How often does a multi-billion dollar industry develop in less than five years?
How often does that industry have the potential to grow to tens of billions within another five years?
The first industry I was involved in like this was paid search. After a couple of hard years struggling to gain momentum in the late 1990s, the category started to explode in the early 2000s and has since become the leviathan of digital advertising, a $58 billion industry globally.
Now I’m working in a new space within digital advertising that has already demonstrated huge growth and has the potential to rival paid search in the next few years. That space is called programmatic.
What is programmatic? Programmatic is a technology that enables buyers and sellers to instantly and automatically trade digital advertising inventory through ad exchanges, marketplaces that operate rather like stock exchanges but for ad impressions instead of stocks.
Like many new technologies, programmatic evolved in response to economic and process inefficiencies.Before 2009, practically all digital advertising (banner ads, video ads, rich media ads etc.) other than search was sold in bulk and executed via traditional manual methods. Buyers bought advertising space in large blocks. As a result advertisers' ROIs were often lower because buyers were unable to exert as much control over the specific inventory they were buying. On the process side, directly negotiated prices, emailed and faxed insertion orders, manually transferred creative assets and high touch; multi-step ad operations defined the approach. For example, back at Yahoo in 2008, we still had a 30+ step process for getting ad campaigns live. This resulted in lots of work, time, back-and-forth and cost.
Birth of a New Industry: Programmatic
Starting in 2009 several companies (including my company OpenX, then later Google) began to offer publishers a new way of selling, or trading, their advertising space. They could do it automatically through a real-time ad exchange. This new model offered several advantages for both buyers and sellers.
Buyers could analyze each ad impression in real-time, determine its value and bid accordingly. For the first time, they could buy exactly the audiences they wanted and avoid the waste of bulk buying. As a result, they could also pay higher prices for the inventory they really did want. Also, because the ad exchange aggregated ad inventory (or “supply”) from many selling publishers, buyers could buy their target audiences at scale across multiple publisher sites in one efficient way. And they could do all this instantly – in true real-time, during the ~200 milliseconds it takes to serve ad space to users.
Meanwhile sellers could “real-time sell”: they could place inventory they had not yet sold or inventory they had sold but on a non-guaranteed basis into the exchange, set a reserve price and see if they could get higher bids from the new programmatic channel. This was huge. Rather than having to guess what a buyer was going to pay you and allocate inventory to buyers based on that guess (which typically was different for each individual impression), the seller now had a situation where buyers were telling them explicitly how much they were willing to pay for each individual impression, in real time. In short, a more economically efficient allocative mechanism had been created.
Finally, there were also significant efficiency gains on the process side. Buyers and sellers could now execute transactions with much lower friction, through automated algorithms and API-level interactions versus multi-step manual processes.
Naturally this fledgling industry faced challenges. Premium publishers worried about channel conflict, direct sales cannibalization, control over price and quality and transparency. In the early years, as with most new technologies, programmatic was adopted fastest by more aggressive, mid-market publishers with less to lose. But over time the concerns were better addressed, both through features and controls but also through dialogue. Now the majority of the Comscore 100 publishers use programmatic as an element in their revenue strategy and most major advertisers include programmatic buying in their portfolio. Other players like Facebook started to adopt programmatic trading, launching FBX in mid-2012.
Along the way, an entire new ecosystem has evolved. On the sell side, several ad exchanges and supply-side platforms like Google's AdX, OpenX, FBX and Rubicon achieved real scale.
On the buy side, a new class of company emerged. Generally known as Demand Side Platforms (DSPs), companies like Turn, Mediamath, Appnexus, The Trade Desk, RocketFuel, Invite (now Google) and Criteo created the ability to bid into ad exchanges in real time — in other words to buy programmatically. Within the major agency holding groups, new organizations called Agency Trading Desks (ATDs) were created to leverage this new way of buying advertising.
And a range of companies focused on data also flourished. Generally known as Data Management Platforms (DMPs), companies like BlueKai, Lotame, Acxiom and Krux focused on applying the rapidly evolving science of Big Data to audience data at scale, helping arm the DSPs and ATDs with the data signals they needed to value immense numbers of ad impressions instantly.
As a result of all this innovation and economic activity, programmatic grew from zero in early 2009 to $6 billion in 2014. By my estimates, this is roughly comparable to the first five-year cycle of the paid search’s rapid growth. In other words, during this first five-year phase, programmatic proved that it can scale.
Programmatic Scale: Formats, Screens and Business Models
During its scale phase, programmatic grew very fast but remained fundamentally quite simple. This is most clearly visible when we break the business down into three critical dimensions: formats, screens and business models.
In the first dimension (ad formats), during the last few years over 90 percent of trading volume in programmatic has been executed within just three IAB standard ad format sizes. On dimension two (screens), the vast majority of RTB transactions have occurred on desktop – and laptop — screens. And in the third dimension of the business model, programmatic has mainly been applied to open market, RTB bidding model where all buyers compete for ad inventory.
In short, programmatic’s first phase has been a relatively homogeneous system focused on scaling.
Today’s Inflection Point: From Scale to Sophistication
Now we’re at an incredible inflection point where programmatic is rapidly becoming more sophisticated because participants on both the buy and sell sides have seen the power of programmatic and want to apply its power to a much broader, more diverse set of opportunities.
For example, the range of ad formats that are traded programmatically is beginning to increase. We’re starting to see trading of a greater array of rich media and more specialized ad units that traditionally were only sold via Insertion Orders (like big homepage units). Programmatic trading of Facebook’s sponsored stories via FBX has become a major category within programmatic. And we’re also starting to see cutting-edge work on programmatic native within mobile.
In addition, screens are rapidly becoming more diverse as programmatic buying is being applied to mobile phones and tablets. Over time we expect this trend to extend into Smart TVs and other connected screens.
And finally, with regard to the business model dimension, we’re seeing a rapidly proliferating range of transaction types such as private marketplaces (in which trading takes places based on more tightly defined rules of participation), programmatic direct (a hybrid approach wherein ads are sold through the traditional sales process but executed through the new programmatic infrastructure) and many variations of these new deal structures.
This kind of maturation cycle makes sense. Many other industries evolve from a preliminary homogeneous scale phase in which they prove themselves before entering a new level of sophistication that emphasizes more flexibility and enables a much more heterogeneous array of applications. A good example is car manufacturing. In the beginning the products come in one form (think Model-T: one shape, one size, one pricing model etc.), but now you can pretty much configure your own car.
Another example is sales force automation software which started out pretty simply but has now become enormously sophisticated, with companies like Salesforce enabling companies to finely tune how their SaaS sales support systems mesh with their operational processes to give them terrific execution leverage.
Far Reaching Potential Scope
As we begin to grasp the full potential of programmatic it’s becoming clearer to me that the potential is truly enormous. The efficiency and effectiveness of the underlying programmatic technology means that it will likely play a vital role in every single digital advertising category (except Search).
Since the industrial revolution we have learned that nearly all industries are transformed by waves of innovation that improve efficiency, often through automation based on technology. These innovation trajectories are rarely totally linear and they often transform parts of industries more than others, but they are usually inexorable.
In the case of digital advertising, programmatic will apply across all screens and formats. Any screen that has a digital connection will be in scope: not just the familiar desktop and mobile screens we’re focused on now but scenarios we are only beginning to imagine. And these scenarios will use data in ways that are not obvious today. Adjusting video ads in digital outdoor screens overlooking the freeway based on the speed of traffic. Firing up a targeted ad in the in-elevator screen that recognizes your phone’s unique ID. Presenting more relevant native ads within your Smart TV experience and removing all those 8-minute workout routine ads that so annoy you. Presenting a curated restaurant recommendation in the heads-up display (HUD) in your driverless car. They will all be programmatic.
In fact, there is no reason programmatic technology has to be limited to advertising. Over time we may see the same underlying real-time, programmatic response prediction technologies applied to unpaid content as well. That restaurant recommendation on your HUD could be purely “editorial."
TiPut simply, the future of advertising likely has two main paradigms. Unique, ultra high-value, custom and complex advertising transactions — sponsorships, large premium guarantees, Super Bowl ads — will likely be handled by people negotiating and working together.
Everything else that is more scalable, with clear parameters and standards, will be handled programmatically through technology. As a result, within the next decade, programmatic may well be bigger than paid search.
This means that anyone involved in modern media or advertising related businesses; whether on the buy-side, the sell-side or even the content side, needs to understand the current state of the programmatic industry and where it’s going. If it's not affecting you today, it probably will be soon.