AI gets a lot of buzz, and like any hot trend, it can get reduced in people’s minds to only that – a trendy buzzword. Don’t let all the marketing pitches take away from the revolutionary effect this technology is having on the ad-tech industry though. AI is here to stay, and it is helping companies all through the supply chain optimize the way they do business, cutting costs, boosting speed and improving results.
More than a year ago, Digiday wrote about how DSPs are using AI for supply path optimization, publishers for inventory tagging, agencies to organize data and dynamically optimize creatives. At the time, many still viewed it as more of a catchphrase than a serious influencer on the industry. Over the year that’s passed, AI has become more well-understood and proving itself to the mainstream.
segmentation and targeting
E-marketer reported that around 40% of DSPs are adding AI features for media optimization to predict the likelihood of winning any bid. This can improve win rates, and decrease the number of non-winning bids, ultimately cutting infrastructure costs. With efficiency at the heart of everything we do at Telefrek, we strive to keep overheads low to offer the best price possible on inventory.
Marketers who understand the importance of big data and the challenges in working with it are certainly warming up to the performance benefits of AI. Machine learning is an application of AI that finds patterns huge datasets and uses them for tasks traditionally considered to rely on human intelligence, especially decision making. With more and more data available, humans simply cannot process it all. ML is a perfect fit for complex big data-based tasks like targeting and audience segmentation. According to a report by Econsultancy and MediaMath, 80% of marketers plan to use AI for those two functions. In fact, half already do.
One major hurdle standing in the way of mainstream AI adoption is the perception that it will replace humans in all our jobs. The reality is that AI enables humans to do their jobs better. Marketers are wise to avoid AI solutions that collect data but obfuscate it from the user. If the ad placements, price paid and audience attributes are transparently shared, the DSP can serve as a discovery tool, uncovering untapped segments that human buyers may not have explored.
To aid in discovery, LucentBid always maintains a portion of budget to test new hypotheses outside of what the training data suggests to bid on. ML that acts entirely on the training data will continually amplify any bias. Human buyers with certain perceptions of who their audience should be may never guess they are biased at all. A continuously updating algorithm, however, will never stop testing new combinations, and can thus uncover some diamonds in the rough.
If you’re ready to test some budget on something new, Contact Us about LucentBid, the performance DSP.