Overview
With each placement Browsi reveals on the page, our engine will send across the predicted for viewability per ad request, using an ad server key value. This process allows publishers to control which demand goes into which viewability tier, and to match high viewability-focused campaigns with real-time high viewable placements, keeping your impression waste to minimum.
Introduction
Viewability has become a key metric for publishers to sell traffic, and a key metric for advertisers to measure campaign success in the context of the CPM worth investing per ad slot on the page. We at Browsi believe that viewability is contingent upon many different factors and not merely by the location of an ad slot on the page. We have developed a machine learning algorithm that can predict the likelihood for a spot on the page to get viewed by a single user in real time before fetching demand for that slot. We now supply publishers with the option of using the predicted viewability and connecting it to your ad server and letting that metric decide which demand, if any, should fill that newfound ad placement.
How it works
Browsi’s engine finds the best location to place an ad in real time for each and every article published.
Browsi analyzes different types of data per each distinct article:
The page layout and elements
Real-time user engagement
Viewability and other ad-related data.
Browsi’s engine uses its machine learning algorithm and the data listed above to understand the likelihood of each and every placement to be viewed, before sending a call to the ad server.
Browsi’s engine then executes a call to the publisher ad server to fetch an ad to the page. When doing so, the engine will send over – using a dedicated key-value pair – the predicted likelihood for the location to be viewable by the user.
In the ad server, the publisher will choose which line items will target a specific viewability tier.
Integration with the publisher
Integration with this feature is simple, and requires only basic knowledge of your ad server’s key value pairs features, and targeting them in line items:
Initial integration
Browsi will supply the publisher with a tag that gets our engine to the page.
Our engine requires running for 3 days on analysis mode on the entire publisher traffic.
After running analysis mode, Browsi will collect all necessary data to build a custom machine learning algorithm that best matches the publisher (models will be built per site).
Ad server integration
After the initial integration, Browsi can start sending ad requests to the publisher’s ad server.
With each ad request that is sent by Browsi’s engine from the page to the publisher ad server, Browsi will add a key called “browsiViewability”
That key will be populated with the appropriate prediction value including the likelihood of a viewable impression.
The value will be a decimal value between 0.00-1.00.
Two numbers after the decimal point (depending on the tiers, see below).
0 means a low chance for a viewable impression.
1 means a high chance for a viewable impression.
The publisher sets up and determines the viewability tiers in ad server.
Option 1 – each possible prediction value (0.01, 0.02…..0.98, 0.99).
Option 2 – gap of 0.05 between prediction values (0.05, 0.10, 0.15….0.90, 0.95).
Option 3 – gap of 0.10 between prediction values (0.10, 0.20, 0.30….0.80, 0.90).
Creating a hybrid model (Changing gaps).
The publisher will use the tiers to target different line items according to his demand and advertisers request.