How It Works

We personally analyze 16,000 features across hundreds of thousands of websites and apps to give you deep insight into publisher quality.

 

Publisher quality is complex, and our rating system is designed to give you insight that is calibrated to reflect your precise needs. Depending on the campaign, the brand, and the users you are targeting, a publisher’s quality rating can be variable. 


Via our hands-on editorial approach coupled with our proprietary data and statistical modeling techniques, we deliver an analysis of tens of thousands of publishers with ratings that are customized to you.

OUR RATING SYSTEM

Our composite scores are broken into three shield ratings: White, Gray, and Black:

WHITE
High Quality

GRaY
Variable Quality

BLACK
Poor Quality

We arrive at our shield ratings through a combination of each publisher’s overall quality and overall safety score.


OVERALL QUALITY SCORE

Our rating system determines true quality by going beyond the basics. We look into the deepest layers of a website to determine not just what is there, but how it is arranged and what it may actually look like to a consumer – a level of detail no one else has even thought to pursue.

Category Fit

Safety Flags

CONTENT BREADTH

Site Design & LAYOUT

Which all contribute to a publisher’s overall QUALITY SCORE:


Safety Scores

Our rating system determines true safety by thinking about it in a human, contextual way. For example, profanity may not be important to you, so long as the publisher is high quality with deep expertise in a specific category. We understand that safety is relative so we adjust our analysis based on a brand's unique needs.

Some of the SAFETY FEATURES we take into account:

Hate

Danger_hate.png

Ad Clutter

Dangerous Content

User Generated Content

Which contribute to a publisher’s overall SAFETY SCORE:

safety_meter.png

Still want more?

Click the buttons below to download summaries of our standard ratings.


MORE ABOUT THE SCIENCE BEHIND OUR RATINGS...

TRUST METRICS MERGES HUMAN AND MACHINE INTELLIGENCE

Ultimately, it's human-powered editorial expertise that distinguishes Trust Metric's data and ratings, and what ensures that our customers can rest easy knowing that their ads are running in appropriate environments.

So the problem we've solved is how to use computation to augment and extend our experts' judgements. To do so, we use an army of computers to collect information about the places our customers are considering advertisting. We analyze millions of websites, web pages and apps, and automatically extract more than 10,000 features that can be used to discriminate the good from the bad and the ugly.

We then apply advanced statistical and machine learning techniques to score target environments along a set of models hand tuned to inform the human scores that we share with our customers.

This frees our experts from the drudgery of reading a million (mostly terrible) web pages a day, and lets our customers know immediately which advertising environments suit their tactical and brand safety goals.