Get More with No-Cost Ad Serving, Maybe for Publishers

Well you knew I would be coming at this one.  Bennett Zucker’s February 16th, 2007 Get More with No-Cost Ad Serving article in iMediaConnection throws out the notion that the value of ad servers is fading out of existence.  While Bennett fails to be explicit about whether he is specifically looking at publisher ad servers or advertiser ad servers, let’s examine both from the perspective he presents.


Bennett says: “no-fee ad serving can bring about technological innovation, and better, cheaper, faster ways of getting things done online.”  True, ad serving has become an increasingly commoditized service.  Zedo and Mediaplex are the two biggest drives of this trend.  Zedo’s approach to giving away three-months of free ad serving and then converting customers to paying clients results in very cheap ad serving.  And Mediaplex is ‘buying’ business aggressively in an attempt to capture market share from DoubleClick and Atlas.  It is not uncommon to see $0.05 or $0.04 CPM rates from either provider for tiered impression volumes in excess of 100M/mos.  I’ve seen Mediaplex go as low at $0.035.  Hard to compete with.


Zedo is easier to compete with.  They are not certified to serve on as many sites, their customer service is not rated that high and their tool is buggy.  Mediaplex has its issues too.  I’ve gone into the competitive features of different ad servers in the past in my “How to pick an ad server series:”


1.      The Ad Server’s Point of View – Selling to the Advertiser and Cost

2.      How to Pick an Ad Server – Part II Interface Evaluation

3.      How to Pick an Ad Server – Part III Rich Media, Targeting & Optimization

4.      How to Pick an Ad Server – Part IV Price

5.      How to Pick an Ad Server – Part V Training

6.      How to Pick an Ad Server – Part VI Customer Support & Implementation


“Today we assume, for example, that it costs nothing in bandwidth or storage or processing power to add new subscribers and web pages. But this notion hasn’t been widely applied yet in the realm of ad serving.”  Actually Bennett, this is false.  With ever-increasing files sizes, rich media and video this is the one thing that is keeping CPMs from falling lower and, in fact, this is what is starting to drive CPMs back up.  Bandwidth is a fixed cost for an ad server.  Granted the more bandwidth you require the better your rate, but a pipe is a pipe and you pay for the bandwidth in chunks no matter what you push through it.  The operating model of an ad server is based on what sells and most ad servers have a minimum CPM that they need to cover to break-even.  This number increases as the files sizes creep up.  Video, especially, introduces a whole new level of bandwidth requirements like rich media.  I know people paying $4 CPMs with Eyeblaster if you can believe that!  Granted, they’re getting hosed.


Publisher ad servers have an interesting situation to consider, and the Google proposition is one that is knocking at their door.  If Google is going to offer free ad serving in exchange for joining the AdSense network, then you do have a problem for the ad servers.  This may be a very enticing offer to a publisher.  Eliminate the $0.08 DFP charge and that could be worth $100K/mos or more to the publisher’s operation.  It could be worth a lot more if they’re a large web site.


So how do advertiser ad servers stay in the game?  For one thing, what are ad servers really all about?  Is it posting ads, rotating, optimization and selection logic?  Really?  Come-on.  There is a reason why there are out-of-the-box ad servers like Ad Juggler and freeware php-Ads-new.  Okay, I think of them on the same level, you may not.  Because it’s not hard to upload ads to a server and select creative and post them.  Adteractive runs over 1 billion ads a month on a home-grown system.  All you need is scalability.  Servers.  Co-location helps so that you have the security of not worrying about going down regionally and you have the load-balancing geographically and potentially internationally.  Those are only subtle differences between the major ad servers and the tier-three players. 


Self-managed out-of-the-box ad servers can’t get certified on the Yahoo!, AOL and MSN’s of the world because of reasons like that.  So if you are using those solutions, you’re out of luck there.  But that is not necessarily an issue for all advertisers.


Ad servers for advertisers are in the game for log processing and reporting.  That is the meat of the game.  The ability to process 70-100 million logs per ad served and return impressions, clicks, post-click events per site, site-section and placement is the heart of what ad servers do.  Taken further, the better ad servers offer robust reporting and analytics that enable advertisers to cross-dissect the data to make empowering decisions.  This is where some of the real differentiation comes into play.


A second place is integration.  Ad servers are starting to step up and recognize that 2007 is about consolidated reporting.  Drawing together disparate data sources into the dashboard.  Blackfoot has been touting this for 2 years.  David Smith has been challenging the industry for over a year to bring this to light.  Ad servers are slowly coming to recognize that this is where they need to be. 


TruEffect began a project with mOne a year ago to build a consolidated report that brought together the search, email, rich media and ad serving data under one roof, within Microsoft Excel.  Nothing entirely different from what some of the other ad servers are capable of doing accept for the fact that it is active, live data that is manipulatable within a Microsoft Excel environment, instead of within a Web-based application and downloadable as .csv files.


Now I know first-hand that publishers are dissatisfied with their publisher ad serving solutions.  See my post: Banner Ads on Google.  And I know that there really is not a comprehensive solution out there that combines inventory management, sales management, forecasting (the wholly grail) and campaign management.  So maybe Bennett can shed some light on that in a follow-up article or perhaps Brad Beren’s can find someone to write a non-self-promoting article on the state of affairs of the publisher ad serving space.  But for now, the largest pubs out there are using a combination of home grown systems or a combination of DPF plus site-side analytics plus proprietary tools plus a room full of analysts; and all of them are blind more than just a couple of weeks out in terms of available inventory.  Crack that code and you get an award.  And a job … probably from Google.


Differentiation is key in every space, no doubt Bennett.  And the key is to make those differentiations known in the industry.  DoubleClick is asleep at the wheel.  Atlas is focused on their purchase of Accipiter and has let their client services slip so badly that their clients are just crying for someone to come and take them away (oh Calgon!).  Mediaplex is aggressive out there, no doubt.  But all they have to offer is a shelf full of the features that you should already expect from an ad server and a price that is intended to undercut DC and Atlas.


And now, the plug J … TruEffect has all the features, and yes, we’re competing on the price points too.  But we built the ad server within Microsoft Excel, TruAdvertiser.xls.  See Maximize your Ad Traffickers’ Value, Re-Evaluate Their Time for a deep-dive into the product with screen shots and all that good stuff.  I know, I named the post poorly, but there is a lot of meat in that one.


Anyway, TruAdvertiser.xls was built around the workflow of an agency or advertiser.  It integrates with other aspects of the business operation, leveraging all of Microsoft Office.  All aspects of the media planning, ad serving and reporting is conducted in Microsoft Excel and other disparate sources of data can easily be pulled in to create dashboard-like reports.  Data can be pushed out to accounting systems or other tools like internal business intelligence reporting tools.  There is not a whole lot that can’t be done with some specialize work.  And we do it everyday with our clients.  Okay, enough … that wasn’t bad (166 words).


So Bennett Zucker wrote us another good article.  And from the publisher ad serving perspective I think he is spot-on.  Commoditized pricing is an indication that there is little differentiation in the space.  If it is just about price, the players are driving head-first into the ground.  If someone does not crack the forecasting code it will remain a one-way ticket.  Google’s attempt to throw their hat into the game with their free publisher ad server for AdSense doesn’t even bother to take a swing at it.  My conversations with them last year yielded that they too didn’t have that problem solved.  So if you want to win, and maybe I can help you here, launch a publisher ad server with forecasting that can look out past 2-3 weeks with a degree of accuracy that can be leveraged for selling.  That is the golden egg.  In terms of advertiser ad servers, it’s not about features, everyone has them and that is why the prices have fallen.  Rich media, video and larger file sizes will drive the CPMs back up.  Integration and consolidated reporting will keep the leaders alive if they don’t stay asleep at the wheel.


Reactionary with Insight

_uacct = “UA-980395-1″;

How to Pick an Ad Server – Part III: Rich Media, Targeting & Optimization

Rich Media is an established component of media buys today.  Having rich media capabilities is a requirement of an ad server.  Eyeblaster and Pointroll are different than flash and shockwave and most buyers want these two capabilities to come together.  DC is the only ad server that has integrated them to any extent with Motif.  The problem is that you have to build your rich media using Motif and then you can’t use the media on any other ad server other than DC.  That means you’re stuck if you use more than one ad server.  Word too the wise, many agencies are forced to use more than one ad server because clients have contracts with ad servers too.  Building on proprietary platforms like Motif can cause you trouble when it comes time to having to run creative on another ad server.  Just because the other ad servers don’t serve Eyeblaster and Pointroll doesn’t mean that they can’t measure, track and report on them.  Most of them can.

Targeting is another inherently essential component.  When evaluating the ad servers make sure that you can do all of the most common forms of targeting.  Geo should be at least DMA (down to the city level) from a reputable source and should also extend to whatever global region you require.  If an ad server does not cover a region you need, ask if they will add it as it’s not hard to do.  Ad servers that are willing to accommodate your needs during the buy cycle can be a great sign of flexibility and indicative of a potentially strong working relationship in the future.  “We don’t do that,” reactions are a sign of inflexibility and will resurface in the kind of customer service you will receive in the future.

Ask about cookie-targeting as well.  This is something that is glossed over the most frequently.  How an advertiser or agency can leverage an ad server’s cookies can be pretty amazing.  The Big 3 will charge you handsomely for custom cookie targeting so watch out for the hidden costs.  But a company like TruEffect and [X+1] who leverage cookies in their technologies can really offer great customer targeting features.  Behavioral targeting is event-based targeting, so call it what it is.  I’ve covered this is in other posts.  See:

• “Audience Screening is also Behavioral Targeting, Event-based and Anonymous Targeting
•  “What’s Wrong with Behavioral Targeting

Using pixels to track someone’s behavior on an advertiser’s site or on a network, and then targeting that behavior with an ad later is valuable.  The results have proven some up-tick in performance but Boomerang and Atlas DRIVEpm charge a premium for this service and the ROI is not there for many advertisers.  In my post:: “Why Am I So Frustrated with Behavioral Targeting – Do I Really Know Something,” I introduce re-targeting.  First Party Ad Serving, what TruEffect calls DirectServe, allows an advertiser to recognize its existing customers, registrants or users while advertising online so that they can be distinguished from an unknown individual and messaged to differently in real-time.  This is not event-based targeting but re-targeting.  Behavioral targeting/event-based targeting/re-messaging should be a common feature.  Re-targeting is a patented capability of TruEffect so forgive me for the plug.

Optimization capabilities are very important.  I have talked in previous posts about auto-pilot and its abuse.  See: “Site Optimization and Ad Serving.”  But the premise of site optimization, being able to weight rotations, placements and banner plays is a must.  All of the dominant ad servers offer this capability.  The only ad server that I hear problems about is Zedo.  Because their ad server appears to be a publisher solution twisted around to serve advertisers and agencies, the ad serving logic does not work properly and the optimization does not work.  I have had clients tell me that when they set their prioritization schemes, Zedo still plays things wrong and poorly performing placements still get dominant exposure even when dialed down.  So ROI–optimization does not work properly, yet.  Auto-optimization features like impression-based, click-based and post-click-based should be available with whatever ad server you choose.  Just don’t intend to turn it on and walk away because that is not what it is meant to do.  You do have to do some work when you use an ad server, that is, if you want optimal ROI.

_uacct = “UA-980395-1″;

How to Measure Video Performance – I inspire People

Bradley Werner took a chapter from my blog in his article How to Measure Video Performance today.  I guess he, like me, is trying to see where the light is with video.

The article highlights the basics of measurement.  Did we reach the right people?  Did we affect the audience the way we wanted to?  He asks the esoteric questions correctly but then only gives us the most basic concepts of impressions as clicks to define them.  Blah.  We need more power Scotty!

Bradly does talk about percentages of video views which is more interesting, but that is a whole lot like an impression.  The advanced measurement is the “…Interaction Rate. IR is the sum of all the interactions, divided by the number of impressions. These interactions include the click-thru’s, replays clicks, unmute clicks, clicks on all the add-on features (submit info, download, watch video 2), clicks within a game, rollovers, etc.”  This is the only meat we get from the article.

So my inspiration brought on a flounder.  I will keep looking for you, but for now the search goes on.  If video is to cross the chasm and bring us the next great thing, we need to see measurement capabilities that are a fusion of broadcast and display.  It has to be rich media to the nth degree.  Interactions plus playtimes plus the ability to place a marker on the user so that we can later recognize, distinguish them and target them in the future as with display advertising.  The golden nugget.  We’ll find it.  Java calls it a muffin.  I wonder if video will call it a scone?

_uacct = “UA-980395-1″;