This promoted content is produced by a publishing partner of Open Mic. A paid-for membership product for partners of The Drum to self-publish their news, opinions and insights on thedrum.com - Find out more
Full-funnel contextual advertising is finally here, and we have AI to thank
April 19, 2023
Contextual advertising’s ability to tailor the content and placement of creative to appeal to target audiences has established its value as an upper-funnel marketing tool, while its on-the-fly optimization to keep ads in step with the customer’s journey has also transformed mid-funnel performance. Where contextual has been lacking is at the finish line, but that’s all set to change.
Advances in the predictive capabilities of AI now allow contextual solutions to recognize where consumers are in the sales funnel and anticipate post-click behaviors at scale. Marketers can utilise contextual AI to plan, launch, and refine their full-funnel campaigns to keep communications relevant and engaging throughout the sales journey, as well as completely privacy-safe.
The evolution from page context to consumer context
Contextual advertising began as a means of matching an ad to its surrounding content to increase the likelihood that it will be relevant to the consumer and grab their attention. Over time, technological advancements have automated this process and expanded its scope to optimize the on-page placement and creative content of the advert as well. The more data was fed into these AI-powered tools, the better they became at recognizing patterns in what makes ads effective, and now their predictive capabilities have reached beyond content to the consumers themselves.
The strength of AI-powered contextual performance advertising is that it doesn’t rely upon pre-determined categories and assumptions about how certain demographics will behave. Instead, it simply predicts how context — both of the surrounding content and of the consumer — will affect performance outcomes. This effectively creates and identifies an infinite variety of addressable audiences on the fly.
As the process is entirely predictive, the individual consumer is not tracked or specifically targeted in the conventional sense. None of their personal data or browsing history is used — in fact, it doesn’t matter who they are at all. All the contextual engine sees is the likelihood that the consumer falls into a particular persona based upon the content of the page, where they were referred from, their device type, local time and weather, scroll speed and device orientation, and various other non-personally identifiable contextual and situational signals.
The personas themselves are derived from machine learning analysis of trillions of data points that build a detailed and constantly updated map of consumer actions. This can then be used to predict whether a consumer is interested in a particular product and where they are in their sales journey so that they can be served the most effective ad in the most effective placement. All decisions are made in real-time and become more precise over time as the model “learns” from prior performance.
For example, if someone is reading a detailed outdoor furniture style guide, their scroll speed indicates that they’re paying attention to the content, but it’s the middle of winter and the weather is poor, then they are likely in the consideration phase. However, if it’s the height of summer and they’re reading an article on where to buy outdoor furniture and what special offers are ongoing, there’s a strong chance they’re ready to buy. Wherever a consumer is in the sales funnel, contextual engines can optimize the advertising experience to lift campaign KPIs.
Post-click predictions complete the contextual package
Even if a prospect that has been correctly identified as having high purchase intent, is served the appropriate creative for their position in the sales funnel, and then clicks on the ad, there is still a chance that the conversion fails at the final hurdle. Historically, this is as far as contextual advertising’s capabilities could extend, but now that machine learning models have been trained to predict post-click behavior, campaigns can be optimized all the way through to the final sale.
This opens up many possibilities. For one, contextual engines can determine whether it is worth serving the ad to the consumer in the first place. If they fall into a persona that consistently fails to follow through to a purchase, then the impression may be wasted. Alternatively, the creative could be tweaked depending on their predicted post-click behaviors to drive up conversions, with one ad for those likely to proceed directly to purchase and another for those likely to drop off.
We are still at the beginning of this latest evolution of contextual advertising, and it will be fascinating to see how brands, agencies, and publishers alike utilize its capabilities. Now that AI-powered contextual advertising works across the full funnel, end-to-end privacy-first targeting and optimization is finally a reality, ready for the fast-approaching cookieless future.