Real-Time Bidding: The Ad Exchange Leaps

Real Time Bidding or RTBWe have exchanges. We have trading. We have media. And, we have data.

What are we missing?

How about “Real-Time Bidding,” or RTB, for short?

With RTB, exchange participants have a new feature coming for the buy side which should make agency business intelligence mucky-mucks and advertisers delirious after watching the development of yield management tools for the publisher or supply-side.

Clearly though, RTB benefits both buyers and sellers.

The Future Is Not Reserved

I can hear some of you… You thought the exchanges were already “real-time” and, to an extent, that’s true. At the very least, if you think of an ad exchange as an ad server, well, then the ad gets served in real-time – as soon as the page is opened by the end user. But, no no no, we’re not talking about that…

For the most part, it’s been a “futures” or “reserved” world to-date where buyers of online display ad media bid on publisher placements in the future. And, the advertiser must blindly trust, for example, that certain targeting parameters for any buy are implemented correctly by the selling publisher or ad network.

What RTB will enable with ad networks and exchanges is:

  1. The purchase of individual impressions using the “buy” side’s data to analyze the impression and bid accordingly in real-time.
  2. The optimization of real-time bidding using real-time conversion data.

As RTB is solved through technological innovation and the gradual adoption of industry-wide standards, performance metrics could leap.

As an industry insider told us, “Demand-side optimization creates a more dynamic environment with more efficient and fewer pre-defined rulesets.” Advertisers will be more willing to buy media – and bid higher for media (making happy publishers) – that they believe will likely perform well given the efficiencies created through RTB. Proprietary data from advertisers can be mapped against any potential media buying campaign to further improve performance.

Many companies are running hard to provide real-time bidding.

In recent weeks, a very limited number of ad technology companies now offer or have said they will start offering RTB. Agency/advertiser trading platforms can’t wait such as Havas’ Adnetik, The Media Kitchen’s Varick Media and Publicis’ Vivaki Nerve Center.

This is no small challenge as today’s NASDAQ stock market, for example, is a model of how buyers and sellers bid in real-time on the stock of public corporations – this innovation likely cost many millions of dollars. Technology companies working on this challenge are still struggling with the intensive data calculations as bidders and impression levels increase in addition to other data parameters.

No doubt, the implementation of RTB will accelerate adoption of ad exchanges and digital media trading – the open auction of online display advertising.

And, any technology companies that can define or discover technology for RTB stand to enjoy a healthy payday.

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!


  1. Very well done. A lot of folks in the industry don’t quite get the distinction that you have made. With Real time bidding, it really makes ad buying a fully transparent process where the ad buyer gets to see every impression, the url, the user, the Geo, the context, any BT data, and historical performance data to make an educated decision about what the impression is worth. What has kept us from this for so long is that this decision needs to be made in 50 milliseconds and is a pretty tough technical problem. It seems like we might finally be there. I am certainly pretty excited about it. Keep up the good work.

  2. I’m a bit of a skeptic as it relates to how widespread real-time bidding will become. As Zach notes, there’s a significant cost involved with making the quick decisions required. Not only do you need the hardware to respond quickly, but your models need to be constantly updating to fully exploit the opportunity.

    Even worse is the fact that you get penalized for being a picky bidder. You have to look at most impressions, but will only bid on a fraction of them. Your hardware needs to scale with the size of the inventory you’re addressing, not the inventory you buy.

    This adds a lot of frictional cost that, for many advertisers, will likely outweigh the ROI benefits gained by bidding in real-time versus updating bids every two hours. Those frictional costs are significant. If there are 20+ buyers in the auction, you’re buying ~20 times the hardware to support real-time bidding versus updates every few hours. (Yes, there are scale benefits, but still real economic problems)

    Does anyone know of a technical solution that addresses this problem?

  3. That is a very good point Tim. Another issue is bandwidth, as you scale the number of bidders in the system, you also exponentially scale the number of messages that are being passed. The solution that people are thinking is the answer to this is to use a cloud. That way it is just a question of scaling instances rather then servers, and bandwidth costs are minimized. It is not a trivial problem though. But I take heart that wall street seems to have solved it at scale. Though I do wonder what it costs. I imagine that the end solution will be professional media buying companies that handle the real time bidding for a enough clients that it makes economic sense.

  4. I have to agree with Zach here. While the technical challenge is great here, there is benefit to performance oriented networks and agencies here that will surpass any hardware and software costs. This is of course assuming that the agency in question is capable of reacting in real time to the inventory changes…

  5. Ohad is right, the buyside of the equation has a lot of work to do in order to be able to make valuation, targeting and optimization decisions in 50 milliseconds. Given that most ad buyers still use a fax machine and a spreadsheet, it is going to be quite a leap for some folks.

  6. Lots of blue skies with the new approach and definitely where the industry is going.

    We have yet to make progress on co-hosting servers and ensuring the efficiency remains as Tim mentioned. ROI will drastically get impacted by introducing the infrastructure to support RTB.

    I find it that most company requesting the functionality are a bit confused on what exactly they are in need for.
    Segment Bidding and a better transparency into the bid landscape can solve most of their needs.

    Furthermore, not many are getting their advertisers input in real time, usually there is a 12 hours delay. So what ended up being a real time decision process, now becomes a 1 hour cherry picking and 11 hours of risk taking and frequency management.

    ON the optimistic side, we are making a great progress as an industry to treat Media and audience as commodity, where the premium buys are futures and contract negotiation becomes futures.

  7. Eran,
    As we discussed at the Clift, the big issue is visibility into where the ads are running on a per impression basis. Allowing those of us on the buy side to see where the ad is running and to who enables us to much for effectively price the value of the impression and pick the right ad to run. The only way that I can understand that this full visibility will occur is with RTB.

    I think the reason that you haven’t seen folks on the buy-side with the capability to handle this sort of bidding is that unless you control a very large network there in not much incentive to build the very complicated tech that is necessary until the RTB exchanges real scale. Thankfully that is finally starting to happen.

  8. There are many limitations, but also great opportunities in real time bidding. A great opportunity is to change bids immediately after a new competitor comes in, or an old one disappears. Also, it handles traffic spikes due to click fraud very effectively.

    Drawback: if revenue per conversion takes 2 days to settle (this was the case with one of my clients), real time bidding can be used only to filter out fraud and update ad position on page, but not for ROI optimization.