It’s a fascinating market, the digital advertising market, and here I want to give a very brief overview of its size and structure, and look at one of the areas, DSPs.
Obviously this is ever so brief and I will expand on the different areas of the marketplace in later posts but I hope you find it useful and maybe for some of you, help spark a few ideas for the new year.
The digital advertising market in the UK has grown at a rapid rate – from £825m in 2004 to £7.2bn in 2014 and on to a first half of 2016 of £4.8bn, up +16% YOY (all figures given in this article are from IAB and PwC). That’s a growth of 4% market share of the entire advertising market to over 40%.
Mobile as a sector within the digital advertising market has had spectacular growth – from just £28m in 2008, then passing £1.6bn in 2014, and on to £1.7bn in the first half of this year, a +56% growth YOY!
Paid Search remains the largest sector at £2.5bn for H1 2016, then Mobile, then Display at £1.6bn. Within Display, Banners make up £602m followed by Video at £474m.
But look at these sectors within the Mobile sector – Paid Search £908m, but Display an impressive £802m, so more than 50% of PC’s value and up 52% YOY.
Split between Advertiser and Publisher services with the ad exchanges and networks in the middle, one of the areas that has always fascinated me are the DSPs. ‘Demand Side Platforms’ spend their time reading analytics data and building mathematical models to optimise campaigns.
More about DSPs
I previously worked in a DSP and one of its USPs was its targeting. Most DSPs will use contextual targeting and behaviorial targeting, but this company also created what they called ‘Cluster targeting’. Basically for cluster targeting we put context and behavioural to one side for a moment and just analysed the data to find clusters of people clicking through to target sites. We graded them in terms of click through rate and then analysed their properties. So instead of laying theories down that to drive clicks to a site they need to have visited sports sites for example, as behavioural targeting does, here we didn’t start with a theory and just let the data do the talking. We would find some unexpected patterns. We wouldn’t try to understand or explain but simply build out the clusters based on these findings and use for targeting the client’s sites.
Sorry, maybe I should have first explained what a DSP is in more detail. DSPs allow advertisers to buy impressions across various publisher sites, but layer in targeting to specific sites based on the site’s content, location and the user’s demos and previous browsing and online purchasing behaviour. Publishers make their ad impressions available through the ad exchanges, and DSPs then select out the sites for the advertiser to buy. Often the price of the impressions is determined in ‘real time’ through online auctions known as real-time bidding (RTB). The process takes milliseconds, as a user’s computer loads a webpage.
Ad Networks also do this (and there is a hell of lot of overlap in the market), but the edge DSPs have is they offer the process of buying, tracking and optimising on one single platform tool.
DSP players include Invite Media, MediaMath (one of my favourites J), Turn, Appnexus and Data XU.
Benefits of DSPs
A key one is buying efficiency – the process is almost fully automated – the negotiations with suppliers, linking to ad servers, setting up tracking, sending insertion orders, monitoring performance, etc; is now an efficient and computerised process. Another is real-time optimisation – now within a day you can see ad performances, by ads, by remarketing lists, demos, etc and adjust immediately instead of waiting for the end of the campaign. Also cost saving – because of the complexity of the bidding process in terms of the bidding criteria it’s not always the case that the highest spender wins. The ability of the platform to handle this level of complexity can mean you achieve low cost buys.
Choosing more than one DSP
Finally there are some arguments for selecting a range of DSPs for your business not just one. Four reasons:
- Technology – DSPs submit their bids for SSPs (platforms that buy on the publishers behalf) firstly internally against a range of criteria (budget, performance, time of day, block lists, minimum bids etc) and must win this before even sending the bid out to the SSP. So a bid in Invite of £2 may not necessarily beat a bid of £1 with Appnexus.
- Size – as you optimise a campaign you will often shrink the audience available on a platform
- Risk management – you might not always have the inventory you want on a certain DSP, or it might even disappear, so you need to spread the risk of always being able to access the quality of sites required
- Analytics and insight – running campaigns across a range of platforms means you can run test and learn activity on a continuous basis and across a wide and representative mix of publishers.
So a very brief article to such a complex area, I hope it has helped in some way, and I will continue to publish more in the next days/weeks.
Read more digital marketing posts and post your own on The E Room.