TV audience targeting must reflect the dynamics of consumer behaviour

November 27, 2022

TV targeting attributes must align with the speed of the consumer, not the speed of the organisation

Audience targeting on TV remains very basic compared to the offering available on digital. As operators have moved into addressable advertising, most have initially sought to sell ad inventory using geo-targeting and subsequently by matching subscribers to purchased third party lifestyle data to offer Advertisers ‘off the shelf’ targeting attributes.

Unfortunately, little attention seems to be paid to data recency. In the case of 3rd-party data, some of it might be more than 12 months old. This has profound implications for the accuracy of targeting.

Even those TV operators who have built behavioural based targeting from their own viewing data need to be mindful of two big problems that may not be immediately obvious in the excitement to deploy analytical resources to explore such a rich data set.

Firstly, viewing behavioural attributes are expensive and slow to develop using data scientists or modellers. It can take some weeks to properly develop reliable and accurate attributes that work. The more attributes you need, the more time and cost is incurred. When we launched AdSmart at Sky, my data science team successfully developed several useful behavioural attributes, but it took a few weeks to create and operationalise each one. When I calculated the headcount cost incurred, it came to tens of thousands of pounds for every attribute.

However, the problems don’t end there. An attribute set needs to be constantly monitored and maintained in order that attributes reflect the changing dynamics of the viewing customer base. If this is not done, the attribute set will ossify and produce declining uplifts. Attribute maintenance also takes time and costs money.

The targeting variables deployed must match the dynamics of the addressable TV ecosystem. In effect, attributes must change and update as viewer behaviour changes. Whilst some behavioural patterns may be slow to shift, and therefore demonstrate established viewer habits, we have also seen very sudden, yet persistent, transformations in viewing behaviour that reveal potential lifestyle changes or intent to purchase. A relatively static set of TV targeting attributes will miss these valuable triggers.

Consequently, the demands of targeting on addressable TV advertising require high levels of automated AI, both in development and maintenance. That means reducing the manual aspects of the analytical process to a minimum. As well as ensuring high velocity of development and adjustment, it also dramatically reduces the cost of offering TV behavioural targeting to advertisers.

This is why we developed the ThinkAdvertising targeting solution to facilitate rapid implementation of an automated advanced analytics engine that creates and updates TV targeting attributes in tune with the dynamics of consumer behaviour.