Recently my team and I were forced to dig deeper into the term Real-Time Advertising (RTA), since our natural first reaction was to ignore yet another industry term. It was interesting what we discovered. A few of us in Smaato Product and Marketing stepped back from building products and running campaigns to look into the RTA topic, and discussed its relevance to Smaato.
Obviously we’re in the era of “need for speed,” and that has forced everyone in our industry to constantly position themselves with acronyms and buzzwords from the lumascape. Some of the most popular ones are:
- Real-Time Bidding (RTB)
- Supply/Sell Side Platform (SSP)
- Programmatic premium
- Demand Side Platform (DSP)
- Social media platforms
And many more.
Similar to any other efficient market, our industry revolves around finding price optimization and efficiencies that are based on supply and demand. The common theme here is that the majority of the companies in our ecosystem are focused on advertisers, the so-called “demand” side, and those advertisers’ desire to find audiences close to real-time that convert for marketing ROI purposes.
The ones that are focused on the supply side are also confused, or are struggling to find their place, and thereby like to attach themselves to the most popular terms such as real-time bidding exchanges or supply/sell side platforms. We can easily write many more pages with confusing but limited focus and that are based on the most popular acronyms; however, let’s zero in on what no one discusses, either due to lack of knowledge or lack of importance: the “ad server.” More specifically, let’s focus in on sell-side ad servers, and save a discussion of buy-side ad servers for another post.
There is always a lot of discussion around demand daisy chains or waterfalls and yield optimization, yet have we really tackled the topic of ad servers as the true backbone of this process? Ad servers have received less attention in the last 5-10 years because they have come to be perceived as commodities. Not to mention that 5-10 years ago the majority of those confusing acronyms did not exist, and definitely were not thought about when building an efficient ad server.
The most important fact to note here is that an ad server, of course, delivers ads on publisher properties. A supposedly boring piece of technology, sure; however, “serving” actually denotes decisioning logic, and that logic has changed significantly in recent years.
Historically this ad delivery decision logic was based on a demand source or a campaign based on price. However, if you flip the coin, today the decisioning logic has to be based on the value of an audience/impression, and data plays an important role in that decisioning logic. The nucleus of such data storage and decision logic is the ad server.
Obviously, mobile advertising growth has gained a lot of its traction due to the fact that more people are on smartphones, and advertisers are naturally going where they can access those audiences and eyeballs. Mobile also happens to be the platform where time spent has gone up significantly vs. desktop.
I think it’s also important to note that mobile has removed the reliance on cookie-based targeting and advertising and has further distanced itself from traditional decision-making on ad servers, SSPs and RTB exchanges based on cookie-based targeting. While all this growth has occurred, the importance of finding an audience in near real-time is what is driving the shift in advertising spend on to those apps or mobile websites where users are spending a lot of their time.
The new definition of a publisher is no longer a large traditional publisher or a large traditional website. Rather, it’s the new generation of content creators that come in many forms, including sites, blogs and apps. This exploding segment of new supply sources/content creators generally exist inside app stores like Google’s or Apple’s and are called application developers. We recently released our mobile-first ad server for this new generation of publishers called SPX (The Smaato Publisher Platform), and we ensured that it had a strong emphasis on real-time decision making and real-time price setting.
What we did not want to build was a mere ad server, as there are many such choices available in the market. It makes us curious to get your thoughts: what would you call such a platform? Is it still an ad server, or is it a Real-Time Advertising Platform?
Now let’s talk about the 2 key real-time features inside SPX and their importance and association to the “need for speed,” as well as how they help a publisher scale without breaking the bank on ad-serving fees or on large ad operations/optimization teams.
The two key real-time marketing features we will talk about here are:
Sure, everyone in the ad-tech industry speaks to the fact that RTB is the perfect solution to maximizing yield. However, RTB is just one demand source. There are always multiple RTBs, direct campaigns and ad networks. Typically a publisher’s Ad Operations team traffics all demand sources inside their primary sell-side ad server in order to create a daisy chain or waterfall of these demand sources. It’s generally a very inefficient method for managing demand. Dynamic Demand is Smaato’s solution for this decisioning logic inside the ad server. It compares all of a publisher’s demand sources and makes a real-time decision on which of them wins. Think of this as a mini-auction that happens inside the ad server across all demand sources, be they Direct, RTB or Network.
Here’s a graphic that shows the inefficient “daisy chain” of demand sources:
Now, here is a graphic that helps explain how Smaato’s Dynamic Demand works:
Automated Floor Price Management (AFPM)
The second problem we’ve addressed in our platform is: how do you price each demand source for best possible performance based on eCPM? In RTB, we call this floor price. Having an RTB floor price doesn’t by itself solve the problem of pricing all other demand sources inside the ad server. If the demand source happens to be a direct campaign, you can easily price that based on agreed-upon price. However, with Ad Networks and AdX, you really are going to only learn about the price based on what they decided to pay you. This challenge requires yield optimization; however, it is something that is difficult to scale with humans if you’re looking for real-time results.
Smaato’s AFPM logic is a pre-auction price-setting algorithm that operates inside the ad server, and it looks at all demand sources and their ability to pay the best possible price for each impression before making a decision on who wins.
This logic has 2 key benefits:
1: The price curve changes in near real-time based on demand and eCPMs. AFPM looks at many variables around demand and supply in order to make a decision on the best possible price for each impression prior to the auction and winner.
2: The process of price setting for each impression eliminates the waterfall decisioning model, as it decides on a price and possible winner prior to even calling that demand source, based on an Automated Floor Price Management model. All of this happens pre-auction.
With the current setup of multiple arbitrage players, not to mention many buyers and sellers labeled as the “real-time” players, it is important not to get confused about the true future of real-time advertising. Real-time advertising over the programmatic pipes is not simply algorithmic decision making; it is also an advertiser’s ability to make clever, brand-specific and timely decisions. From a publisher’s point of view, the real-time controls are even more important because they need the ability to define the price for their content based on perceived value.
Smaato’s Dynamic Demand and Automated Floor-Price Management illustrate that an intelligent, beating, ever-learning, ever-improving heart can truly sit at the core of an advanced Real-Time Advertising Platform – and thereby deliver higher eCPMs for today’s publishers. We’d love to tell you more if you’re interested, so please feel free to get in touch.