Building an Ad Server to Serve Your Application

Last Thursday my talk on Bootstrapping Your AdServer provoked hems and haws at the AdMonsters Network Forum (Dec 9th, 2010) in New York City. My intention was to provide a bit of insight to the challenges that publishers encounter on a regular basis; especially those in the midst of a growth spurt. However, I think it was taken to be an indictment of the contemporary ad network.

Like the OEM ad server, ad networks serve a diverse clientele in terms of both volume and genre. They intend to meet all the needs of most publishers and at least some of the needs of larger publishers. Indeed, they have a place in the revenue stream of almost every ad-supported website in existence. My aim was to arouse a sense of independence in publishers and to remind them that developing a fantastic product should be their main focus; to take into account ad serving only second to achieving the former. My mindset can be best summarized in the following statement:

Develop your ad server to meet the needs of your application. Don’t develop your application to meet the needs of your ad server.

That should seem obvious, but all too often we see websites and web applications designed to fit the ideals espoused by the IAB, DSPs, networks and ad servers alike. This standardization has a few results, some positive and some negative, depending on your point-of-view. One positive result is that delivery and verification are more easily achieved. Standardization also allows for smaller advertisers to create rich advertisements and distribute them widely. I’d worked on the buy-side for some time before my current tenure at and can speak to the ease at which this is done. However, the constant demand of brands and agencies, and thus high-volume publishers to produce “first-to-market” offerings, has left the publishers with few out-of-the-box resources to depend on.

I am not a detractor of the standardization movement, nor do I care if networks and the rest of the ad-serving ecosystem take custom creative solutions into account while building their offering. That’s not what networks are here for. As long as brands demand innovative ways to interact with customers, clients and patrons the art of digital advertising will continue to evolve. By their very nature, networks and ad servers will not be able to accommodate bleeding-edge custom creatives. This is in no way an indictment of either. I only seek to remind publishers of that great American mindset, in which we break through the constraints of technologies built for the masses in order to achieve new heights in audience experience.

Why Should You Care?

Like it or not, the Ops community is at the heart of this battle, especially ad ops firms that do a lot of the custom creative trafficking on behalf of mid to big cap publishers. Until recently there has been no counter-weight for the heavy demand-side infrastructure being built and bought. I am happy to say that the Yield Optimizers and the RTBs have come to the aid of the publisher and then some.

But digital advertising inventory is quickly becoming a faux commodities market. There is a glut of low-quality inventory and as publishers innovate to drive more traffic to their applications, OEM ad-servers lag behind in the technologies needed to serve these custom ad units. DSPs are creating value for the advertiser by identifying market optimal prices for segments on web properties and bidding on them in real time. This is driving prices down across the board and giving digital advertising a greater ROI. That’s good for our business overall because it opens more doors to smaller advertisers. On the other hand, yield optimizers and the RTBs are creating value for the publisher by showing them all of their options at once and making the split second decisions necessary to grab the highest bid during any given moment. In the fog of this war, service providers (I use that term interchangeably with “ad server”) are trying to build products for the masses. Their strategy is to sign contracts with publishers that give ad servers (and their newly formed networks and exchanges) license to the unsold inventory. It becomes a race to grab the most inventory. This is one of those beautiful cases where rationality is actually increased by speed, and the two bargaining parties want two different but not mutually exclusive conclusions. The wealth of inventory means that advertisers can meet their CPA goals with or without you. They decide their price, tell the RTB what that is and the RTB says, “hey that’s the best price possible for this publisher, we’ll take it” or “move along to another property where you can get what you are looking for at the price that you want.”

This makes buying and serving standardized ads incredibly easy. It makes monetizing, even low quality content, incredibly easy. It makes adding revenue to a high-volume site, (you guessed it) incredibly easy. But it doesn’t serve the need of the brand or the agency to produce bleeding-edge advertisements that catch the attention of the CCP (customer, client or patron) in a unique way. Only direct deals that push the limitations of the website or application can achieve that and because of this simple rational constraint, inventory glut will only be a problem for websites that conform to the constraints of DSPs, networks and ad servers alike. happens to work in a competitive vertical. Advertising to music-enthusiasts has never been easy but our custom units produce unheard-of user interaction. Campaigns for artists and record labels routinely see CTRs (or interaction rates) in the high teens. You can quote me on saying 10% + . We receive reports from our brand partners that praise our custom units as the highest interaction rate they have ever seen. I’m happy to brag about the achievements of my colleges after a few beers on any night. In this case, it (finally) serves a purpose; to illustrate the accuracy of my claim: Custom creatives are the future for big Publishers. Creative, A.D.’s, C.D.’s, salespeople and developers working with creative and digital agencies will produce the user experiences that are necessary to bring the lion’s share of media buys out of traditional media and into the digital arena. Your website will either suffer financial stagnation as the RTB environment races towards equilibrium or be part of cohort that blasts through that $40 Billion wall and absorbs some $100 Billion spends that have been doing nothing to increase brand interaction on those boring old mediums.

Breaking Through

Here is a bit of guidance to get you started.

  1. Build your “dream” ad unit first and then figure out how to serve it. This harkens back to my thesis. Ask yourself how to create ads that feel like features and encourage the active participation with a brand while showcasing the uniqueness of your site.
  2. Don’t confine yourself to what the sales rep tells you about standard operating procedure. They may not have what you need right out of the box but don’t be afraid to ask them to get creative. At GS we use multiple ad severs to achieve the unique delivery mechanisms and accountability needed for our top brand and agency partners. This was achieved by asking everyone at our service providers to get creative – they did.
  3. Consider using APIs to create custom reports that showcase your unique successes. Don’t expect agencies or brand managers to dig through all of the information available to them and find the unique “something” that your site delivers.
  4. Buy data to supplement targeting. Buying data is no longer controversial and importing to supplement targeting and improve user experience may be worth your while. I participated in a friendly exchange with Eric Picard (Traffiq) at last week’s Ad Monsters Network Forum in NYC. He argued that we have dug ourselves a hole by showcasing our superb abilities to target with obscene granularity. We created a demand for complexity that Print and Television can’t compete with. Now that we have, all publishers must meet those demands or risk not partaking in the riches.

The usefulness of Ad Networks, DSPs, yield optimizers and RTBs is nowhere near outlived. We are just starting to reap the benefits of them as advertisers and publishers. However, publishers must find unique solutions to the problems of innovation. We must spread dependencies across service providers and internal staff to produce rich experiences for users and advertisers alike.