We’ve taken ‘programmatic’ to mean any channel where instructions are programmed in before being executed by a machine. This has led to us to buying Facebook, LinkedIn and PPC as well as our core RTB.
I’m wondering if ‘content serving’ is promoting posts on Facebook or LinkedIn or putting content like news headlines or blog posts into banners for wider distribution. Or all three and more of course.
The quick answer is it can mean different things to different people. I tend to think of ‘content marketing’ with the same ambiguity many have applied to the term ‘marketing’ itself.
The communication channel is central to marketing. It conducts’ content of various types, from a range of sources, to different audiences. Their response, and the implied influence of this communication on future actions, is tinder to the correlation/causation debate that underpins the industry.
I suppose the logical way of answering this question is to look at programmatic technology as simply the delivery platform, which is currently replacing many of the automated buying technologies already used in specific channels, such as social media, search, mobile, video and TV.
It is, in effect, the platform behind the communication channel and as such is distinct from the content it carries.
Obviously it’s not as simple as that, as we can use RTB technology to alter, delete, concatenate and generally optimise the content actually carried in the communication channel, to the extent it ultimately becomes entwined within the technology itself.
Technology and people
And this takes us ‘neatly’ to the role of technology and people within programmatic technology! I would like to say that technology is the over-riding requirement in delivery, but as any media planner armed with the humble excel spread-sheet will know, that is not entirely true.
There is a fundamental requirement at a number of levels for data access, a trading platform that can interface into supply and buy side architectures and an agreed protocol of programming future traders.
However, we are still effectively pushing the boundaries of addressable media, with the current buying points representing only the tip of the opportunity.
Visualising trading in RTB
At the heart of RTB trading is a human trader who ultimately optimises delivery against a series of trading strategies, not too dissimilar from their peers in the financial world.
Depending on the objective of the campaign, a series of strategies will be initiated to access potential opportunities pooled across networks and private market places. ‘Decision trees’ are used graphically to represent the multitude of trading strategies and ROI opportunities available over time.
The number and size of the ‘nodes’ represent the opportunity and are then refined and amended until acquisition costs decrease to target levels. The period of time this takes depends on the volume of spend & conversions available, the trading strategies and the conversion time lag.
We use Decision Trees to show how the platform learns and how we optimise:
The nodes, their colour and size represent dimensions we can use to optimise against and improve performance.
There are many different dimensions that can be optimised against:
Insights in the mist
We can split these dimensions further into variables to increase the granularity of optimisation. Of course we are still uncovering variables with the obvious candidates still dominating the optimisation process.
Publisher type, sector/channel of publisher and so on still drive the tip of optimisation but interesting insights are starting to emerge which can alter the strategy and planning process enormously.
As a media company, we tend to focus the trading tools & strategies against a number of media objectives which we will discuss below.
This makes it easier to map against the existing media strategy and demystifies the approach for our clients.
Prospecting is the technique used to drive brand new users in a campaign. We can learn a lot about new user behaviour from prospecting strategies and use this to further inform our campaign set ups. We can find lapsed users, audiences that look similar to existing audience segments or are connected in some measurable way to one another. Prospecting is the sharp tip of the CRM machines that ultimately retain and convert initial purchasers into loyal customers.
Our programmatic platform categorises domains (websites) and pages dependent on their content. These are then allocated a category (channel), which can be broad, e.g. fashion, or more targeted.
Programmatic traders then select which channels to run our ads against within the platform. These contextually relevant placements help build awareness and drive response.
Pages are categorised in real-time based on a combination of semantic keyword targeting (including sentiment) and algorithmic probability lessons to create a dynamic site list on which we use to serve our adverts.
Retargeting uses cookie data to actively re-market to individuals who have visited a client site but perhaps not converted.
They represent an opportunity to engage further and in many ways are extensions of the CRM cycle using personalised information to make the experience more relevant and more compelling.
Private Market Places are becoming central to aligning key publisher inventory and data into an integrated buying opportunity for clients.
If spend is sufficient to justify the necessary time and resources, we can start to create exciting and real-time media products that automatically adapt to user behaviours and market trends.
These are specific sites with inventory that we can access programmatically. This has traditionally been display inventory but in the future I expect this to include content, CRM, rich audience data & display inventory from large retailers looking to add incremental revenue.
Look out for my forthcoming final instalment, which will look at the role of media attribution, case studies and a forward facing perspective on how RTB will be incorporated into client side trading desks.