How The Adbank Ecosystem Addresses Publisher Fraud

adbank
adbank blog
4 min readMay 2, 2018

--

The following article was written by Adbank’s architect and development lead, Dr. Fodé Touré, to expand on the technical innovation behind our proprietary Publisher Reputation Score (aka “PR Score”) system and how it will breed a new standard of trust in digital advertising.

After removing the middleman, the relationship between the advertiser and the publisher becomes so simple that it turns into a game (gamification).

Thus, in addition to putting the advertiser in confidence through our fraud detection algorithm, using tools to facilitate and secure the distribution of ads, we build the reputation of the publisher through a new concept: Publisher Reputation Score (PR Score).

After each campaign, a Publisher earns points allowing them to build their reputation. The calculation of the points takes into account the current campaign and the recurrence factors (Example: if the Publisher has 3 stars in 2 consecutive campaigns then the current star number will be doubled (multiplies by two).

Thus, the equation used to calculate the number of points is as follows:

“star” is the star number received by the publisher (given by the advertiser) for the current campaign.

“factor star” is the recurrence factor related to the star number.

“model” is the performance achieved according to the model (click, impression).

“N model” is the limit according to the model (1000 for impression, 1 for click, ….).

“factor model” is the recurrence factor related to the performance achieved according to the model.

“match” is the number of perfect matches (keyword, geographical area, …).

“factor match” is the recurrence factor related to the number of perfect matches.

“N match” is the maximum number of parameters for a 100% matching.

The reputation of the Publisher also takes into account the penalties it receives. The penalty may be due to a bad report (from the advertiser) or to a blacklist appearance (from fraud detection algorithm).

The penalty is calculated as follows:

“E” is a bit (0 or 1) informing if the publisher had (or not) a penalty during the campaign (fraud detection algorithm) or at the end of the campaign (advertiser report).

“factor penalty” is the recurrence factor related to the penalty.

For the anecdote, according to a popular saying, there are 99 days for the thief and one (1) day for the owner.

With Adbank, the thief (fraudster) is caught the first day.

Hence, almost one day for the thief and all the life for the owner in the Adbank ecosystem. Thus, the reputation is built in 99 days over a period of 100 days.

361 is not a prime number. On the other hand, 361 is a semi-prime number (also called bi-prime or 2-almost-prime) because it is the product of two prime numbers that are not necessarily distinct. Indeed, 361 = 19 x 19, where 19 is a prime number and by coincidence 361 is a perfect square and 19 is its square root.

The new concept of Adbank (PR score) creates the first almost perfect publishers that tend towards perfection.

After 361, there are 4 days left for the thief. Perfect, that also coincides well because 4 is a perfect square since 2 x 2 = 4.

The variables used in the calculation of the points will be used to produce an analytical report allowing the publisher to improve the performance of their sites.

--

--