Scale & Precision in Contextual
Our mission at Advanced Contextual is simple. We believe individual privacy is important and doesn’t have to compromise good business. We provide publishers and brands with tools to better understand their audiences through content consumption and dynamic intent data. We do this in a way that preserves both scale and precision, not tied to any single identifier.
Achieving precision at scale has always been the “sort of” attainable goal in audience or content targeting by brands and publishers. Developing scale has always been easy. Developing relevance via precision and at scale is not so easy. That’s always been the knock on contextual. There’s nothing inherently imprecise about scaled contextual signals. It’s all in how it’s used and deployed.
There are a few basic concepts which drive both precision and scale in contextual:
Focus what is being targeted – two dimensions here; what is targeted (topic, keyword, entity or mindset) and what contextual trigger is used (i.e. how well-defined a system’s read of the page is). Targeting “cloud” as a keyword is too ambiguous, and will result in matching both computing and weather content. Additional surrounding context is needed to disambiguate this term, and how a contextual engine develops that surrounding context is critical in balancing the tradeoff between precision and scale.
Avoid false scale – it’s easy in contextual to develop irrelevant scale that would look relevant on the surface. If you target “home improvement” and there’s a story about that on Yahoo!’s home page, some contextual platforms would recommend and target Yahoo!’s home page even though the story was only there for 15 mins. That would become a lot of wasted impressions.
Find related topics via audience analytics – we’re moving to a world of targeting cohorts of people and that’s a strength for contextual. The contextual engine has to be able to find related topics at the audience level (life insurance content readers also reading about family planning) in a privacy-safe way. Creating a consumption graph of people who start with “life insurance” then consume other related topics lets you find the audience and target the pages. With the right permissions you can target the audience too.
Here’s an example.
A B2B brand wants to target cloud security content which has certain organizations (entities) associated with the endemic definition of the audience. They have to be able to target the phrase “cloud computing security” and do that on pages which have the named entities they want. From there they have to listen to only article-level content to avoid high-reach false scale content. With that done, the contextual engine has to find other topics that cohort reads, then target the exact pages inside those topics.
All of this has to come together in one workflow. We can’t expect publishers or brands to pick and choose different aspects and expect the right result. This is a programmatic environment and content consumption is counted in the millions of impressions a second. An easy workflow with immediate programmatic distribution capabilities has to be present for advanced contextual to grow to a dominant signal.
All this is to say that an advanced contextual engine must foster scaled and precise targeting of various triggers in a well-defined way, across relevant content with insights into that audience’s total consumption, not just endemic.
Contact us below to learn more.