"機器人欺詐"給廣告業帶來鉅額損失

Marketers will lose $6bn next year to fraudsters who are using networks of robots to exploit the online advertising industry, according to the largest study to date of digital ad crime.

一項對數字廣告犯罪行爲迄今最大規模的研究表明，明年，使用機器人“觀看”廣告的欺詐行爲，會給廣告商帶來60億美元損失。

The Association of National Advertisers, whose members spend more than $250bn on marketing each year, conducted the study with White Ops, an ad fraud detection company.

這項研究由美國全國廣告商協會(Association of National Advertisers，簡稱ANA)與廣告欺詐偵探所White Ops共同開展。ANA旗下成員每年用於市場營銷的開支超過2500億美元。

The US association found that a quarter of online video ads were “viewed” by bots — automated computer programmes that are remotely operated by fraudsters — rather than human beings.

這家美國行會發現，四分之一的在線視頻廣告是由機器人而不是真人“觀看”的。這裏的機器人是指那種由欺詐者遠程操縱的自動化電腦程序。

The study, which analysed 5.5bn ad impressions, found that one in 10 display ads were viewed by bots rather than real people.

這項研究分析了55億次廣告投放，發現每10個顯示廣告中就有一個是由機器人而不是真人觀看的。

Ford, Pfizer and Nestlé were among the brands that took part in the investigation, which provides the clearest sign yet that fraud is rife in the fast-growing online advertising market.

參與此次調查的包括福特(Ford)、輝瑞(Pfizer)和雀巢(Nestlé)等品牌。這次調查得出迄今最明確的結果，即在快速增長的在線廣告市場中，欺詐現象十分普遍。

To take advantage of marketers, fraudsters create software that mimics the cursor movements and mouse clicks of humans, giving the impression that a person is visiting a website.

爲了欺騙廣告商，欺詐者設計了能模仿人類操控下移動光標和點擊鼠標的軟件，從而留下有人在訪問該網站的印象。

After installing the software on their own servers or on a hijacked computer, the fraudsters are able to direct the bots to visit certain websites. This allows them to profit from selling ads on their own sites or by driving fake traffic to a site of an affiliate.

在將軟件安裝在自己的服務器或被挾持的電腦上之後，欺詐者就能操縱這些機器人訪問特定網站。這讓他們可以通過在自己網站上銷售廣告，或將虛假流量導向關聯公司網站獲利。

“A bot can look like a sports fan, someone with a six-figure income, someone interested in buying a car, or a grandparent looking for holiday gifts for grandchildren,” the report said.

報告稱：“機器人的行爲可以表現得像體育愛好者、收入6位數的人、想買車的人、或爲孫子孫女尋找假日禮品的祖父母。”

White Ops and the ANA worked with 36 of the association’s member organisations to analyse digital advertising campaign traffic between August 1 and September 30.

在報告中，White Ops和ANA與該協會36家成員機構合作，分析了今年8月1日到9月30日之間的線上廣告活動的流量。

The researchers inserted special software into the ad units of the advertisers and then applied proprietary analysis techniques, alongside data from third parties, to identify bots. Marketers rarely conduct such a detailed level of forensic analysis of their online advertising campaigns.

研究人員在廣告商的廣告單元中植入特殊軟件，再使用獨家分析技術，並綜合第三方數據來辨認機器人。相比之下，廣告商很少採取如此細化的方式，如此詳盡地分析其在線廣告活動的數據。

The ANA estimated that advertisers will lose approximately $6.3bn globally to bots in 2015, after applying the bot levels observed across its study to the estimated $40bn spent globally on display ads and the estimated $8.3bn spent globally on video ads.

在把研究中觀察到的機器人比例推廣到全球的顯示廣告和視頻廣告之後，ANA估計，2015年全球廣告商因機器人而遭遇的損失約爲63億美元。目前，全球用於顯示廣告的開支估計爲400億美元，用於視頻廣告的開支估計爲83億美元。