Insights

The Path to Cross-Screen Measurement

James Rothwell VP, Marketing, Comcast Advertising

Introduction

Cross-screen measurement is vital for advertisers to understand the impact of their campaigns, but it poses many obstacles. True cross-screen measurement relies on multiple data sets –panels, census data, and attribution – to be accurate, trustworthy and complete. By using a combination of these data sets and strategically leveraging the strengths of each, advertisers can gain a more accurate view of media consumption across individuals and devices, as well as better understand how cross-platform media investments drive business outcomes.

The good news is that progress is being made, especially by publishers and operators where there are deterministic datasets and capabilities to map cross-screen exposure. The area of focus for the industry to enable marketers to have a cross-publisher/operator view is around the data rights to support broader measurement and frequency management.

What follows is an explanation of the key unique data sets, their pros and cons, and how a combination of all three will ultimately drive true cross-screen measurement.

Scroll down to learn more or download the full report.

Click or Hover On the Charts Below to Interact


Three Valuable Data Types

Panels

Panels have been used historically to measure TV viewing across a National representative sample, and have supported decades of advertising trading, focused on traditional age- and gender-defined audience segments.

Census

Census data sets such as set-top box data and ACR are less representative but their large scale enables a more accurate view of increasingly fragmented viewing behavior and a broader set of audience segments.

Attribution

Attribution methodologies focus on tying media exposure to business outcomes and enabling marketers to better understand ROI and optimize current and future media plans to maximize results.


 

PANEL PROS

  • History / acceptance
  • Single source in terms of breadth of data collection of HH media consumption
  • Designed for national representativeness
  • Measures individual-level viewing

PANEL CONS

  • Expensive per HH (to recruit and maintain)
  • Not statistically reliable for more granular viewing, niche audience segments, and hyper-local audiences
  • Panel churn impact on time-based trends
  • Cross-device viewing is a challenge
  • Audience definition disparities exist between measurement and targeting providers

CENSUS PROS

  • Large-scale media consumption data
  • Efficient per-HH costs
  • Enables measurement of more granular audience viewing and / or segments

CENSUS CONS

  • Expensive to license needed data sets
  • HH- or device-level measurement (but no individual-level viewing)
  • Must weight data to achieve coverage and representativeness
  • Data rights / restrictions may require modeling or aggregated reporting

ATTRIBUTION PROS

  • Alignment of media consumption and action
  • Scalable in terms of cross-screen activity
  • Optimization opportunities mid-flight
  • Opportunities to model / validate business impact relative to media measurement proxies (incremental reach / effective frequency)

ATTRIBUTION CONS

  • Multi-touch attribution is complex, expensive
  • Cross-device data may be incomplete (no common ID)
  • Methodologies are inconsistent across providers
  • Historical performance may not be reliable future signal
  • Industry experts often disagree on attribution model application (i.e. which model is best in a given scenario)

 

Note: Companies with 100+ employees that use at least one digital marketing channel; an attribution model is a way to differentiate the respective contributions of various marketing channels to a desired outcome; includes first- and last-touchpoint models and more complex multichannel models; multichannel attribution models are attribution models capable of attributing marketing credit to more than one marketing channel or touchpoint to differentiate the respective contributions of various marketing channels to a desired outcome; models can include both digital and nondigital channels and touchpoints

Source: eMarketer, December 2020


Applying Data for Better Measurement

 

The path to true cross-screen measurement is not an either / or story.
It is an “AND.”


No one standalone data set can provide any single marketer a true view of how media was consumed across all individuals and all devices at any moment in time, or how advertising then impacted outcomes as a result of exposure across that media.


That means the available data sets need to work together, reference and validate each other and often be merged to provide a fuller picture.


Alignment between attribution modeling and panel-plus-census-data-based media measurement proxies can also help to be more efficient in terms of costs and ongoing analysis so advertisers can make smarter investment decisions, optimize reach and frequency and drive results.


In short, the path to cross-screen measurement involves more sophisticated and expansive data sets and modeling techniques, using multiple mechanisms and methodologies to get closer to the truth.