cellphone, smartphone, vpn, streaming, sport, online, live, app, person, vpn, vpn, vpn, vpn, vpn, app
Photo by StefanCoders on Pixabay

Incrementality

Part of CTV attribution and incrementality

Reviewing attribution windows in CTV reports

Audit CTV post-view and post-click windows, credit rules, event counting and data cut-offs before comparing attributed actions.

This is an editorial pen name used by the publication, not a claim of an independently verified personal identity. Original editorial credit: Measurement Desk.

undefined

An attribution window is the period after an eligible ad impression or click in which a recorded action may receive credit. Post-view and post-click are separate clocks: each starts with its eligible exposure event, while the conversion event and credit model determine whether a later action qualifies and how credit is assigned.

In CTV the post-view clock does more of the work, because streaming lacks the deterministic, click-based data of the open web (stackmatix.com). Two CTV-related window settings appear in the guidance: MNTN Performance TV in Northbeam, where the setup calls for attribution and look-back windows of 30+ days, and Kochava's re-engagement trackers, where impression lookback windows are typically set to 24 hours.

Review effective post-view and post-click durations separately, alongside each report's event definition, credit model, counting method and cut-off. MNTN's Performance TV windows and Kochava's re-engagement lookback sit in different configuration contexts, so they are not a like-for-like post-view and post-click comparison. A credited action does not show that an ad caused it.

Find the window behind each CTV report

Record effective post-view and post-click durations separately, along with the event that starts each clock, the conversion event counted, attribution model, counting method, report time zone, event-date convention and treatment of repeated actions. Compare those fields side by side before reading totals.

CTV reports draw on ad platform data from sources such as Roku and Amazon DSP, and measurement partners such as IAS and Nielsen (stackmatix.com). Every platform has its own rules for what a conversion is, which interaction gets credit, how long that credit lasts and how much of the customer journey it can see (inbeat.agency).

Check the effective configuration for the relevant campaign, placement or tracker rather than assuming a platform default applies everywhere. A label is not a universal duration, and providers may define labels differently.

A longer window cannot recover an impression or action the matching system never recorded.

Post-click and post-view clocks commonly differ in length: a paid media guide notes that reports may apply different attribution periods (inbeat.agency). Record both durations before comparing totals, because the shorter clock will credit fewer actions.

For Kochava re-engagement campaigns, view-through attribution for impressions is disabled by default for existing apps and enabled by default for new apps, so advertisers should review the configurable impression lookback windows and target conversion events, and check that campaign trackers are measuring impressions.

Separate window settings from credit rules

Identify the attribution model used for each CTV report. CTV measurement guidance argues teams should move beyond last-click and adopt probabilistic and time-decay models, with time-decay still crediting earlier CTV views (stackmatix.com). If reports use different models, differences in credit cannot be attributed to windows alone.

A conversion can follow several ads, and different models can assign credit to different eligible interactions. Another channel can receive credit for the same event in a separate report.

Northbeam’s Clicks + Deterministic Views model assigns fractional conversion and revenue credit across all channels in the customer journey.

In Northbeam, MNTN performance only populates when Clicks + Deterministic Views is selected and MNTN is chosen as the platform (help.mountain.com). Keep that model selection visible alongside the window settings when reviewing MNTN totals, because both the eligibility period and credit allocation rules affect what the report shows.

Reports may use different event-counting methods, such as counting every event or limiting repeated events. A window comparison is incomplete if the event or counting method changed too.

Northbeam’s Accrual Performance dates revenue and transactional credit to the marketing touchpoint; Cash Snapshot attributes credit to the conversion date rather than the impression date. Keep accounting mode and the report cut-off visible when reviewing totals: a different assigned date can move credited revenue into or out of the period being read.

Compare window contexts like for like

Kochava's view-through attribution announcement says impression lookback windows on re-engagement trackers are typically set to 24 hours, adjustable by the advertiser for deterministic and probabilistic matches. That guidance says impression-based re-engagement can be particularly relevant to OTT and CTV campaigns, where ad views are more predominant.

MNTN performance in Northbeam requires Clicks + Deterministic Views to be enabled and MNTN selected. MNTN's setup guidance calls for attribution and look-back windows of 30+ days, a different configuration context from Kochava's typical 24-hour impression lookback, so the two are not a like-for-like post-view and post-click duration comparison.

MNTN in Northbeam runs attribution and look-back windows of 30+ days, which should match your sales cycle and the typical CTV consideration period (help.mountain.com). Northbeam does not attribute Visits for MNTN, because CTV cannot pass UTMs.

Hold the event, eligible inventory, model, counting method and windows constant where possible. To isolate a window effect, compare reports with other fields matching and show each configured duration beside credited counts; flag other differences rather than attributing the whole change to the window.

If settings changed during the reporting period, check whether the change recalculates earlier events or applies only to new events. A report spanning such a change needs particular care.

For each post-view or post-click clock, an eligible impression or click is not mature until its full configured window has elapsed by the report cut-off. Compare cohorts only after that duration has elapsed; otherwise, the latest exposures have had less time to receive credit.

Describe the result as actions credited under the stated rules. Incremental sales require a suitable control or comparison.

More in Incrementality