品牌安全手册
下面是小编为大家整理的品牌安全手册（2022年）,供大家参考。
WHAT’S IN
THIS
PL A YBOOK?
INTRODUCTION
To build brand love, you need brand safety 10 key takeaways
BRAND SAFETY
TODAY
Third-party cookies Third-party measurement Deepfakes Transparency Too much brand safety?
MEDIA
CHECK-LISTS Connected TV Digital Out-of-Home Location Data Audio Gaming
CONCLUSION 2 2 3
5
6 7 8 9 10-11
12
13-15 16-18 19-20 21-23 24-26
27
4
T O
A Y
BS AR AF ENTDY
AN OV E RVIEW
OF
PO L ITICAL, SOCIAL
AND TECHNOLOGICAL SHIFTS
CURRENTLY
IMPACTING
BRAND
SAFETY
AT
A
MACRO LEVEL
cookies measurement deepfakesaudio gaming digitaltv
THIRD-PARTY COOKIES
ARE
BEING DEPRECATED
The EU’s General Data Protection Regulation (GDPR) came into force in 2018 and, while it disrupted the digital advertising industry, many believe that the true disruption may be provoked by the yet-to-be-adopted ePrivacy Regulation.
The regulation foresees — in some of its forms — default blocking of all third-party cookies at the browser level and establishes browsers as the ultimate consent management platforms. Anticipating this regulatory development and motivated by privacy considerations and changes in user behavior, Firefox and Safari started blocking third-party cookies a year ago and Google announced Chrome would join the choir in two years’ time. The U.S. is anticipating its own federal privacy regulation, propelled by the California Consumer Privacy Act (CCPA).
The demise of third-party cookies, though, is due to the regulatory pressure the industry
is facing, starting in Europe where ad tech has been subject to regulatory guidance or deep-dive investigation. It is also an inevitable technological development driven by a change in consumer habits and technology adoption (moving toward smartphones and other connected devices).
But, while the death of the third-party cookie has been getting a lot public attention, it is
only part of a broader transformation the industry is going through. New capabilities will open new advertising channels and data sets for advertisers. With new rules on how data is used and how ads are delivered and measured, execution and measurement of primary media investments will also change.
The industry has already started looking into different solutions to continue to reach the right consumers and optimize media spend. Many of them rely on more privacy-friendly solutions like “clean room environments” (secure environments in which no granular data access is possible) or innovative machine learning techniques like federated learning (applying machine learning models to consumer data without leaving their device).
These new methods, although imperfect, would allow advertisers to receive aggregate reporting about campaign performance while complying with the relevant regulations.
6
AND
INDEPENDENT
THIRD- PARTY MEASUREMENT
MAY BE
ON
ITS WAY
OUT,
TOO
A potential pitfall may be the downfall of independent, third-party measurement. GroupM, however, remains committed to privacy-enabled independent measurement and will continue working closely with our partners to address this issue. Not only is this important for brand safety, but it’s also necessary to uphold simple, yet crucial, transparency about what inventory advertisers bought and how their campaigns performed.
Simply put, if measurement companies cannot access data directly and can only gain access
to already-processed aggregate data from the publishers themselves, measurement may stop being impartial. While it may appear that this problem is limited to audience verification, content verification
may not be exempt if content categories or URLs are seen to be revealing personal data.
Complying with new data protection regulations is complicated by companies across Europe having to navigate confusing guidance and conflicting
interpretation from Data Protection Authorities. This not only yields additional compliance costs for companies, but the different levels of protection will be confusing for consumers. Finally, it would be misleading to assume that data protection compliance becomes easier as third-party cookies disappear. It is important to recognize that data processing is not limited to cookies.
The deprecation of third-party cookies is an opportunity for the industry to forge a new
and more collaborative privacy-enabled future. To do that, the industry must find
a new
way to steer the right advertising to the right people, moving away from existing identifiers and creating new, privacy-friendly ways to advertise.
7
FAKES
ARE
PROLIFERATING, AND
GROWING
MORE SOPHISTICATED The trend of online disinformation, otherwise known as fake news, has been dominating news headlines since 2016 with no signs of slowing down. The online disinformation actors are particularly active during election periods and, as we enter into the final
stages of the U.S. presidential campaign, we are likely
to see more of it. The fear and uncertainty stemming from the COVID-19 pandemic has also paved the way for bad actors to promote and benefit
from online disinformation.
The industry has played its part by devising many solutions, including new artificial intelligence-driven technologies to recognize fake news and by establishing partnerships with fact-checkers and journalists. But now we face a new and disturbing danger emerging from the depths of the dark web: deepfakes. Through the use of machine learning technology, pictures, videos and voice recordings of a person can be ingested and then, most often fraudulently, manipulated into a deepfake video or audio recording.
Apart from obvious misuse for political purposes, deepfakes can be used to misrepresent senior representatives of any company making public statements that could not only influence
a share price, but erode any other communications effort deployed by the company. Disinformation in general can also be used to spread harmful rumors about brands that serve to damage their image. 8
COMPUTER
VISION, BLOCKCHAIN
&
BIOMETRICS MAY
TRANSFORM TRANSPARENCY The promise and appeal of new technologies to help us address inappropriate adjacencies, fraud or transparency remain strong. While the artificial
intelligence models are successful in helping to identify IP-infringing content or spam, identifying online disinformation or hate speech remains a challenge. Human contempt has many nuances that have eluded technology up to now and detection relies heavily on human review.
As technology evolves, we are hopeful that nuances and human bias will become more identifiable.
Machine learning has driven a revolution in content analysis, evolving the practice from simple keyword avoidance to sophisticated sentiment and semantic analysis. Still, most video content evaluation is conducted via metadata and audio file
analysis. In 2020, true computer vision that can see images and process video exists only within bespoke video players and the biggest social platforms. Even then, it cannot yet be processed at adequate scale and speed.
On a different note, there is early evidence
that blockchain can help deliver better transparency to ad placement and distribution of ad spending. Tests run in the U.K. and the U.S. promise to deliver the first
blockchain- powered insights into media buys.
Voice- or facial-recognition verification technology will be discussed and tested in months and years to come that will lead to a whole new set of ethical and regulatory debates about processing of biometric data, even just to fight
fraud.
CONTENT
ANALYSIS
:
STEADY
PROGRESS
NLP:
Natural
Language
Processing
9
THERE
IS SUCH A
THING
AS TOO MUCH
BRAND SAFETY
FOR NEWS
PUBLISHERS
For the past several years, global marketers have been focused on reducing their brand safety risk by adopting a range of policies, tools and tactics. There is now a concern that some advertisers have taken excessive brand safety measures, which may be negatively impacting reach, performance and the health of diverse, newsworthy, legitimate ad-supported environments.
During the coronavirus pandemic, many advertisers want to avoid being associated with any related content, whether positive or negative, and publishers have been vocal about the signi fi cant demonetization of news as a result.
Such revenue losses should be put in the right context. For over a decade, algorithmically driven social platforms have led to a mass exodus of advertising revenue from the pages of many online and print publications. And, as people continue to use social platforms as virtual newsrooms, advertising investment follows them. This has severely impacted news publishers’ ability to afford to place sufficient journalists on the ground, especially in local communities. Many have been forced to fold up shop, merge with other publications, or severely reduce their reporting staff.
Aggressive use of avoidance technology has taken much of the blame for news demonetization. This technology ensures that when ads are bought programmatically, there are parameters in place to ensure brand safety because adjacency to controversial news is an emotionally charged issue and marketers, understandably, want to protect their brands. Since the majority of stories in credible publications are focused on the pandemic during the crisis, semantic avoidance practices have led to decreased revenue for publishers
at the time they are most needed. This led to a GroupM brand safety recommendation for clients to limit COVID-19 avoidance to negative associations, like death counts, mortality rates, etc. By focusing on positive
associations, we can support credible news publishers and deliver quality reach to our clients. GroupM has long advised clients against any blanket blocking and has worked with industry bodies, such as IAB U.K., to provide best practice guidance on how to use avoidance technologies appropriately.
Semantic avoidance technology is not the enemy. These technologies can actually make
it easier for advertisers to support news by enabling each marketer to identify their own sense of brand suitability. The issue is how thi...