Browse Prior Art Database

Method and System for Detecting Smart Cheating in Advertising Campaigns

IP.com Disclosure Number: IPCOM000244255D
Publication Date: 2015-Nov-26
Document File: 4 page(s) / 73K

Publishing Venue

The IP.com Prior Art Database

Related People

Pengyuan Wang: INVENTOR [+5]

Abstract

A method and system for detecting smart cheating in advertising campaigns is disclosed The method and system detects the smart cheating by computing a Smart Cheating Ratio SCR with confidence intervals The sign of the SCR identifies whether there is smart cheating and the absolute value identifies the severity of the smart cheating

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 37% of the total text.

Method and System for Detecting Smart Cheating in Advertising Campaigns

Abstract

A method and system for detecting smart cheating in advertising campaigns is disclosed.  The method and system detects the smart cheating by computing a Smart Cheating Ratio (SCR) with confidence intervals. The sign of the SCR identifies whether there is smart cheating and the absolute value identifies the severity of the smart cheating.

Description

Generally, advertisers like to be aware that money is efficiently spent on an advertising campaign or specific targeting plan of the advertisers.  For example, the advertisers require information regarding number of extra conversions, purchases, or brand searches yielded by the advertising campaign or the specific targeting plan comparing to the situation where no ad campaign was running.  Since the overall conversion rate does not discount the conversions that happen without advertising campaigns, the information provided to the advertisers may not be accurate.  The advertising campaigns may perform smart cheating such as targeting users who will convert even without advertising campaigns.  In such cases, the advertising campaigns and the targeting plans are not efficient in yielding extra conversions.  Due to the smart cheating, the advertisers may wish to distinguish the advertising campaigns that cheat from ones yielding real extra conversions.  Since, there is no metric that can quantify the distinction, the advertisers cannot compare the campaigns objectively.

Disclosed is a method and system for detecting smart cheating in advertising campaigns.  The method and system detects the smart cheating by computing a Smart Cheating Ratio (SCR) with confidence intervals.  Sign of the SCR identifies whether there is smart cheating and the absolute value identifies severity of the smart cheating.

The method and system quantifies the smart cheating based on the difference of a naive amplifier which may have been inflated by smart cheating and a true amplifier with smart cheating effect eliminated.  The naive amplifier is formally defined as follows.  In an observational dataset, suppose  users indexed by  are observed, each with a characteristic vector with  elements .  A user is either exposed to advertisements or not.  The exposure indicator is represented by , where  indicates exposures and  indicates no exposure.  Hence, the users are divided into two groups: the exposed group and control (i.e., unexposed) group.  The user-level outcomes, for example purchases, are indicated by , where  indicates success action(s) and  indicates no success action.  Naive success rates of the exposed and control groups are  and  respectively.  The naive amplifier is simply the ratio .

However, the naive amplifier is biased since the control group may have different user characteristics than the exposed group.  To obtain the true amplifier on the exposed users, the control group is matched to the exposed group,...