In terms of media planning – the process by which advertisers decide where, when and how often an ad should run to maximize engagement and ROI – some of the key AI use cases include its ability to create bespoke media plans within minutes and to work more efficiently by automating data analysis, targeting personalization and campaign optimization. For example, during development, one media agency conducted “human vs. machine” parallel tests, in which the agency’s AI tool and an experienced team of media planners were tasked with creating a media plan that would optimize reach over three overlapping audiences. The planners were unable to achieve the task, as each time they improved reach for one audience, they lost reach for the other two. The AI media planner, however, was apparently able to solve the problem in 90 seconds.
AI can not only identify patterns and insights that are otherwise invisible to human perception but also reduce the time, effort and agency resources needed to optimize marketing campaigns to their fullest potential. By relegating such labor-intensive tasks to AI-powered tools, advertising agencies can focus on the art of media planning – the creative thinking, intuition and strategy needed to create impactful campaigns and media placements, which AI is unable to replicate.
AI may also become a game changer in the world of programmatic buying. By using AI to analyze audience data and adjust bids in real time to reach the most valuable audiences at the most effective times, advertisers can gain more efficient, targeted and cost-effective ad placements and improved campaign performance.
The largest global agency holding companies and other media agencies have already begun unveiling their own AI platforms for media planning and media buying, and many recognized the potential of AI and how it would revolutionize the agency business model years ago. Back in 2020, Group M global chief executive Christian Juhl predicted that in five years, Group M would look “more like a software company than it will a media agency. It will automate, it will have really hard technology connections with the major media providers in the world... We will have more people doing programmatic and AI and algorithmic optimizations than we will have sending IOs [traditional insertion orders for ads].”
We are starting to see evidence of global agency holding companies investing significantly in this space, either through research and development or corporate acquisitions in order to stay ahead of the curve, including WPP’s acquisition of Satalia, an AI technology company; Publicis’s launch of its AI platform, Marcel; and Dentsu’s launch of its M1 AI platform.
Buyers beware: Advertiser concerns with automated media planning and media buying
With the undeniable advantages of using AI for media planning and buying and the increasingly widespread adoption of AI platforms and tools by advertisers and media agencies, what could possibly go wrong?
Lack of transparency, bias and confidentiality
One of the biggest risks for advertisers using such AI tools for the purposes of media planning and buying is that such tools pose a “black box problem.” Because AI models rely on intricate algorithms that are not easily understandable to humans, the inner workings of how these systems process data and generate predictions or decisions create transparency issues around choices made which could lead to potential bias and confidentiality concerns.
AI tools are only as good as the training data they’re given. If, for example, an AI tool has been trained on biased or insufficient data, this can lead to a potentially narrow, skewed view of an ad’s target audience or discriminatory practices relating to which ads are served to a target audience. The AI tool can also be trained by digesting media plans that have an inherent bias towards the agency’s or other third parties’ objectives, which, because of the “black box” nature of the tool, can remain undisclosed to the advertiser.
The training of AI can raise confidentiality concerns, particularly when AI tools are used across multiple brands and clients. For example, there is the risk that sensitive advertiser data and insights could be shared across multiple agency clients, which could lead to such information being used to benefit advertiser competitors.
Data privacy
With AI-powered media planning and buying tools relying on collecting and processing large amounts of data to learn and make predictions, including personal information, significant concerns arise about the collection, processing and storage of such data. Brands and agencies must ensure they implement robust security measures to protect collected personal information from unauthorized access, breaches or misuse.
AI-driven contextual targeting could potentially be a prominent player in a cookieless future, as it can provide an expedited way for marketers to find the most relevant places to run their ads by analyzing webpage content, keywords and other contextual signals to infer how a user may react to an ad rather than relying on (or collecting) that individual’s personal data. As such, AI-driven contextual targeting may help avoid the privacy compliance requirements associated with other targeting methods and may also be seen as less intrusive by consumers. However, only the future will tell whether the use of AI-driven contextual data will prove to be as effective for ad personalization and audience targeting purposes compared to targeted advertising based on personal data.
Ownership and advertiser data
Another large risk for advertisers to consider before agreeing to their agency’s implementation of such tools is ownership. Does the advertiser or the agency retain ownership of an automated media plan? In the event that the agency retains ownership, what happens when the same AI tool that has been trained on such advertiser media plans is then used for competing brands within the same holding company? Not only is there a risk of ceding ownership rights to media plans, but advertisers also must be wary of ceding their data (i.e., campaign data and analytics data) as agencies look to consolidate and build the largest data libraries possible for their AI platforms. This could lead to a data race that could leave the advertiser in last place unless they clearly define their ownership and access rights with the agency, particularly to their customer data. Given that the end is in sight for cookies, owning first-party data and the digital relationship with customers is more important than ever.
Brand safety and invalid traffic
For several years now, AI has been primed as a potential solution for brand safety and invalid traffic concerns. However, so far, AI tools have not been fully effective in combatting such issues. One of the reasons for this is that fraudsters continue to employ new, novel techniques that are capable of mimicking human behavior, which requires AI models to further evolve and update their training data to learn and better differentiate between legitimate and fraudulent traffic.
AI has so far not been the silver bullet that advertisers hoped for in addressing brand safety issues either. This is, in part, due to the increasing volume and complexity of online content (including nuances of the language and culture within), which make it difficult for AI algorithms to accurately determine the context of an ad’s placement, resulting in ads being placed in inappropriate or offensive environments. As such, a combination of using AI algorithms and other automated tools to identify and flag potentially harmful content, as well as human oversight to review flagged content, is likely the most effective way for brands to mitigate against potentially problematic content.
The best antidote for brands in relation to brand safety and fraud seems to be to rely on their own tags and data audits of demand-side platforms and ad server log-level information rather than rely solely on black box tools. Accordingly, advertisers will need, at least in the medium term, to continue to contract for as much access as possible to this data.
Advertiser best practices
Before using AI tools for purposes of media planning and media buying, advertisers should consider implementing the following guardrails within their agency agreements:
- Advertisers should require disclosure by their agencies of the use of AI tools, how they operate and are trained and how the agency prevents/protects against the issues identified above.
- Advertisers should require human oversight of all tools and final decisions.
- Build in a sufficient amount of advertiser control and oversight over not only the data inputs but also the instruction parameters used for the AI platform or tool to help ensure greater transparency and mitigate against potential bias issues.
- Clarify the ownership rights of automated media plans and ownership and usage rights of data fed into the AI tool.
- Build in robust terms relating to limitations as to the agency’s ability to use the same AI tools for competing brands within the same holding company.
- Continue to ensure that the advertiser has full access to campaign and impression data regardless of the sales pitch on AI tools’ effectiveness!
