Purpose
This article compares InnovidXP's approach to our three key methodologies: Spike linear model, impression-based linear, and CTV/OTT.
| Model | Spike Model | Impression-based Model | CTV / OTT |
| Description | 1 spot aired on live TV, with 1 million viewers watching simultaneously. |
1 spot aired on live TV with viewers viewing at the same time. Innovid takes a unified approach, combining total response measurement with household-level impression data. Innovid uses total response measurement to ensure complete coverage of all responses, then applies impression-level data to cross-validate and isolate the precise uplift driven by a TV campaign. |
1 impression sent to 1 million individual households, each viewing it at a different time. |
|
Methodology |
Establishes baseline web or app activity, then isolates post-ad traffic surges and applies a probabilistic score to each session within a brief response window, accurately attributing immediate online conversions to specific linear TV spots. | Connects household-level viewing data from smart TVs, ACR, and publisher logs to online responses over a multi-day window (typically up to 7 days), using a time-decay algorithm to measure both immediate and long-term campaign impact. Utilizes 1:1 mapping with the Vizio/Inscape panel and applies an upscale factor for greater accuracy. |
Direct 1:1 impression-tracking tags and cross-device household IP mapping to precisely link streaming ad exposures to digital conversions on any device within the same household. |
| Image | |||
|
Attribution window |
Up to 15 minutes after the spot aired. | Up to 7 days after the impression was delivered. | |
| Response data | Website first-party pixel or app integration. | ||
| Delivery data | Spot data: post log file. | Impression data: pixel on ad server. | |