applies to its tested markets, period and intervention; it does not automatically describe a later national buy.
For a repeat or expansion decision, compare the tested conditions with the next proposed campaign. Where evidence is weak, specify the narrower conclusion it supports and what a better test would need.
Key Metrics in CTV Attribution and Incrementality Testing
- Attributed Conversions
- Recorded actions credited under matching rules
- Incremental Sales Estimate
- Change in outcome attributed to CTV exposure (from test)
- Test Duration (GeoX)
- 7–14 weeks (planning to analysis)
In this guide
- Comparing exposed and unexposed markets carefullyPlan a CTV market comparison by selecting credible controls, checking the delivery contrast and reporting an effect estimate with uncertainty.
- Separating household correlation from causal sales impactSee what household CTV-to-sales matching can show, why exposed buyers may differ and when a control is needed for a causal claim.
- Reviewing and comparing CTV attribution windows in reportsAudit CTV post-view and post-click windows, credit rules, event counting and data cut-offs before comparing attributed actions.
- Reporting a CTV result when measurement coverage is partialShow measured CTV outcomes beside eligible delivery, exclusions and unknowns without extending a partial finding to the whole buy.
