Benjamin Tice will be speaking at ATS London Tuesday September 10.
Twenty-plus years after the birth of the internet, programmatic buying and selling and RTB is now a multi-billion dollar industry. The story is different on mobile, where app stores launched just five years ago and RTB is nascent. Many of the recognised DSPs from the display world are beginning to dip their toes into mobile, and many of the more recognised publishers are only just starting to carve out portions of their inventory to exchanges to monetise on their behalf. The promise is significant, however, with RTB expected to account for around 25% of all display spend by 2015.
This has led to an opportunity for an increasing number of smaller mobile DSPs and Ad Networks with bidding capabilities (for ease, we’ll call them all mobile DSPs), to take advantage of an open market without existing dominant players. As such, much of the RTB spend in mobile today is shared between a growing number of these smaller, less recognised, often ‘mobile-first’ organisations.
Until now, Mobile DSPs have focused on providing clients, agencies and trading desks with access to a platform that delivers the efficiencies of buying mobile inventory programmatically and the benefits of buying media through auctions in a liquid marketplace – essentially, providing advertisers with first-in-market advantage. What has been lacking, though, is the ability to provide the same level of sophistication in buying and optimising as is currently offered by the recognised DSPs in the display world. This includes using data to inform targeting strategies, conversion tracking post-impression/click then optimising, private marketplaces and so on.
It’s not surprising that in many cases the most attractive traffic to a Mobile DSP today is the least expensive and highest performing (CPI, CTR, CPA); but the current methods for sourcing this type of inventory are elementary. Mobile DSPs that only look at historical CPMs and CTRs of publishers will find themselves up against strong competition from other DSPs operating in the same way. When multiple DSPs bid on the same inventory, for the same reason (because it’s inexpensive or performs well), they will soon compete to such an extent that the inventory is no longer affordable/effective. The DSPs that will succeed are those that can interpret more than just historical price and performance of a publisher’s inventory; the winners will differentiate by utilising all the data that is available to them to make more informed bids.
This brings us onto an issue regularly discussed when referring to the use of data for targeting in mobile – the general absence of the cookie. This is arguably a moot point in the mobile ecosystem, given that 80% of time spent on a mobile device is in an app rather than on the mobile web so in-app inventory therefore dominates mobile RTB traffic.
As the leading provider of in-app analytics, Flurry is in a unique position to append its own first-party and proprietary data to devices by creating Flurry Device IDs for the one billion users seen across its network of 350,000 apps. Demographics are inferred from a registered panel of 40 million users (when one considers that comScore and Nielsen use panels of thousands on the web, the richness of the data becomes clear). Going a level deeper, Flurry has built 35+ Personas (e.g. Young Mum, Sports Fan) that can be appended to a device as a result of app usage, exceeding twenty-five times more than the average in that particular Persona category. Suddenly, based on app usage data, the DSP has gone from buying an impression for a 25-year-old female in the UK, to buying an impression it can be certain will reach a first time mum who lives in Birmingham who is a value shopper and loves to travel. The Persona data is bundled with the impression, sent to the DSP from the Flurry Marketplace to inform the bid strategy at no incremental cost to the DSP, and the winning bidder pays the winning bid. While third-party data from vendors such as BlueKai or eXelate exist in mobile RTB, scale and reliability are issues, as these providers are accustomed to working in a web environment where data is appended to cookies. DSPs will also find additional costs associated with buying third-party data, and this affects their margins considerably.
At Flurry, we see only a handful of DSPs using all the data that is available to them. Fractional Media, one of Flurry’s Marketplace beta partners, saw a twice over increase in their ROI when they started using sophisticated targeting data. Additionally, and perhaps more importantly, Fractional saw little or no increase in the cost of obtaining the media, as there is currently little competition between DSPs using audience data to inform their bids.
As in any auction setting, applying quality information to the supply increases the value of that inventory; the more you know about an impression, the more you will be willing to pay for it. This knowledge enables DSPs to have a more streamlined approach to the inventory they bid on, eliminate wastage in their campaigns and realise the benefits of programmatic trading. As the market matures, and DSPs become more adept at using data signals in their bid strategies, we will also start to see publisher CPMs grow. Growing publisher CPMs will encourage more quality inventory to enter exchanges, and as this happens, DSPs will benefit as we can start to move away from terms like ‘remnant’, ‘non-premium’ and so on.
One thing is certain: as time spent on mobile continues to eclipse the web, the desktop DSPs will get in the mobile game more aggressively, and they bring experience and existing models for doing programmatic effectively in the desktop world. In order for Mobile Programmatic to realise its potential, mobile-first DSPs need to continue to develop their data capabilities in more sophisticated ways. Those that don’t won’t be around for long.