Oliver Thylmann’s Thoughts

Two blog posts got me back on a certain subject I thought I’d write about again. First of all, Mark Cuban posted “I couldn’t resist … Youtube vs. Broadcast.com 10 years ago” in which he compares YouTube and Broadcast.com. Yes it is not a 100% comparison, but interesting none the less. Then I found “Ad Networks: Inventory vs. the brand” by Mathew Ingram (via Simon). From that post just see this part from the end:

Maybe it’s just the spillover from the sub-prime mortgage meltdown, but in some cases packaging remnant inventory and selling it through an ad network reminds me of the Wall Street practice of bundling underperforming or questionable mortgages together, and “securitizing” them in order to unload them onto outside investors. That kind of strategy works really well — right up until it doesn’t.

Back in the first bubble I was running a statistics service for webmasters and we were using Flycast, which very soon folded because that network kind of did not work. Next up for us was Tribalfusion, and they survived. One of the features I really liked about them was that I could put ads into their system on my site myself for free. Very cool indeed. Seems a bit like Google’s AdManager today. Then there was Doubleclick for the big sites, which was not just a service to use and pay for as a publisher or agency, but a real network selling your inventory, until they stopped doing it when the marketing money ran out. So we had networks, and we had targeting, and we had lots of reach and bundling over different bigger sites.

I do believe that we are running down a similar road and we’ll see if lots of those networks out there will survive today. There is a lot more reach out there of course but one of the biggest differences we have is that the cost side has really really changed. You can have more server power and reach for stuff like behavioral targeting and costs for ad servers has gone way down, as has bandwidth. This allows both better targeting and and created networks that can run on the unsold inventory, because even with several cents CPM you can still run profitably. Also there are more real marketing dollars out there obviously.

Sadly the latter gave rise to ads that are not something you want on your site. The thing is that often times, cheesy ads that have higher click throughs to worse conversion but yield higher ecpms in the end. That is why you have these weird flashing banners, or texts that are totally unreasonable like “You are the 1.000.000th Visitor and you Won!”

As always these are interesting times, and I 100% believe in networks with a clear focus, as I wouldn’t run one otherwise. But we are focusing on filling inventory with marketing budgets from small local players in a performance marketing manner, making you other money. Others allow for exploiting your inventory through behavioral targeting techniques, making you more money. Still others are bundling smaller sites, who are allowing them to make money at all. All have to watch out what they stand for and that they are really providing value over and beyond the other solutions out there. And all have to make sure they can compete with Google, on which ever level that is.

The first one I’d like to briefly talk about is AdScale from Germany, which is really another copy cat, this time from AdBrite. Having launched yesterday, it obviously still has very very little traffic on there, which is really the thing a network needs. But with one of their Investors, the European Founders Fund, has also invested in Performance Media, I presume that to fill up rather quickly, but this depends on how the two systems can really interact, and with AdScale needing their code on the site they serve banners on, this might not work out as (possibly) planned. We will see.

The interface is really nice and I can easily set up my own site to become a member of their network. I will then choose a category and also choose keywords my site is about. These keywords can then be booked by advertisers. Having gone through the advertiser process first I was wondering if they were reading out the content of publishers live and serving banners similar to Google, which would be a huge undertaking, but that does not seem to be the case. At the same time, this makes the keywords a little bit a mute point, and the problem becomes that with each publisher choosing 5 keywords for any one placement (I presume to be able to give more than 1 placement), the keyword density will just be too low for efficient booking on the advertiser side.

The really nice thing is that you can give AdScale a fallback ad. So you could say that you want a CPM of 5 EURs on your blog, as an average over the banners they serve, and otherwise serve AdSense. Again, the problem is that either advertisers book with very high CPCs or you they need very good targeting, both not really being the case from my first looks. They do have MyVideo and others (many investments by EFF and Holtzbrinck Ventures I might add, both investors in AdScale) but based on the millions and millions of ads MyVideo serves, the current AdImpressions AdScale has point to the fact that they are just booked on few sites per default and can book on MyVideo if there are deals in the system that MyVideo wants (becoming very similar to Performance Media). It’s really the chicken and egg problem that advertising networks always have. And remember that we had advertising networks en masse in the first bubble and they slowly disappeared again.

There is one much more interesting startup that plays with an idea that has been in my head for a long time. I talked about it in my post from July entitled We need a small little change to banner serving.

The problem I am starting to see, especially for the smaller sites, is that they will have to choose one network, like AdJug to serve their ads. Then they are quasi locked in. I personally want the highest paying ad running and I don’t care where it is coming from. I can now go into my installation for OpenAds on my blog and choose to run AdSense ads or AdJug ads or like I set it now, Ormigo Ads (just testing my own stuff here, nothing for your to test, sorry) when people come from Germany. It’s all sub-optimal though. Because for any given page, AdJug might have the better ads, or we might or somebody else might or I might choose to just run AdSense, or even opt to have a CPL AdSense campaign shown when it really fits super well.

What we need is a VISA kind of market for advertising. In general I would love to do this as a start-up, but the thing is that it’s not really a start-up as it is not there to go public or get sold or anything else to make money of as an investor. It’s a separate entity that will make sure that the ground rules are clear.

This is where PubMatic comes in. What they do is a first step in the right direction, again a lock in, but the right direction. You give PubMatic your accounts for YPN, Blue Lithium, AdSense and others. They will then monitor how much money you make when displaying certain ads and certain pages and in the end, will start serving those ads that work best for any given page. Now that is cool! That’s the performance bit that I believe in, and that makes it easy for a publisher to always run the best ads. The thing we now still need is an open specification, a cloud of ads somewhere, where I can gather information of what runs best and choose it. That’s the next step. The thing is that the publisher business is 99% about profit, but they will have to take some strategic decisions soon in terms of which networks they work with and which client relationships they want for themselves, that’s for the big ones at least. But I do love a company addressing the problem.

What do we hear there? TechCrunch alerted me that AOL seems to have bought Tacoda. The New York Post is talking about it too, citing prices between $200 and $300 million. the NYP article was actually linked right from the Tacoda Homepage. We had some turmoil some weeks ago in the ad server market and AOL only got the small player in AdTech (not the conference, the ad server ;)) And now it seems like they got themselves an ad network. Actually it really goes into the behavioral targeting world, about which  Jeremy Liew has some good thoughts.

The thing is though that this might hot up the behavioral ad market too.  It is important to note though that Tacoda also is a network, even though I am not sure how the network is booked, e.g. do they get the rests from publishers because they can monetize them better (Update: seems like it.), do they pay fixed up front and hope to make a margin, or are they first choice and simply really pay more. AdWeek has a good article on behavioral ads, in that they might even click worse, but have higher conversions afterwards.

The thing is that AOL already has advertising.com, and they are a network of sorts, and I don’t know what Tacoda ads to the equation. The thing is that some big publishers might cancel their relations with Tacoda due to a possibly buy from AOL. We really need an independant player in that market.  wunderloop might be one (Disclosure: Michael Kleindl is Chairman of wunderloop and Investor in Ormigo) or nugg.ad.

One thing is important though. Whoever gets to see people on niche sites, gets to target them on the big non-focused sites. So I might agree with Jeremy that that is the really important part for the behavioral market guys. Happy to learn more.

Now this was interesting. I posted about the Banner Server Market Changing early in the day and a few hours later, Microsoft bought aQuantive for $6 billion. This resulted in my post being quoted on alarm:clock and linked on techmeme.

The interesting thing is that there is no big banner server provider left really. There is Tribal Fusion as  Venturebeat posts. That doesn’t really fit though I have to say. Tribal Fusion is not the same as DoubleClick or Atlas or Ad Tech. Those are banner server providers and Tribal Fusion is a network, so it would not be what Microsoft needs. As I said in my last post, there are a few small ones left, but no big ones with huge customer contacts and big potential ad impressions handeled. Remember that DoubleClick does not really sell ads on the spaces it handles on publisher sites, but is just a technology providers. Same for Ad Tech bought by AOL and same for large parts of aQuantive, Atlas namely.

As some have already put it, this is for MS to stay in the game. They don’t want to go to Google, AOL or Yahoo for running banners on their properties. So this was a must buy. Just imaging Yahoo! buying them or another agency like WPP buying 24/7 Real Media just now. This would be hell.

The best comparison of the buying action I have seen on Pauls blog coming from zenrob. The important thing is the money spent per ad impression served.

• Google/Doubleclick: $0.0107
• WPP/24/7: $0.0032
• Microsoft/aQuantive: $0.0272
• Yahoo!/Right Media: $0.0142

For me the Right Media deal doesn’t really fit in here because this is a mix between a network and a service providers, really more of a market place and actually very close to my VISA model, so I do really like the Right Media buy. Best one of them all as a gut feeling.

But anyway, as you see, Microsoft paid a real premium for aQuantive, but they did so to have a very big foot in the market, and really in three. For one they now have Atlas as a banner server system for agencies and publishers, DRIVEpm as an arbitrage system buying ad spaces in bulk and then running performance based campaigns over them (can be very lucrative business, I know, but might become dangerous as people want to buy into those spaces that publishers could get to buy standard CPM deals, but that’s another post), and avenue a razorfish an interactive agency. Microsoft is now a bigger ad play than they were before and they need that knowledge to exploit the hell out of all the properties they will have soon. DoubleClick on the other hand was bought for the relationship with agencies only, because the technology is just not that amazing (actually the interface was so damn slow when I used it it was a pain).

If you want to read a few more posts I suggest going over to this alarm:clock post with a roundup of views of other people.

And by the way, I think DRIVEpm is a large part of that deal, because it is how you can push search ads on sites. But let’s see where it goes.

(Update: A few hours after I wrote the following, Microsoft buys aQuantive, one of the last banner servers not owned by the big ones. See more at Techcrunch. I’ll keep the article as it is for now It seems that all the big ones now have their Banner Server. )

There was a time when banner server companies were something special. You needed a DoubleClick (bought by Google) because if you used somebody else, the counting differences between the publisher banner server (DFP) and the agency banner server (DFA) would be too big and result in friction. I never really understood that part because it can be reasoned away, no longer being a problem, but it seems that it is too much of a pain to use a sometimes better banner server.

But I am loosing focus. More Banner Server Providers came along like Falk eSolutions (bought by DoubleClick) whom I met ages ago in a pitch at the place I worked at back then. I have to say that they were technologically superior … oh but the counting differences. Then you have Ad Tech (now bought by AOL) and just yesterday WPP acquires 24/7 Real Media. With all the “bought by” behind the names it becomes clear that Banner Serving is a hot topic at the moment and that it will likely move to become even more of a commodity. After the last bubble (no we don’t have a bubble at the moment) prices have dropped considerably for what you pay in CPM for delivering your ads. The problem of course is that the potential companies to work with are thinning slowly bug surely (unless you want WPP, Google or AOL be your Banner Server Providers ;)) so these prices might increase again. A few are left though like Adition here in Germany which seems to be a good solution, and of course OpenAds. Then there is the mightly aQuantive with Atlas Solutions and also emediate. There are likely others I forgot (leave a comment) and of course there are lots of targeting companies at the moment that just plug in to a banner server, like wunderloop. Above that there are the market places like Right Media (bought by Yahoo!) who have a great demo by the way!

Lots of companies out there, all growing well, and things are moving closer together. Companies are slowly seeing that internet is core to what they do and they are investing into that. The best sign of that is WPP going into the Banner Serving Space.

The problem is that simple banner serving is no longer special. Service is what distinguishes and special features above the standard. Having built one specialized banner server already, and now building another system I would call a solution rather than a server, I can tell you that building your own can make sense in many cases. Especially if like us you have the knowledge in-house including one of the original Falk eSolutions developers. This is especially true if you need something different than a standard top of the fold big size banner.

And this is where the space will likely be moving in the future. We need APIs more than ever in this space as there will be more small systems, specialized for the portals they are used on, as well as some big players with lots of power behind them. We need a real market place not created for making money but for facilitating performance. This is a performance play people! We can measure things and as soon as we can we do we are moving in the performance mode. Currently this is the internet space, but it will become standard in many other places because rather sooner than later (probably later and bigger because we always overestimate but that’s another point ;)) lots of things will become measurable. Digital Paper, TV with a backchannel.

I really see a VISA like structure to be established as a non-profit. Problem is that only the good ones would win. Transparency would be king. And we are not there yet. We are still in the wining and dining business and it’s probably going to remain that way for some time. But it’s good seeing the buying action because it shows that things are changing and ad serving is becoming central to the strategy of the big ones. That opens us up for new thinking. Looking forward to it. But first I need to revolutionize the local market.