With the UK’s investment in programmatic set for double digit growth, pushing spend towards £2.5 billion in 2016, it’s no surprise the demand for greater transparency and highly viewable, bot-free and brand-safe impressions within the programmatic ecosystem is at an all-time high.
There has been a legacy issue with real-time bidding (RTB) where transparency into what inventory buyers are bidding on is brought into question. As buyers balance programmatic trading with potential risks to brand safety, ad fraud, viewability targets, price, and other criteria – all in a fiercely competitive market – transparency is absolutely essential.
Media quality vendors have recognised this need for transparency and created intelligent, pre-bid targeting segments using data points from billions of impressions, which can be implemented across campaigns to provide highly accurate targeting that fits with a media buyer’s requirements. But it is vital for buyers to understand how these targeting segments are generated and how they can be used to best effect.
Mapping scores
So how are pre-bid segments created and how do they come to be put into the hands of media buyers?
Media quality vendors have highly scalable platforms and rating engines, fuelled by data points from billions of impressions that allow these intelligent targeting segments to be created. Data is collected from billions of unique URLs every day, each with its own associated set of metrics. Some of these are largely static, such as a URL’s brand safety scoring – how much risk that URL content poses to brand advertising. Others are inherently dynamic and based on user behaviour, viewability and fraud being the obvious examples.
All these metrics are grouped together and analysed, and a predictive score is created for each URL and placement. A basic map of scores is then produced, which the demand-side platform (DSP) will use as the selection logic for buying media.
Optimising for targets
These pre-bid segments must then be exposed to the buyer, typically through a DSP or a maintained bidder. Because a typical ad auction must be executed within 10 milliseconds, it isn’t possible for a DSP to provide the media quality vendor with a constant stream of all bid request URLs in real time. It would create too much latency in the delivery chain if a request was sent from a supply-side platform (SSP) to a DSP, then referred to the media quality vendor to supply a URL rating before the DSP could finally conduct the auction.
Instead, the media quality vendor’s URL scores are held by the DSP in a continually updated cache – a component that stores data so future requests for it can be served faster. The DSP can ask about a particular URL at any time – the response is conducted via an API which can scale to the queries per second (QPS) needed. This allows the DSP to create the universe of scores offline and allows instant access to the pre-bid rating through a simple look-up rather than any transmission of data.
Finally the DSP must visualise the segments in their user interface for client targeting. Typically, segments are shown in the UI with a description, for example, ‘target top 10% of impressions in view for the longest period of time’, or ‘exclude ads that have a high risk of fraud.’ Buyers can select multiple segments to ensure that while brand safety and fraud risk is minimised, they are optimising towards their viewability targets.
Analysis is key
While pre-bid segments go a long way to helping buyers reach their campaign targets they don’t necessarily provide a guarantee in all cases. For example, preventing delivery on suspicious or non-brand-safe URLs is an immediate win, but the competition for highly-viewable impressions is fierce and supply for these is not unlimited.
As with all RTB buying strategies, one should take the time to closely analyse winning bid prices, segments, and volumes, both at the campaign and macro level to ensure targeting segments are working effectively.
With pre-bid segments the issue of transparency in RTB has vastly improved and thanks to the addition of key media quality elements in programmatic buying, such as viewability, buyers can now be confident the inventory they are bidding on is far more accurately represented.