3 ways artificial intelligence will save adtech
As the adtech ecosystem continues to grow and evolve, we can expect to see advertisers utilizing artificial intelligence (AI) and machine learning (ML) in three essential ways. In fact, the very future of adtech may depend upon it. Read on...
AI will play a pivotal role in helping advertisers achieve their KPIs
1. AI will develop cookieless lookalike models
Assuming Chrome does not push back its removal of cookies, 2023 will likely be the end of the already constrained cookie-based targeting era. Growing solutions such as universal IDs and first-party publisher data show promise, but the reality is that the adtech industry will also need to refine the ability to target and measure performance in the total absence of user-level signals – a task AI can help with.
One of the more promising solutions for this future lies in AI’s ability to create lookalike models for brands based on smaller sets of known users that do perform well. In some ways, AI has already been creating lookalike modeling for cookie-based targeting, just using a different set of audience data that will soon be much more scarce such as demographics, behaviors and interests. The good news is there are many privacy-compliant signals available including the type of website or app, geography, type of device, time of day, local weather, dominant political or other attributes of the region, keywords on the page, sentiment of the page and time on page.
In the future, AI will allow brands to learn about small sets of data available through user-level signals, then extrapolate based on non-user level signals to find larger target segments. For instance, AI may learn that your brand performs best on Android users in urban settings in the evening hours on content related to food and sports, then automatically adjust the programmatic buying filters to focus spend on those users.
This future is inevitable, but the question is what part of the adtech ecosystem will develop this type of AI? Historically, most targeting, and certainly most audience targeting, happened at the DSP layer or through DMPs that were integrated with the DSP. More recently, however, SSPs are getting involved in integrating first- and second-party audience data. Whoever holds the audience has more durable value, so the door is open for DSPs, DMPs and SSPs to develop the most sophisticated and performant lookalike models.
2. AI will improve supply path optimization
Supply path optimization is and will continue to be an important tool for ad buyers to improve performance by buying through the most performant path. Today, this process generally involves a manual analysis of tons of quantitative and qualitative data points from SSPs and publishers that ultimately results in an advertiser or ad buying group selecting a handful of its top preferred SSPs.
The reality is that exchanges are not a monolith of quality or lack thereof. Advertisers choosing a big exchange cannot assume that every path will therefore be the best path for their goals. Moving beyond the exchange-level to each unique supply path, however, is too complicated to be analyzed by hand and can ignore all sorts of important but complex data signals that quickly reach a data volume that is nearly impossible for humans alone to analyze.
The future will likely include a mix of manual and AI optimization where advertisers choose a shortlist of exchanges based on broad qualitative and economic factors, and then rely on AI to optimize at a more detailed path level, targeting their goals for price, performance, fraud and more.
For example, if an advertiser is bidding on a banner on CNN.com, when done manually, an advertiser might opt for the exchange with the most efficient CPM. What AI can improve on is predicting that the same banner served through exchange B will have a slightly higher CPM, but will result in a lower CPC or CPA because of improved rendering or other enhancements to the banner that historically improved performance.
For video, an advertiser might have a select few exchange paths approved to transact video on CNN.com. However, the current state will not necessarily distinguish between an instream ad that plays before a user-selected CNN video and a similar-looking impression that actually plays on mute in the bottom corner of the page. AI will be able to decipher those paths and either remove the lesser performant option or at least reduce the CPM bid.
Both DSPs and other parties such as Jounce Media are already working toward paying attention to and optimizing based on deep SPO signals, but we have much room to grow here and the role of AI is just getting started.
3. AI will supercharge dynamic creative optimization
Currently, running a dynamic creative optimization (DCO) campaign is a conscious decision and tech partners are isolated to a few DCO companies. Typically, a decision tree is mutually agreed upon and the creative agency provides a few creative parameters from which the DCO company can optimize toward. In the not-too-distant future, AI will improve the sophistication of this process by ingesting even more potential optimization parameters and creative options to automatically generate an ad for any given impression.
Instead of making a conscious decision to ‘run a DCO campaign,’ this type of decision-making process could eventually be built directly into DSPs so that all campaigns have at least some elements of DCO. When setting up a campaign, advertisers would upload components into a DSP and, like a self-driving car, let the DSP take over choosing the right creative for each impression.
From a creative standpoint, advertisers could upload a variety of images, color palettes, slogans and headlines, and AI could even develop similar assets the way Netflix uses AI to develop and test multiple thumbnails per title. AI could also construct headlines for advertisers using a set of keywords and historic performance.
Then the AI will discover and predict which creative combination will perform best based on the parameters available at each impression call and continually learn, adapt and improve the performance of a campaign.
This is a future that some enhanced SSPs are already exploring. By finding the correct performing headlines or captions and enhancing assets with those headlines to match the look of the site, the parameters by which AI can optimize are expanded.
The promise of AI is not too far away. In the not-too-distant future of adtech, AI will play a pivotal role in helping advertisers achieve their KPIs by creating a privacy-centric, regulatory compliant lookalike model for a larger set of users. Advertisers can save money, reduce media waste and optimize supply paths at a level of efficiency and detail impossible for humans, further improving overall ROI. AI can also automatically generate the right ad for any given impression, freeing up resources for advertisers to work on the next campaign.
Curt Larson is chief product officer at Sharethrough.