Traditional in-person marketplace vendors can infer many things from just looking at you. Is it raining, time for lunch, and you have two kids instead of an umbrella? The news stand next to you is suddenly stocked with umbrellas instead of sunscreen.
As large sections of the web loose any notion of persistent identity, advertisers started talking about contextual targeting again. This suggests targeting the circumstances of the pageview, rather than user attributes. Let’s look under the hood.
There are actually three types of “context augmentation” one can perform:
a) page context
b) request context
c) user context
Page context is the easiest to comprehend, and it is what people historically meant under ‘contextual targeting’ – what is the page about, based on understanding the language and sentences or bags of words. Topics, categories, sentiment are well understood, but have been mostly used for blocking spend and brand safety.
Request context is about time and location of where the pageview is being rendered. Is it raining in Utah right now? Is there a big game in town and the local team is winning? These incidents can modify consumer behaviour, we can take advantage of it in advertising.
User context is trickier, but there are things we can “guess” from how the user is behaving on the page. Are they moving their mouse pointer quickly or slowly? Are they clicking a lot or little? Can it indicate mood? Would coca-cola want to target “happy” users?
In all these cases, we start to have a conversation about brand identity and brand alignment, more than user identity and targeting. Which is progress back to the basics of selling.
Leave a Reply