Tech On Me: Can adtech wean itself off MFAs?

This week's focus 

IPG Mediabrands has released the fifth edition of The Media Responsibility Index (MRI), which now includes programmatic companies alongside social-media platforms.

In 2023, programmatic advertising accounted for over 70% of global display and video revenues, growing by 10% from the previous year. The latest MRI revealed a small performance gap between average platforms and top-performing demand-side platforms (DSPs), indicating high consistency in safety and data ethics measures.

To further enhance media responsibility, MRI 5.0 recommends that programmatic platforms avoid made-for-advertising (MFA) inventory, standardise the definition of MFA sites, remove all advertising from MFA sites and improve communication with advertisers about safety measures.

The MRI also found that social-media platforms have improved their performance over the past five years, with their average MRI score increasing by 21 points. The gap between the group average and the benchmark decreased from 16% in MRI 1.0 to 5% in MRI 5.0.

My take: MFA has been an ongoing issue for the industry and has been in the spotlight since the Association of National Advertisers (ANA) released its Programmatic Media Supply Chain Transparency Study. A few weeks ago, DoubleVerify found a 19% increase in MFA impressions year-over-year. While some advertisers have significantly reduced the number of sites they use, others still prioritise reach over quality, which can be inefficient.

Therefore, it is encouraging to see the ANA and Trustworthy Accountability Group’s TrustNet launching quarterly benchmarks to help advertisers gauge their campaigns' performance against industry standards in MFA activity, media quality, CPMs, sustainability, and DEI. This first benchmark, analysing data from 11 ANA member brands, showed that media spending on MFA sites dropped from 15% to 4%, and the average number of websites per campaign decreased from 44,000 to 23,000.

In other news

You can now ask Meta AI questions about India’s elections

Following the completion of India's election, Meta has lifted its restriction on election-related queries through its AI chatbot. Users can now inquire about election results, political figures, and officeholders. Initially, during the election period starting in April, Meta directed users to the Election Commission’s website for such queries.

In contrast, Google continues to enforce global restrictions on election-related queries, including in India, despite launching its Gemini AI app, which supports nine local languages. Due to ongoing global elections, Google's policy points users to Google Search for election-related information.

My take: Meta's approach could offer a richer user experience by providing direct answers through the chatbot, while Google prioritises safety and accuracy by directing users to search results. 

Given the high stakes of election-related information, Google's method might better manage risks related to misinformation and bias. Meta's strategy, while user-friendly post-election, must ensure it effectively handles accuracy and bias during the election period. Meta's adaptive strategy may resonate well in markets like India, where timely political information is crucial. However, Google's uniform global policy provides a standard approach that can be easier to manage at scale.

Both platforms' strategies highlight the importance of balancing user accessibility with the need to prevent misinformation, especially in politically sensitive contexts. As AI tools evolve, it will be interesting to see how these approaches are refined to serve users better while safeguarding against potential risks.

Snap takes wraps off real-time image model 

At the Augmented World Expo, Snap unveiled its augmented reality (AR) technology advancements, including a real-time, on-device image diffusion model that generates vivid AR experiences. Co-founder and CTO Bobby Murphy highlighted that the model is small enough to run on smartphones and quick enough to re-render frames in real-time based on text prompts.

Snap also launched Lens Studio 5.0, offering new generative AI tools to AR developers, significantly reducing the time required to create AR effects. These tools enable the creation of realistic ML face effects, custom stylisation, and 3D assets in minutes. Creators can also generate characters and face masks using text or image prompts through Face Mesh technology. In addition, Lens Studio 5.0 includes an AI assistant to help AR creators with their queries, aiming to streamline and enhance the AR development process.

My take: The ability to run a real-time, on-device image diffusion model on smartphones addresses the current limitations of generative AI in AR by enhancing speed and responsiveness, which are crucial for creating immersive AR experiences. 

Including an AI assistant in Lens Studio 5.0 is also a good move. It can help creators overcome technical challenges more efficiently, making the platform more user-friendly and accessible, especially for those new to AR development.

AI avatars of creators now in TikTok ads

TikTok has introduced generative AI avatars for branded content and ads, including custom avatars of creators, brand spokespersons, and pre-built stock avatars using licensed actors. This initiative aims to help creators and brands expand their reach and personalise global campaigns. In addition, TikTok is launching an ‘AI dubbing’ tool that allows content translation into 10 languages, facilitating broader audience engagement.

My take: Creators' control over using their likenesses is crucial, especially concerning copyright issues. Remember how Scarlett Johansson called out OpenAI for using her voice without permission? Allowing creators to set their rates and licensing terms helps mitigate potential exploitation and ensures fair use of their digital likenesses. I have previously covered how brands and creators can navigate and avoid copyright infringements.

Advertisers in APAC worried about fraud and generative AI

Generative AI has significantly contributed to a rise in new ad fraud schemes globally, with a 23% increase in 2023, DoubleVerify has found. APAC saw the highest fraud rates in CTV and mobile apps, with 37% of fraud occurring on mobile apps. While trending down overall in APAC, brand suitability violations rose by 14% in Southeast Asia due to intense news cycles. However, globally, a 17% decrease in brand safety violations was noted, attributed to increased use of pre-bid protections.

Fifty-four percent of APAC media buyers view the rise in low-quality content and MFA sites as disruptive. Despite high performance on isolated KPIs, these sites deliver lower overall attention, with MFA sites showing a 7% decrease in display ad attention and 28% less on video.

Sixty percent of APAC marketers believe AI-driven campaign optimisation enhances media quality. The rise of retail media networks (RMNs) has been noted, with RMNs showing significantly lower fraud rates and brand suitability violations than benchmarks.

My take: In 2023, cybersecurity researchers identified what appeared to be the first AI-powered malicious ad campaign targeting LinkedIn aimed at stealing sensitive personal information. While not a significant threat, this incident was a precursor to more sophisticated attacks. Fast-forward to the present, that prediction has come true as bad actors actively use generative AI to create fake domains and generate fake traffic. So no, Tom Hanks did not promote a dental clinic, nor did YouTube star Mr Beast launch an iPhone giveaway.

Most read Pride-related topics 

As Pride Month enters its second half, Taboola has shared data from its global network of publishers, which includes over 500 million daily active users. Australia ranks sixth globally for ‘Pride’ readership, with 71,000 page views in the past 90 days. The USA leads with 2.4 million page views.

France's Pride parade was the most read-about globally, with a 1,058% increase in readership, followed by Germany and the UK. In the US, the brands that saw the most significant increases in readership due to their 2023 Pride campaigns were Vans (31%), Madewell (34%), and The North Face (78%).

The top trending pro-LGBTQ shows in the US by pageviews were the Eurovision Song Contest with 1.1 million page views, RuPaul’s Drag Race with 212,000 views, and the dark comedy Hacks with 186,000 page views.

My take: Brands that have previously participated in Pride Month reduced their involvement in 2024, and there is increased hesitation among brands that have not yet engaged meaningfully. However, as Vans, Madewell, and The North Face have shown that consistently supporting the LGBTQ+ community can build long-term loyalty among LGBTQ+ consumers and allies. Consumers are more likely to support brands that align with their values and demonstrate genuine inclusivity.

Looking ahead

Tech MVP 2024

Tech MVP is back, this time in its fourth year. A quick recap: Tech MVP includes two lists: Most Valuable Professional and Most Valuable Product. The Most Valuable Professional list spotlights the most exceptional technical talent in the industry. It is open to individuals who hold a technical role and are central to developing technology for their business.

We seek individuals who work at agencies, platforms, publishers, tech suppliers, and startups. MVPs should possess deep technical knowledge and ability and be passionate about driving the industry forward.

The Most Valuable Product will honour the region's most technically innovative products. This list is open to businesses that operate within or provide a service related to advertising and marketing. We invite businesses to submit products developed in Asia-Pacific that demonstrate originality, scalability, and business success.

My take: Look out for the 2024 launch and the jury's announcement in the upcoming weeks. In the meantime, read up on 2023 Most Valuable Professionals Lu Kenn Wong, Raigon Jolly and Eric Yu, and 2023 Most Valuable Products Veridooh, Publicis Groupe and Vonage.