By: Lauren Douglass, VP Marketing, North America at Teads
Artificial Intelligence has become an integral part of our daily lives, whether we’re aware of it or not. For consumers, AI helps from outright utility such as speech recognition and chatbots to subtler applications such as smart traffic lights or estimated wait times, AI has an invisible hand in near everything we do, and the advertising world is no exception. Advertisers are now using AI to personalize offers, optimize media spends, and better segment and target audiences, and the potentiality of what can be done seems limitless. For an industry that has long demanded that the customer be known—sometimes even before the product is—machine learning presents a wildly employable opportunity to glean information on audiences and channel those findings back into improving returns.
We’ve already covered how AI is being used to structure the vast amount of data being collected through the clicks, swipes, and taps that smart devices and IoT yield, as well as the advances in hardware and computing that have allowed this to happen; we’ve also touched on how digital advertising is applying these findings. Artificial Intelligence is such a gift to the industry that even the slightest integration can produce results, but to truly take advantage of the possibilities, the proper algorithms must be developed and deployed, with no data left for scraps. This is where the Teads approach comes in.
The Teads Approach
At Teads, we deliver ads to more than 1.5 billion people a month across a wide-ranging inventory from the world’s best publishers. With that quantity of data generated on a daily basis, we have a unique opportunity to move beyond the simple CPM model and utilize predictive advertising to deliver actionable insights and business results.
Using a sophisticated, adaptable AI framework that learns and improves with experience, we can feed in what we know about a specific user, advertiser, publisher, or context to create robust recommendations for the targeted audience. From simple user features such as gender or affluence to contextual features such as the browser being used or the time of day; from the category of the publisher’s website to the creative of the ad itself; we factor more than 50,000 variables into the algorithm that decides who will see an ad and when, all with the same goal in mind: achieving pre-defined marketing objectives. From completed views in upper-funnel to site-visits that drive consideration.
How Our Predictive Advertising Works
When it comes to ads that generate results, we’ve done this before—billions of times before. We have a tremendous number of examples showing what has worked, when it has worked, and who it has worked with. In order to fully harness such a large amount of data, we leverage machine learning to extract optimized combinations of targeted impressions that are most likely to deliver outcomes: each ad served is a business case that can be assigned a value, and each future ad has a desired value based on outcomes such as conversions or engagement. Our framework takes a huge number of examples of the desired value that occurred in the past and analyzes the variables involved in order to find which combination of features are the best to predict success in the future.
Our commitment to driving guaranteed outcomes for clients is made possible through an array of Teads proprietary models and technology:
1. Guaranteed Outcomes. To drive cost-effective and valuable campaigns, we use predictive AI to serve ads taking into consideration dimensions like browser, device, user data, publisher etc in order to best predict the likelihood of accomplishing a KPI. To ensure incrementality, our proprietary Ghost Ad technology measures queries generated from ads shown to the target audience, and queries generated from the same target audience who weren’t shown an ad.
2. Lookalike Audience Targeting. Our unique view of premium media content consumption allows us to find real specificities and patterns. We have seen that within a given segment seed, people are often consuming similar type of content. Our lookalike algorithm uses this fact to find relevant statistical twins. When a preferred audience is too small, we extend the reach and maximize relevancy by finding users that look like users present in the seed list. Using a 360° view of the users in our engine, we allow the discovery and identification of similar users (statistical twins) via models built on a daily basis. Our machine learning algorithm helps us:
- Gather browsing history data across our publisher network, including time of visit and type of visited pages (i.e. sports, politics)
- Regroup users with similar browsing behavior into 300 clusters using a non-supervised AI algorithm
- Evaluate each cluster to find where the advertiser’s targeted seed users are represented with a high percentage, indicating the cluster contains similar users and thus creating a lookalike audience.
3. Dynamic Creative Optimization. With creative powered by AI, different versions of creative from the same campaign are automatically generated based on dynamic signals that determine what would work best in the given context. Using predetermined inputs (such as the weather, the time of day, or the audience segment), the creative generation engine can change things like the template and format type and the text and positioning of the CTA in order to personalize content to the user’s context. That way, a car shopper in Florida doesn’t have to watch how a car handles in snow, and a potential vacationer can see what tickets cost out of their home airport.
Better Outcomes for Your Brand
Every brand has its own business requirements and aims, and there’s no one turnkey solution that can satisfy them all. What Teads can provide is an evolving, full funnel solution that’s tailored to a brand’s marketing objectives, even as priorities and audiences shift. Our predictive AI leverages data to deliver personalized ads and better outcomes and to forecast our audience’s future behaviors. Rigorous testing and real-time analytics allow us to test whether audiences will engage with an ad, delivering unique KPIs for the brand. Combined with our performance-based pricing and cost-per-completed-view (CPCV) programmatic buying platform, the Teads approach focuses on giving advertisers the confidence that every cent of a digital media spend maximizes its potential return.