Generative AI and tools such as ChatGPT and DALL-E are just the latest example of how Artificial Intelligence (AI) is automating once manual processes. In digital media for instance, AI has been applied to a range of media buying and selling workflows to enable cost savings, efficiencies as well as more intelligent audience and creative processes. One underutilized application of AI has been forecasting. By applying the power of AI to forecasting and CPM optimization, there’s an opportunity to improve ad monetization across channels including video.
As video ad spend continues to grow, many are looking to adtech to provide enhancements to ad planning, delivery, and measurement, leveraging automation where possible to provide efficiency, improve ad experiences, and drive profits. Examples of applying AI in adtech include media campaign optimization, personalized comms based on predicted behaviors, workflow automation, as well as intelligent audience curation and targeting.
AI provides speed, accuracy and efficiency in certain activities that a human can’t, saving time and money, freeing up human capital to focus on things such as strategy and creative production. Applying those benefits to forecasting could open up new opportunities. Traditionally forecasting is primarily linked to avails and the aggregate number of ad impressions available for ads. In a dynamic environment with constant changes in demand and supply – depending on time of day, week, month or year – accurate inventory forecasting can be a valuable tool for publishers and advertisers.
Artificially intelligent forecasting and optimization
Price floors are often based on static data and implemented at increments of time that can lead to missed opportunities – whether that’s underselling inventory or not selling it at the market rate. Impression level and quality can be vastly different at 8am to what it is at 8pm, just as they can be different on boxing day compared with random day in January. As a result, avails and their associated CPMs vary across the time of day, week, month and year – all of which inventory forecasting must consider with past data to estimate levels of supply and demand for media planning and CPM optimization.
For instance, if we know that on average the CPM at 8pm EST is 20% more than 8am EST, then the floors for programmatic should adjust to accommodate the forecasted clearing price of this inventory. Similarly, a media owner might have a $15 CPM in the ad server as their floor; at 8pm they might want to sell at $20 and at 8am they might want to sell at $10 and then get an average rate target of $15.
However, It’s impossible for a human to understand and adjust the price floors every hour, every day within the context of days, weeks, months, years, with fully data-informed decisions, but AI can. When AI-driven forecasting and optimization is run inside an ad server that is connected to an SSP, we can automate the process of setting, analyzing and adjusting clearing prices in real-time with more accuracy than a human ever could. AI can constantly run in the background, adjusting floors as required leveraging all historical CPM data from a publisher to improve and self correct over time.
Combining dynamic forecasting with real-time CPM optimization amplifies the benefits, automa in ting media owners’ ability to value impressions accurately, improving fill rates and reducing unsold inventory. Meanwhile, the AI-fuelled insights can provide sales teams with intelligence on the historical and predicted value of inventory to make more informed decisions.
Why AI driven forecasting is the past, present and future
AI can take infinite amounts of past data, present it for improved insights, predict the future with more accuracy and efficiency than a human, and most importantly automate the processes to put that information into profitable action. With growing AVOD, FAST and live streaming ad opportunities, continuously growing CTV budgets and new buyers investing in the medium, the ability to forecast avails more accurately to optimize CPMs in real-time based on those forecasts will enable improved monetization.
At a time when the industry is buzzing around the possible applications of AI tech, it’s important to focus on how AI can have an impact now while innovating for the future. There’s already examples of AI tech that can be applied to video ad workflows and monetization such as creative review automation, audience curation, and intelligent impression forecasting and optimization. Combining AI with streamlined offerings can further maximize the value in terms of driving cost efficiencies as well as profitability. For example, forecasting inside of an ad server connected to an SSP allows us to set, analyze and adjust the clear price in realtime for improved performance.