Ad Graph is our way of processing, storing, and building new ways of analyzing the world’s advertising data. Ad Graph powers our models and translates the raw data we capture into actionable insights for our customers.
To deliver the data we use a combination of:
Historical Data
Since 2012, we’ve collected millions of digital ads across thousands of web pages. This data powers our underlying models to feed our automated processes.
Customer Feedback
Our clients include thousands of users that help power our Ad Graph with their digital marketing data.
Bidder Intelligence
Real time bidding technology informs our CPM models across the globe.
Industry Research
We constantly tune our data capture techniques and CPM calculations to reflect the state of the market. We incorporate data from public financial data, industry reports, and analyst estimates.
Data Partnerships
We partner with industry leaders to access and incorporate their estimates for CPM and impressions into our machine learning models.
The process starts when Pathmatics collects digital ads from the web.
We rely on both panels and crawlers to collect a statistically representative sample dataset. Together the two methods enable us to provide the most robust picture of the digital advertising landscape.
Devices & formats collected:
Desktop & mobile display
Desktop & mobile video
Facebook
Instagram
Twitter
Youtube
Regions covered:
United States
Australia
Canada
Germany
New Zealand
United Kingdom
During the enrichment process we transform the raw data into insights.
On a daily basis, we apply patented machine learning models to millions of records:
Ad Analysis
We record, analyze, and organize the raw data collected for further analysis and modeling.
Audience Modeling
We weight our data sample to adjust for any bias in the sample. We scale to estimate impressions.
CPM Prediction
We estimate CPM rates for each creative. We tune our CPM estimates for size, position, time, type, and marketplace dynamics.
Category Automation
We have trained hundreds of thousands of machine learning models that identify and categorize advertisers and brands.
Inventory Scoring
We analyze the type of inventory from publishers to review any potential brand quality violations.
Final Estimations & QA
All data comes together for final modeling, spend estimates, and testing before final insights are delivered to end users.
Finally, we report the data and insights and help you answer who, what, when, where, why, and how.
See individual brands, compare multiple brands, or explore entire categories.
Creative Strategies
See messaging, CTA, sizes, audience profiles,, and flighting for every creative
Device Strategies
Track and compare impressions and spend across all devices & formats
Targeting Strategies
Understand Facebook audience insights across gender, geo, and OS
Buying Strategies
See how and where advertisers are buying
Share of Voice
Gain visibility of advertiser share of voice
Trends
See historical spend and impression for advertisers over time
