Notice: This blog is no longer updated. You may find a broken link or two

You can follow my new adventures @mikeonwine

Saw this John McCain ad on a random blog this evening:

Reminds me greatly of the giveaways you normally see on myspace… seems they forgot to leave out the “get a free lecture on public policy!!”.

Landing pages even look the same — a simple form with a ZIP & email. I wonder if they’ll sell my email address for $2.00 and use my Zip for later retargeting later.

What’s interesting though is that it looks like the McCain campaign is doing some serious remarketing on “clickers”. There are pixels there for the Right Media Exchange,’s Leadback and Atlas!

I was just browsing on my gender test blog post and noticed the following ad:

So normally you’d say — so what mike, it’s just an ad for senseo! Well this ad is special because I was just searching for Senseo coffee pods earlyt his morning (both via Google & I find it highly unlikely that this is a pure coincidence. Here’s the clickstream (broken into multiple lines and shortened for legibility):[...]

Funny that there is a nice “q=a-googlewomen” inserted into the landing page — I swear the gender test showed me to be a man!! So this is all purely speculative of course, but if this is indeed true then we suddenly have the world’s best behavioral display network to compete with.

In my last post I talked about redirects and the limitations they posed for online advertising. In this post I’m going to discuss the various workarounds that people have put in place to address the problem. I’m primarily going to focus on serving speed & the “one-way” challenges of redirects as there isn’t much that can be done to address the fact that redirects are inherently insecure.

Fundamentally when you are talking about “integrating” two different serving systems there are two key driving factors: tracking information and influencing the decision. In most cases tracking information is for either transparency or behavioral information. Influencing the decision is also often related to behavior as it’s close to impossible to transplant data from one cookie domain to another. So let’s look at three little ‘hacks’ that people do to make systems play nice.

Replacing a redirect with a tracking pixels

One of the primary reasons why agencies operate their own adservers is to be able to track the # of impressions that their publishing partners are serving. Yet a much simpler method of counting ads that doesn’t involve an entire redirect is to place an “impression tracking pixel” that the publisher (or network) shows on every ad call.

Let’s compare how an impression-tracking pixel impacts the serving speed compared to a more traditional redirect If both a publisher and an advertiser are using a serving system and are using standard redirects (or ad tags) to integrate, each ad call will result in three consecutive requests:

Call 1, Publisher’s Adserver. The first call obviously goes over to the publisher’s serving system. When the publisher’s adserver chooses an adserver he returns to the page either a 302 redirect or some javascript from the advertiser’s serving system. Effectively redirecting the user.

Call 2, Advertiser’s Adserver. In this call the advertiser receives the request, logs an impression and then spits back some HTML (or javascript) that tells the browser to download the ad, generally from a content delivery network (CDN).

Call 3, Creative Content. In the third call the browser requests the actual GIF/JPG/SWF file from a content delivery network and displays the ad.

If each call takes 250ms the total time to serve this ad will be 750ms as no call can occur until the prior one has fully completed. The browser doesn’t know where which advertiser adserver to request content from until he has fully received the publisher’s redirect and then the actual SWF file can’t be downloaded until the advertiser’s adserver has responded with the appropriate HTML.

When an advertiser passes the actual creative over to the publisher and includes an impression tracking pixel the ad-call sequence changes a bit.

Call 1, Publisher’s Adserver. The first call is still to the publisher’s serving system, except rather than returning the actual ad-tag (or redirect) to the advertiser the publisher’s system returns two things. One a call to the actual creative (SWF/GIF/JPG) hosted on his CDN and then a second pixel call to the advertiser’s impression tracking pixel.

Calls 2&3, Creative & Tracking Pixel. Because the user’s browser received both instructions it can now proceed to download both the creative and the tracking pixel in parallel.

If each call takes 250ms then the total time to serve this ad will be 500ms, 250 for the first publisher call and then another 250 for the creative & tracking pixel in parallel.

Benefits: The biggest benefit of the impression tracking pixel approach is that it decreases the time to serve an individual ad — something that will reduce discrepancies and increase the quality of the end user’s experience.

Drawbacks: Of course there are drawbacks with this approach. Trafficking both a creative and it’s associated impression tracking pixel is more work than a simple ad-tag. It also requires more coordination between the advertiser and the publisher

“Forking” an ad-call with an impression pixel

Impression pixels do not have to solely exist instead of a redirect. In fact, they can also be used to supplement existing ad-calls and let third parties “listen in” on an impression stream without giving them the power to manipulate an ad. For example, let’s say that a network uses the Right Media ‘NMX’ platform for their network adserving but is unhappy with the amount of information that he receives from the reporting systems. Said network’s advertisers demand more information from their media-buys — what can he do?

Obviously one answer is to add a redirect to every ad call — this quickly becomes complicated though as the network will have to host the actual gif & flash files, somewhat defeating the purpose of using a 3rd party network adserver. Also, many publishers won’t accept third party served ads from unknown adservers, posing an additional challenge when trying to buy tier-1 inventory.

Another way to address the issue is to insert a tracking pixel into every creative. On every ad-call the network fires off a pixel to his own tracking server which contains the key information that he is interested in — say — creative ID, publisher ID and referring URL. He can then log this information and provide the advertiser with the detailed reports showing exactly where his ads appeared without having to reinvent an adserver!

Using AJAX to load 3rd party content

AJAX has gained a lot of popularity recently. One way that people can integrate two serving systems is by inserting a small bit of javascript that loads the 3rd party content required to either track information or make a decision directly into the browser. For example, let’s say a network wants to know whether a behavioral data provider has valuable data on a user. If the behavioral partner he receives the impression and if not the publisher sends the impression over to one of his regular networks. He could insert a small bit of javascript that sends a request browser-side and loads the behavioral advertiser’s cookie data from the third party domain.

There are some benefits to this approach — timeouts could be set on the actual request limiting the potential slow-down. Also, since the third party isn’t in the redirect stream he doesn’t get full control over the entire impression. The downside is that the decision is made on the client side. This introduces a whole new set of challenges around tracking & writing cross-browser compatible JS.

All these methods are hacks

You’ve probably realized by now that most of the workarounds involve tricking the browser to load some content from different serving systems. Some schemes that I’ve seen get incredibly complex. I remember a customer who wanted to use Right Media’s serving system as the source of inventory but then inject behavioral information from his own cookie domain. The end result was that impressions came into the RM adserver, were redirected to his adserver and then were immediately redirected back to the RM serving system. That’s three ad-calls before an advertiser had even been selected!

That’s about it for now … Stay tuned for part III….

Not too long ago I wrote a post One buyer, many cookies, now what?. In that post I promised that I would write about how buyers can best operate in the coming market with multiple platforms — and hence multiple cookies & adservers.

A quick refresher

In prior posts I have stressed the following:

  1. An exchange or ad platform, fundamentally an adserver
  2. Adserving is still tied to one cookie.
  3. Behavioral targeting is tied to knowledge about a user
  4. Knowledge about a user is stored in the cookie (whether as a unique user id or the actual data)

If that didn’t make sense — go read some of my older posts first.

Behavior is the future!

A single exchange with a single cookie space would have enabled true global internet wide behavioral targeting. Lets imagine I wanted to remarket to users. All I would have had to do in a single platform market is add my visitors to one behavioral segment on one exchange and then place buys on that segment. Sadly this is not likely to happen since the three major internet giants swooped in and acquired practically every adserving/exchange/marketplace up for grabs. So what would I do in a world with three platforms? Lets talk about some options.

Option #1: A global user-id

Imagine this — an independent entity that offers a global user-id database. Every marketplace, ad-network or exchange subscribes to this UID service and syncs their user-ids with the global user-id. So even though Google might think that you are user #12345 and Yahoo might think you are user 54321, I can use global UID database to map your mikeonads UserID of #164 to Yahoo’s 54321 and Google’s 12345. I then simply signal to Yahoo that user 54321 is a user and similarly to Google. Now each exchange knows who my users are and theoretically I can then target campaigns to my users.

Chances of this happening? Pretty close to nill. First off you’d need an independent company to provide the global UID space because none of the giants would want to give up control to another on this front. Then this conglomerate would have to get past scrutiny from the FTC, FCC, FBI, CIA and who knows who else before being able to launch, and then finally somehow convince end users, who would have probably gotten wind of this from all the press coverage, that this wouldn’t be an invasion of privacy.

Option #2: Massive user-segment mapping

In theory this solution isn’t too different from the above except that it requires browser-side communication of segments. Instead of having a global user database which allows companies to merge and map their data you simply signal each individual piece of information to each platform. If I were Revenue Science, Exelate or Tacoda I could sign up with Google, Yahoo & any other source of supply and set up my user category mapping in each system. Then, when I decide whether to add a user to a segment I fire off pixels (or whatever method to add a user to a segment) for each system.

Lets go back to remarketing to users. Some people might find this a little confusing, so I made a little diagram:


Lets walk through the steps (btw, this is hypothetical, I’m not actually tracking you like this!)

  1. You come to
  2. The html for the site contains three img tags that point to various platforms
  3. Your browser loads the Google pixel
  4. Google updates it’s user database and adds info to your cookie
  5. Your browser loads the Yahoo pixel
  6. Yahoo updates it’s user database and adds info to your cookie
  7. etc.

Why does this have to work this way? Cookies! Each company has a different UID and each UID is stored in a cookie. For Yahoo or Google to store data on this user they must know who the user is, for which they need access to the cookie, which means the user’s browser has to request content from their servers.

The above is actually a perfectly feasible model and practiced a fair amount today. The problem is that it’s difficult to scale — sure if all I want to do is tell each platform that you visit my site it’s pretty easy. But what if I have thousands of different user segments and then have weights and scores on each?

Option #3:Tagging users based on value

Instead of building a mapping of a thousand different categories across five different platforms there is another solution — flagging users based on value. Lets say I rank my behavioral segments into ten different buckets, from very low to very high value — for example, users interested in credit-cards are far more valuable than users interested in harry potter. After having assigned a value priority I can then create ten different segments in each supply platform that I want to work with. Each time I have access to a user’s browser I fire off one of the ten segments to each of the supply platforms to signal the value I place on this user, much like option #2 above. Note that I still place the exact category in my own user database.

Once I have flagged users based on value I can then place a media-buy with each platform at different price points based on the ten user segments. For example, I can have one $5.00 CPM campaign for all the high value users, credit-card and auto buyers, and a $0.50 campaign for the less interesting behaviors — low-income family, under 18, etc.. Each time I win a campaign I have the platform redirect the user to my adserver where I pick my own ad to serve based on the exact categories that the user in. The major draw-back of this approach is that it forces third party adservers. To some extent this is inevitable but this is non-ideal from a creative review, discrepancy and user experience perspectives.

Final Thoughts

Even though the industry won’t be standardizing behind one platform there are several methods of enabling cross-marketplace behaviors. Obviously there are some privacy concerns that will have to be ironed out, but those are independent of the platform being used. Also, if you own a site and are interested in remarketing to your users in ways I describe above you should check out’s LeadBack service which allows you to do just that!

I’d like to continue my series. If you haven’t already, be sure to read Part I and Part II first.

After my first two points I received multiple questions around the lines of “Who will make money off of this?”, and “Who benefits most?”, “How will ad-networks survive in this environment?”. Well, I thought we’d take a look at the various types of players in the market today and discuss how they will thrive/survive/die in the exchange environment. When discussing each of these I imagine a world in which there are two or three major ad-exchanges. Say, Googleclick, Righthoo & Micro7 … Any business that wants to play has to in involved with one or more of the exchanges as in this new world, 95% of all inventory gets sold on the exchange.

Read the rest of this entry »

Check it out —

Ok, so what’s the deal here? There’s clearly no website here, all I see are ads that are related to the New York Times. Well, what you’ve accidentally stumbled upon is a page managed by the new and highly lucrative Domainer industry. Domainers are companies that acquire massive numbers of domains, create pages filled with ads, and then rake in the cash. According to this CNN article the domain ‘’ rakes in almost $300 per day in advertising revenue. In a year that means by simply placing ads on this page, someone can make almost $110,000! Crazy right?

Now here’s the best part. When I went to, the second ad on the page was for the real New York Times. Check out the URLs we hit when we click:

New York Times Tracer

So first off, notice that the ad is being run through Google. Ok, so Google likes the Domainer space. Next we hit ‘’, Doubleclick’s adserver. This implies that the advertiser paying for this click I just performed is using their own adserver. Next, we go to ‘’ and I see an offer page for getting the real New York Times delivered to my house. The really interesting piece here is that chances are that the New York Times is paying for this click as part of their Home Delivery ad campaign. Chances are that the majority of visitors to ‘’ actually meant to go to the ‘’, and the actual New York Times is now paying to redirect users to where they intended to go.

Considering you can buy a domain for $8.95, that’s not a bad deal right? So, is this a scam or a valid business model? Lets look at it from two different perspectives, the advertisers paying for the ads and the people looking to buy domains.

The Advertiser: So you might say, “Why does the New York Times have to pay to send users that mistyped a domain to go to the real site?” You’re right, that kinda sucks, but then again, why didn’t the New York Times buy all potential spellings of their domain? Even so, now consider the case where the New York Times advertises on, another parked domain. In this case, the user is clearly looking for a news bit didn’t actually mistype the domain. If you owned a news site, wouldn’t you want to bring this user to your property? Exactly! Parked domains provide some of the most targeted and highly relevant users in the world, and advertisers are willing to pay big bucks to show you ads.

The Domain Buyer: Here is where I think more debate comes in. For the average Joe, it is now practically impossible to get any sort of normal .com domain name. All of the good ones have simply been taken. But on the other hand, these Domainers do bring a certain efficiency to the market. It’s hard to argue that the true value of ‘’ is $8.95 when you can make $300 a day in advertising profit! Creating domain marketplaces creates efficiency. The domain names that people will type in themselves because they are that good will cost in a lot of money whereas ‘’ was still freely available a month ago =). So, if prices only go up how is this good for the buyer? Well — because key domains become available. If were $8.95, some mom & pop shop in Kansas City will buy it for their little kid’s candy stand. The thing is, of the thousands of people who go to every day, very few of them will live in Kansas City. On the other hand, a large candy producing corporation wants to help it’s brand with a short domain name and can also better monetize the thousands of visitors that will naturally come to every day. So yes, there’s no more ‘free lunch’ when purchasing a domain, but if you get funded for a legitimate business, you no longer have to worry that you’ll have to settle for ‘’. Instead, if you think it’s valuable enough purchase a domain like ‘’ from it’s owner, or domains like ‘’ from Domain Marketplace companies like BuyDomains.

The one thing that I would like to see is a centralized marketplace for domains. Currently there are too many methods, just try doing a Google search for ‘Domain Marketplace’.

More posts to come on this topic. There are definitely some questionable topics here — search rankings via fake content, etc. etc. which need to be discussed. But at the core — the true value domainers bring to the internet industry is efficient pricing of domain names. And hey, if you could make a free $100k/yr just because you own, wouldn’t you?

Battle over the Cookie

March 5th, 2007

The latest craze in online advertising nowadays seems that everyone wants to build an ‘exchange’ or a ‘marketplace’. Sure, there are certain efficiencies that a marketplace provides, but in all honesty, it’s all about the cookie.

You see, there’s no need for more marketplaces, there are plenty of them — Adsense,, Tribal Fusion, and any other ‘ad-network’ that you can think of. The problem with each of them is that they are ‘closed’ marketplaces. Not just from a ‘business’ perspective, but from a technology perspective as well. Sure, anybody with credit and a legitimate offer can buy from any ad-network, but lets say you want to buy based on your own data. It’s simply not technically possible to do that today. Well, sorry, it is technically possible, just a royal pain in the a$$ if you do.

Lets say you’re a marketer and you have some information about a set of online visitors that is extremely valueable. E.g., lets say you know for a fact that they want to buy a car because they took part in one of your promotions. Now, first off, lets just assume that every network has behavioral capabilities (which they don’t). If you wanted to buy your users when they hit inventory that’s available on one of the many marketplaces (networks) out there, you would have to pass that data into your users’s cookie for each marketplace’s cookie. You see, when you come to this site, you get a cookie from I can write whatever I want into that cookie. I can’t see any of your others cookies nor can anybody see my cookie when you go to other sites. So, how do you get data into each Marketplace’s cookie? Sadly, this is not something you can “push out” to the user, you can only set or change cookie information when a user comes to you. E.g., if after-the-fact you decide you want to work with Marketplace D when you’re already working with A,B & C, then the only way in which you can tell put your data into Marketplace D’s cookie is when the user comes BACK to your site!

So how do you actually set information in another person’s domain if you can only access information in your own cookie space? Rather simple actually, anywhere on your page you include a link to a tiny 1×1 gif image to the 3rd parties server with some information attached to it. For example, you could put: would then interpret that as “marketer1″ just told me that this user “likescars”, let me store that in my cookie. Can you imagine putting 20 pixels on your promotion page just to gather data? Well, that sounds like a pain doesn’t it? Well it is — and hence, most people don’t do it. So, you say, I hear there are these new ‘exchanges’ out there (e.g. Right Media has one, DoubleClick is building one, and several others are trying). Well, what really defines an exchange? It’s a platform that lets buyers, sellers and brokers all work together using the same system. And what is the only way that will work in online advertising — a single cookie domain.

If all the mini-marketplaces out there worked on the same platform, then our marketer earlier wouldn’t have much of an issue reaching his audience. He would simply put his data into the exchange cookie and then buy from whichever sellers on the exchange have ads to show to his users. It’s not just behavioral targeting that benefits from an exchange, many other factors as well.

Lets say a second marketer wants to reach as many unique users as possible with his new ad campaign. So, he goes out to all the major ad networks and places a buy with the top 10 networks and portals and informs each of them that they are only to show each user his ads once and only once. Well, ignoring the fact that some people clear cookies and hence they lose the ‘frequency cookie’, how many different ad networks serve you ads on a given day? If you go to one site, they might be working with, another with Tribal Fusion, etc. etc.. Well, if this marketer is paying big bucks, each network will choose to show you this ad, but only once of course! The problem is, none of these mini marketplaces knows about the other, so nobody will know if you first saw the ad on the NY-Times from when you show up on Yahoo. The end result, you might actually end up seeing the same ad many times! Guess what — a unified cookie solves that problem!

So why the title “Battle over the Cookie”? Well, people are starting to catch on that if they own the cookie, they own the market. Think about it, if one exchange succeeds in capturing a massive percentage of the market, the barriers to entry will be practically insurmountable. What value is your exchange if you can’t provide services like global frequency caps, cross publisher behavioral targeting, etc. etc.

So who’s fighting? Well, my employer (Right Media) for one. Doubleclick is said to be launching a marketplace, AdECN is another. I’d expect others to start soon — as long as someone hasn’t won, there’s still a chance.

Doesn’t it sound scary that an entire industry is so dependent on a little piece of data that many users hate? How many spyware-detection programs flag cookies as ‘desktop software’? Well, in case it hasn’t gotten to you, it’s terrifying and the industry is scared. If Congress were to pass a law that said 3rd party cookies are illegal, what would we do? There are some interesting technology solutions out there that claim they can identify you by your computer ‘fingerprint’ that might be able to help, but in the near-term, the industry would be screwed.

So kids… hate online advertising? Clear your cookies. Like the websites you vist? Keep ‘em =).

So you own a small site, you have about 5000 page views a day, and you start thinking to yourself, “Hey, it’d be awesome if I could make some money off of my great website.” So you start shopping around for advertisers and quickly get the sense that all small sites work primarily with google. Now why would this be? There are four key ways in which advertisers can monetize your inventory:

  1. Branded Display — e.g., Niche-networks
  2. Contextual — e.g. Google Adsense, Yahoo, Quigo
  3. Behavioral — e.g. Tribal Fusion, Tacoda
  4. Performance Display — e.g. Fastclick/Valueclick, CPA Empire

Before I delve into these into more detail, everyone has to understand something: advertising is all about reaching relevant eyeballs. Whatever the method used, the end goal is the same — if I’m selling sexy shoes, I want people to buy sexy shoes. I can go about this a couple ways but at the end of the day, I’m going to compare the amount of money I spent on advertising and the amount of profit I made from the sexy shoes that I sold. The smaller and the easier it is to purchase my good, the more likely I am going to be focused on ROI & performance. If I’m Netflix selling $4.99 subscriptions, it’s easy to track who signed up from where as I will have tens of thousands of people signing up per month. On the other hand, it’s difficult for Ford to track ad campaign performance… it’s hard to prove you bought a new Ford F150 because you saw an ad for it on

Ok… so eyeballs it is… lets look at the four methods above now:

Brand: If your site is good enough to be on the brand wagon you’re in luck. Branded advertisers have high standards on the inventory they will run on. The demand that it is relevant to their campaigns (e.g. Ford will want to run only on Auto sites) and they are willing to pay a premium to get there. Campaigns are generally CPM with little or no concern for performance. The problem here is that unless a site’s users are extremely desireable it’s hard for a site with low volume to get onto a branded site list. Ford wants to sell cars — to do this they want to find car buyers and associate the ‘Ford’ brand with their target audience. They might run campaigns for a new sexy sportscar on gaming sites and ads for more fuel efficient sedans on the New York Times travel section (who knows?).

Contextual: Thanks to Google this has become the big mama of online advertising. Contextual engines scrape your pages content and choose advertisers that match the content. Got a site about shoes, expect to see shoe advertisements. Now, going back to the eyeball idea, if I’m selling shoes, I want people interested in shoes. If you have a site that reviews the latest and sexiest shoes, then I will want my ad on there. On the other hand, if you have a personal blog and are complaining about how the shoe salesman smelled bad at the mall today, I probably don’t want my ad on your page. Contextual rocks on relevant page content and sucks on most Web 2.0 sites. When was the last time you saw a relevant text ad on

Behavioral: Unlike Brand and Contextual, this is the first type of ad where we don’t care much at all about the page-content. We have an idea of what the users eyeballs are like and want to show him ads that he is going to enjoy the most. If you’re not famiiar with behavioral networks, read my post here. The biggest challenge here is for a site to get INTO the behavioral network. Even though for the purposes of serving one ad they don’t care what your site content is like, they will most surely want to use your site and your users to expand their reach & user information databases. Tribal Fusion, for example, requires 2k uniques per day, “professional site design” and a “highly active user-base”.

Performance Banners: Ok, so I saved my favorite (and perhaps worst for a small site) for last. Performance based advertising means that somebody is tracking clicks and/or conversions (actions) on their campaign and either only paying out on an actual “action” or adjusting bids on different sites according to the performance. Don’t be fooled by networks that pay you on a ‘CPM’, unless the ads they are showing are brand advertisements, somebody somewhere is tracking ROI & performance, and it’s impacting the CPM you’re getting paid. So why doesn’t this work for small sites you may ask? Well, long story short… small sites suck. Ok, it’s a bit of an exaggeration, small sites suck for performance would be a more accurate statement.

Lets go back to eyeballs. Your site has eyeballs, and I’m going to pay you everytime one of your eyeballs buys a pair of sexy shoes. Lets say the shoes cost $30, my production cost is $0.50 (yay for sweat shops!) so I’m willing to pay you $14.50 for every person from your site that buys one of my pairs of sexy shoes. Now there is one other advertiser in the network I’m working with he is selling sports gear (hats and stuff) and is willing to pay $10.00 to the publisher for every purchase.

Now, when a users comes to your site, which ad should the network show? Lets say your site is brand new to the network. Well, this network has no contextual technology to tell me what the site is about and he also doesn’t have a behavioral engine, so what can the network do? The only thing he can, test his campaigns on the site and learn what the performance is like.

Ok, for the next 5000 impressions (1k unique users?) the network randomly chooses between the sexy shoes and the sports paraphernalia. Now what… what do you think the chances are that somebody actually bought a pair of shoes or a hat? On the average site, I would imagine, right around 5-10%. Think about it, if the average CPM on your site is around ~$0.50, and this offer were to be competitive, you shouldn’t be expecting a single conversion until you’ve run at least 30,000 impressions. For better statistical accuracy on the performance of this campaign on one site, think more like 300,000 impressions. Starting to see why small sites suck? The amount of volume necessary to ‘optimize’ the average campaign is simply too much for the measly 5k impressions a day. Imagine if the network had 500 campaigns. How can learning be done on 500 campaigns with just 5000 impressions a day?

Of course there are various techniques to limit the amount of learning that needs to be done, but the fact remains that significant volume must be sent to an performance advertiser before one can get a good idea of the actual performance.

So, what should the small site guy do? Think about your eyeballs & page content. Is your content very relevant to the eyeballs? Then go contextual. Do you have particularly valuable eyeballs? Go for behavioral or brand. None of the above? Try performance based networks, but don’t expect good CPMs until you get high enough volume for some sort of learning to be done.

Next up? Not sure, probably some thoughts on learning techniques, merging behavioral and performance, and other ideas…

How do behavioral networks work?

February 28th, 2007

I really wanted to write a post about the difficulties in monetizing small websites but I realized to best have that discussion I would have to explain behavioral networks first! So, here goes.

A few months ago I was trying to explain to my girlfriend how advertising works. After a couple discussions she said a rather interesting thing — “How come the travel industry seems to be the biggest advertiser out there? All I see are travel ads, no matter where I go!”. Well, turns out my girlfriend is a travel fanatic and it’s behavioral networks that have picked up on the fact and no matter what the content of the site she’s visiting they manage to find a relevant travel related ad to show her. Try it, go to and click through to a couple articles. Then go to a random section of the site, say ‘health’, and chances are you’ll see a travel ad! (see screenshot below) Sounds like voodoo or big brother right? Wrong!

NYT Travel Behavioral
Second behavioral ad on the new york times.

And peeking into my cookies, indeed there is some travel related fun in there (hard to tell what all means, since most cookies are encrypted):

NYTimes Cookie Data

There are many different behavioral players out there in the world each with varied success. Perhaps the most successful behavioral company on the internet is this small company based out of California called Yahoo. There are also a couple ad-networks that dabble in this space, one of the most popular is Tribal Fusion. Each approaches the challenges in a different way, but the core of how it works is the same for all.

So how does it work? It’s actually an amazingly simple concept. Lets say we have four websites:


I am a user, and I visit each of these websites. Each of these websites also works with a brand new (not real of course) behavioral network called “BigBrother2.0″. On every page of these sites there is a little thing known as an ad tag that points to BigBrother’s adserver and requests an ad. Just for fun, lets assume that gizmodo also know your age & gender and passes that information to BigBrother. So you visit the first page, and an adrequest goes to BigBrother as follows:

When your browser requests the URL above, the following sequence of events happens:

  1. Browser Requests content from
  2. Server requests cookies (if any)
  3. Browser sends over cookies
  4. Server does some crunching, picks and ad
  5. Server returns new cookie data & url to get get ad
  6. Browser shows ad

Now the magic here is in the cookies. Even though you are visiting the site, the cookie is under the domain ‘’. In step 5 where the server returns new cookie data, they’ll put in there when you last visited and how many times you’ve been there today.

Now lets say I spend the morning checking out my tech-news, all the while receiving ads from Each page I view gives them a little bit more information about me, and even though I’m on three separate sites, because the cookie spans all three they get a holistic view of what I”m doing.

So now I go to This is a big site. Normally it’d be difficult to pick a good ad to show you. You’re one of the 18 billion myspace users and all you really do is upload photos and post silly comments about your friends on their blogs. BUT, when Myspace decides to let BigBrother2.0 show an ad something interesting happens. Lets walk through the steps of this new ad call in more detail:
  1. Browser Requests content from
  2. Server requests cookies (if any)
  3. Browser sends over cookies
    1. Cookie contains your age & gender (25 & male) (encrypted)
    2. Cookie shows you visited 12 times this morning
    3. Cookie shows you visited 3 times this morning
    4. Cookie shows you visited once this morning
  4. Server does some crunching, picks and ad
    1. Server plugs user data (25, male, 16 visits to tech sites) into a ‘profile engine’
    2. Profile engine spits back ‘categories’, lets say ‘male, technology’
    3. Server looks for ad campaigns targeted to ‘male & technology’
    4. Server picks highest paying ‘male & technology’ targeted campaign
  5. Server returns new cookie data & url to get get ad
  6. Browser shows ad

Instead of showing you a random ‘punch-the-monkey’ ad, BigBrother is able to show you a highly relevant and targeted advertisement that you are far more likely to click on. The end result? Better ads for you and more money for the websites that you visit.

You can see from the above that building a behavioral adserver really isn’t the most difficult thing in the world. All you need to do is track history & have some sort of ‘mapping engine’ that says that is a tech site. The real challenge for behavioral companies is the classic chicken & egg problem. Without a rich user-base, the ad-network won’t be able to sell deals. Without deals the network won’t be able to convince publishers to put their tags on pages. This makes it very difficult for new behavioral players to enter the marketplace, although emerging marketplaces such as the RMX will make it easier in the future.

So what’s my opinion on all this? As long as the data that is being stored is encrypted and no Personally Identifiable Information (PII) is stored I’m all for behavioral targeting. As I mentioned before, the more relevant the ad is to my interests, the more pleasant my browsing experience is going to be.