“The Sell Sider” is a column written for the sell side of the digital media community.
Today’s column is written by Alessandro De Zanche, an audience and data strategy consultant.
Some could argue that media brands’ first-party user-level data is of high quality, but probably not unique if compared to Google’s, Facebook’s or Amazon’s.
What nobody can disagree on is the uniqueness of the context and the environment of quality media. It is symptomatic of the sad state of media companies that there have been few deeper and broader efforts to extract as much monetary value as possible from their content and context.
Contextual advertising capabilities need to be upgraded and refined, applying the same level of granularity and intelligence deployed toward the collection and processing of user data. We must move the focus from the user ID to the content ID – meaning a unique ID assigned to a single piece of content, such as a page, video or sound file – as a building block of a taxonomy and portfolio of contextual segments.
The single, most fundamental element of a sophisticated contextual targeting product would be natural language processing and/or video and audio recognition capabilities. They should be leveraged to fully capture the content of the page or video and identify as many reliable and relevant data points and topics as possible to be assigned to the related content ID. An article or video about Syria could relate to many different topics, such as news, history, architecture, culture, food or travel, sometimes all at the same time.
Deep and accurate categorization provides the opportunity to reliably cluster content IDs by multiple and different metrics in a very granular way, augmenting and multiplying the power of a contextual taxonomy. That is just the basics.
There could be dozens of data points of varying depth assigned to a single content ID, such as the individual publisher or broadcaster (when implemented across media alliances) and its reputational, viewability and overall advertising effectiveness scores. Section- and page-level values for the overall quality and reputation of the author, section and topic, including their popularity and ad effectiveness, could also be assigned to the content. And information about how advertising verticals perform against different content IDs could be used; automotive, for example, may perform well with certain content IDs but not with others.
Quality and reputation are subjective concepts, but the industry can and should develop credible tools to try to capture their value. Credder and Deepnews.ai are two examples.
Even the nature of the content should be also turned into a data point and attached to the content ID. Is it breaking news, an exclusive investigation, or a long read commentary? The more granular and broader the spectrum of information about a piece of content, the deeper the understanding and potential for differentiation and value extraction.
By treating the content ID in the same way we would approach a user ID, we can create personas built on artificial-intelligence-detected navigational patterns, sequences and outcomes; certain patterns, for example, could lead to higher engagement with certain products or conversions.
Intensity of behavior, the time of day, day of the week and broader seasonality, such as holidays or sporting events, should also be added to the picture, along with geolocation, if available and if users give consent. User-level data points otherwise used for audience targeting could be turned into content ID data points, such as how many men vs. women visited a content ID.
These “personas” would consist in virtual behaviors dynamically linked to content rather than people. To minimize the storage of user data, this intelligence could be applied in-session where a certain sequence of content IDs by an individual user, cut to the level of granularity required, would lead to a certain type of ad being displayed. The combination of content IDs, rather than the single content ID, would determine which ad is displayed.
Even “content look-alikes” based on content IDs with similar data points could be introduced to forecast effectiveness before the content is published.
If media brands fail to smarten up contextual targeting and don’t provide advertisers and agencies an opportunity to invest in, they will keep spending their budgets elsewhere. And in light of privacy laws and impending verdicts on GDPR-related complaints against RTB, ad tech will fill the void and build “context platforms”, where we would witness history repeat itself with media owners being relegated to the role of ad-slot farms – exactly what happened in programmatic for audience targeting.
As it becomes evident that producing quality content and contexts cannot be supported anymore via programmatic open marketplaces, the future of premium media brands, with heritage or newly created, relies on strategies that maximize their uniqueness.
But the foundation of this needed change must be reflected also in their hiring strategy, which should start looking more for deeper media and content knowledge, before and alongside ad tech expertise.
Media brands must shift the responsibility for their monetization strategies from programmatic plumbers focused on optimizing their own commoditization to visionary media architects tasked with valorizing and monetizing their unique DNA.
Follow Alessandro De Zanche (@fastbreakdgtl) and AdExchanger (@adexchanger) on Twitter.
