The birth and evolution of programmatic advertising – where did it come from and what does it need to succeed

The key point we need to make is that programmatic per se is an execution method that enables you to govern interactions faster and at a greater rate. It doesn’t necessarily mean those interactions (or those bids for digital space) are any or always smarter.

The reasons for this are clear when you take a look at the origins of programmatic.

If we cast our eyes back to the pre-digital advertising landscape, advertisers were forced to use context as a proxy for the audience who consumes the content against which they were buying an ad. This means making assumptions based on who was thought to be interested in that content. A classic example of this is that certain newspapers have a certain audience and niche magazines have particular audience personas. When digital advertising first emerged, it was this model that was initially applied.

However it soon emerged that this model was inefficient for all concerned.

Marketers had to buy inefficiently, using models such as section takeovers, share of voice, and so on, leading to significant wasted spend.

It wasn’t good for publishers either. Publishers relied on sales teams selling their entire inventory, up front – every impression not sold was lost revenue – and large operations teams to traffic campaigns and try to optimise their yield across potentially competing insertion orders, all while trying to match their advertisers’ KPIs.

This produced two challenges. Publishers wanted to make sure they were getting the best rate on every impression and marketers wanted to make sure they only purchased the people they were interested in. This is the key: buying PEOPLE, not touch points; and to do this data was required.

Identifying individuals

Fortunately the means existed to trace the individual behind a given impression. All devices and browsers have a unique identifier (a device ID or cookie) that is appended to every request for content. As these identifiers are sent with every request, there are two immediate upshots:

• It is possible to build up an understanding of an individual based on how they engage across a site, by linking together requests with the same identifier. Moreover it is possible to define audiences – collections of identifiers – based on common patterns of engagement.
• It is possible to serve specific content to a specific identifier. This allows advertisers to buy the audiences – collections of identifiers – that are most valuable to them.

Adding auctions to media buying

As has long been established, a fair way of ensuring that a supplier gets the best price for a unique product is via an auction process. The same is true for an individual impression generated on a given publisher site. In this context, the auction process must be extremely quick – nearly instantaneous – to ensure that an advert is returned quickly enough so that the user sees it and the consumer experience is not interrupted.

In the last few years the technology has existed to support these requirements. The rise of “real-time bidding” allowed publishers to run an auction on any impression generated on their sites, thereby ensuring that they got the best market price for each impression; and it allowed advertisers to bid only on impressions generated by users who were of interest to them.

As mentioned above, there is no way a marketer can actually manage all of these different strategies at scale across all their platforms and partners, in real time. Instead marketers started to define strategies that could be processed automatically, matching audience types with predefined and pre-designated bid values: a programmatic approach to buying individual impressions generated by individual consumers.

What’s next in this space?

Of course, in order to be smart about who I’m buying I need to know what that identifier means to me: I want to aggregate all my data so that I can make the best decision possible. I need a tool that allows me to transport my consumer insight from whatever repository it lives in and make it available in the programmatic landscape where it can be used in real time.

The answer to that question is the DMP and data-informed decision making – which is how programmatic as an execution capability really comes into it.

To find out more about DMPs read the white paper, check out this recording of a webinar we recently held or visit our DMP website.

This article is an extract from our recent white paper on DMPs, data-informed decision making and programmatic advertising. Download your copy here.