Since the first real time bidding (RTB) exchange platforms were introduced a few years ago, there have been changes occurring in the online advertising industry. Today it successfully delivers billions of monthly impressions for advertisers, but how is it really changing the industry?
In my view RTB has grown rapidly because it benefits everyone. It means improved yields for publishers; increased returns on ad spend for advertisers; improved client performance and the next step towards an automated sales process. Reduced costs and improved efficiencies are powerful incentives as well.
It currently represents a small percentage of overall online EU ad spend, but the uptake in the US over the last couple of years is a good indication of what we can expect in 2012 and beyond. The expected growth in RTB spend this year in Europe is likely to follow the same trend as in the US last year – 140 to 150% growth year on year (Source: IDC, 2011)
Trading desks and small agencies
From what I’m seeing in the market currently agency trading desks are really leading the charge with RTB and there’s no doubt this trend will continue, but alongside that the smaller agencies and independents are able to move faster to take early advantage of the opportunities that real time offers. They may be working with smaller budgets but they are nimble enough to capitalise on this advanced form of ad buying.
Skills to deal with data
One way I’ve seen the industry change is the technical ability of those entering it. New recruits to this sector are now likely to be astrophysics and expert mathematics graduates; such is the level to which RTB and data-based audience buying draws on ability to work with numbers. Whilst I would not go so far as to say that human relationships are no longer important, we are seeing a trend of an automated trading layer being introduced.
Direct response advertising is never going to be ‘plug and play’ and has always been heavily built on data, but RTB exacerbates this further and is the reason why we’re seeing so much technical ability coming into the industry. Many in our optimisation team are maths and science grads, and we see agencies valuing these kinds of skills too. There is always going to be a place for an optimisation specialist who understands the advertiser’s objectives, knows how to analyse user behaviour and has access to data and targeting opportunities that can be utilised to create bespoke segments for advertisers.
Yields for publishers
I mentioned earlier that RTB has meant improved yields. There was a fear in the market from the publisher side that RTB would deflate their eCPM (effective cost per thousand) which is used to measure the effectiveness of a publisher’s inventory being sold. But actually the reverse is happening based on the richness of data on audiences available and the quality of demand from advertisers.
RTB continues to grow significantly and all the statistics attest to that point. Those involved need to be aware that once you move into an RTB world, one of the key distinguishing factors will be data. There’s lots of third party data out there that can be bought in the open market where some of the data will become de-facto requirements. However, in my opinion, true differentiation will come from media buyers (or anyone providing services to end clients) who can harvest their own data for the purposes of targeting more effectively. There are numerous examples out there that I’ve seen, but certainly here at Yahoo! we spend a lot of time and effort refining our own data modelling and harvesting. 700M+ users globally creates a lot of data – and products such as true search retargeting (where we own user query volume - unlike many others), lookalike modelling of fully-declared information and more which are critical advancements for us in 2012.
Of course, the big flaw remains – if everyone is being judged on a last click / view model, then data mining to help drive more value further up the funnel (prospecting) is largely negated in value as retargeting scoops the last click or view. Obviously the attribution model needs to be addressed, but that’s a blog for another day….