What is programmatic advertising ?
When companies like Yahoo! started off the concept of advertising was very new. Someone somewhere compared the "page views" to front page, inner pages and other areas in the news papers and niche magazines the age of internet advertising was started. I am not sure where and when this started so I will skip that part. Essentially, we had images or GIFs shown in reserved spaces of numerous sizes. Deals were arbitrarily inked based on the skills of the advertising sales people. This era can be seen as the dawn of "banner" advertisement.
Soon companies started giving out specially targeted advertisements and charge based on regions, time of the day etc. Various standardization efforts like IAB has created definitions, fair pricing strategy and sizes like 300x250 for the ad-banners.
Companies like Google soon started targeting based on not just "geo-fences" but they started grouping users according their interest, behaviors and income groups. As the internet usage exploded, the available "real estate" for showing advertisement increased.
From lets say, mid 2000s we have a scenario where number of "advertisement buyers" aka ad-agencies are competing for premium targeted space. At any point in time there were way too many advertises (Buyers), Sellers (publishers) & Middle-men.
The number of parameters to choose from increased. Now before showing every advertisement, those who serve the ads were comparing against
- geo-location
- time
- income group
- age
- sex
- sexual orientation
- buying behavior
- buying behavior of friends and family
- events in life
And much more decision parameters. Numerous services popped up offering speed, efficiency, anti-fraud technologies. The result was the simple dynamic banner ads replaced by persolized ads matched against numerous rules and selected from a group of competing publishers.
The rule engines, data processing engine for collection and aggregation, matching and finally delivery where all combined and addressed as Programmatic Advertising.
Real Time Bidding (RTB)
An extensive list of technologies in and their definitions can be found here : https://cloud.google.com/solutions/infrastructure-options-for-building-advertising-platforms#terminology
. As the complexity of serving ads as well as the competition increased, fast, real time decision making and ad-serving using AI, CDN and massive data sets decided the winners. Google, Appnexus and The Rubicon Project emerged as the winners. Needless to say Facebook and Atlas Platform has carved its on niche among all this.
In my view the entire ad-serving still remains looking at the server logs, comparing against existing data and making decisions in near-real time & is defined as Real Time Bidding.
Header Bidding
-- The RTB infrastructure has some of most complicated looking architectures one will come across. They store enormous amounts of data, make decisions at the zillionth of a second, but still the experience was poor and slow.
(Well, just to scare!)
This is when a simple mechanism to contact the backend servers before the content of the page is loaded was devised. This simple "trick" ensured that before the content is even transferred to the users browsers the decision on which advertisement to be shown was taken.
The Steemit Potential
The portal Steemit.com provides a unique audience which can be categorized as early adopters, innovative at the least. Obviously further classifications can be made. Though I don't have the exact numbers, looking at the pubicly available data, steemit.com is a very lucrative medium for the advertisers.
How to do it ?
On a high level, RTB with Header bidding enabled with pass-back tags is the way to go. RTB and Header Bidding are no-brainers. Making use of the pass-back tags ensures that we have a high efficiency interms of how many percentage of the page views has relevant ads. Needless to say, a property like Steemit deserves very high, say, 70 - 80% fill rate. In simple terms, 700 to 800 pages out of 1000 pages will have advertisements.
Where to start ?
My personal favorite is TheGuradian.com. They have an excellent UX & ads are seldom shown and are mostly relevant.
(1. Top banner, 2. The header bidding is activated)
A summary of the ad providers are shown below. I think Google is shown as the last one due to the low rates. The comment also talks about not using a bidding wrapper. My best guess is they are handling the header bidding manually and choosing the networks based on per-determined priorities as opposed to 100% programmatic. I feel this is the BEST approach if we have someone constantly monitoring and tweaking the ad-delivery performance.
Quick Analysis of Steemit.com
Currently steemit.com is configured to show a top level horizontal banner and a Below The Line (BTL) vertical advertisement. Its important to move the vertical banner to the top of the page (ATL) to ensure better rates. Considering the fact that we can't show 300x250 ads easily, this is a must IMHO.
Further, we need to enable Header Bidding right now !
Without any specific setting I am seeing the top banner not filled. One strategy is to keep the rates low for the top banner in the RTB or create a private market place in the SSP platform and "lure" advertisers in. Once the AI networks are aware of the advertising space with specific targeting and demographic, we can increase the prices and also go for in house ad selling.
Paid Content
There are portals like Earn.com doing peculiar paid content and advertisements triggering engagement. Such a model is IMHO the best model for Steemit.com as we have vibrant community.
A common model of paid content is the following from The Guradian:
Selling of Merchandise:
Selling of items/SKUs/utilities interesting to the target audience using https://www.stackcommerce.com/ is a method used by Slashdot.org Personally I have found this as the least annoying advertising means.
Conclusion
Targeted and useful advertisements can bring in revenue to a portal with such unique, organic content like steemit.com I hope this article gives pointers to anyone interested.
Actionable list
Here is list of items that can act like a quick reference. I will update with any inputs in the comments too.
- Enable Header Bidding
- Get approval from multiple ad-networks : Appnexus, Rubicon project & others
- Optimize for maximizing the fill rate
- Use header bidding script in a "water fall model" as opposed to a wrapper (@techslut suggested "waterfall")
- Google Adsense if using should be the last pass-back tag in the waterfall to maximize fill rate
- Optimize the page lay out for more Above the Line ads
- Change the layout for ads in the promoted content: More like Native Ads (@techslut)
- Integrate paid content model
- Integrate with StackCommerce or similar services
- Explore models like Earn.com