PubMatic has made its excellent report, distributed to attendees of its Ad Revenue 2009 Conference on October 8, available to all. Beginning on book page 36/pdf page 39 and running more than 20 pages is “The 2nd Channel Ecosystem,” a special section that I researched and wrote.
AdRevenueReport_PubMatic
Great lineup for next week’s conference all about helping publishers successfully navigate the challenges and leap the hurdles of our insanely complex business and tech ecosystem. Conference details here: Ad Revenue 2009
I’ve enjoyed a great opportunity to work with the conference’s tireless organizer and champion, Eric Klotz, PubMatic’s director of marketing, and with CEO Rajeev Goel, Chairman Amar Goel, and the whole team. I hope to see you there as well!
Part of my contribution to the 64-page report that attendees will receive is a nice reworking of my inventory naming chart, reproduced here.
WELCOME TO THE 2ND CHANNEL
Welcome to the 2nd Channel!
Attention, watchdogs. You are sniffing up the wrong butt!
Mark Rogers Photography on dogster.com
Consumer watchdogs, privacy advocates and others who prefer government regulation to industry self-regulation of data collection, cookie tracking and behavioral targeting put entirely too much faith in government.
Ironically, they’re even more wrong about how well any of this actually works in practice today on the web.
As an industry insider, I do all that I can within the bounds of safe computing to optimize my computer to allow cookie-setters, data gatherers and targeters to show me what they can do. Most often, the results are only fair or poor. On rare occasions, a truly relevant message appears that surprises and delights me and, more important, offers me something I’m actually interested in at that moment.
If it’s usually difficult for me to see the effects – good or bad – from the use of my data for commercial purposes, how is it possible that people with no inside knowledge are dead certain that malevolent forces are out to destroy our privacy via online targeting?
1. It’s not easy to be targeted. 
Every browser, anti-virus and pc optimizer program, among other widely-used applications, makes it increasingly difficult specifically to allow your machine to be targeted. This simple example comes from a routine scan (AVG, free version) of a machine which has been set and reset numerous times to ignore every one of these widely known, safe sources of ad deliveries and/or audience analytics. A typical consumer would take the virus warnings seriously and would promptly remove and prohibit such common cookies, which help make browsing more convenient and relevant.
2. On the internet, nobody knows you’re a dog.
The New Yorker cartoon by Peter Steiner generally is still true 12 years later. Sure it’s possible for hackers, law enforcement and others to piece together a story from your many web activities. Since you purposely leave your life’s details all over social networks, it hardly requires a visit from the internet czar to figure out what you’re up to. You can still choose not to reveal anything about yourself online by, well, refraining from using the web, email, IM, etc. Then only your supermarket, credit card issuers, banks, department stores and cataloguers, U.S. Postal Service and many, many others will know all your details. Just unplug your computer and you’re safe at last.
Do you worry that we’re only a step away from Orwell’s omniscient telescreen? I choose to obey the law and not worry. I’d rather enjoy the benefits of the web, including legitimate commercial uses by companies that are guided by respectful consumer-friendly principles and profit-driven innovation.
3. Behavior – online or off – can be misleading. After reading “Find Me Again: A Blizzard of Retargeting,” I visited several sections of the Skechers site, beginning with the women’s shoes section, likely to be the most valuable and heavily-tagged and trafficked part of the site for research purposes. Next step was to go to my Yahoo! Mail account and, sure enough, the parade of Skechers ads – all targeting women – began appearing. Six weeks later, I still see them all the time.
One visit to the women’s shoes department and they retarget me to death. I assume they can do this because they pay only for specific actions generated from ads on inexpensive and plentiful exchange and network inventory.
To the company’s credit, Skechers has been very open about their marketing efforts. In addition to the Behavioral Insider article, there are Skechers case studies from AudienceScience and MyBuys, among others, and open discussion at conferences. If the privacy fearmongers would read some of these, they might be further inflamed by the casual references to targeting consumers. Or they may get the idea that this practice is widespread, innocuous (even if some find it annoying) and, at its best, it results in a satisfying experience for the consumer and the advertiser. If they were to read further, they might even come to understand that this popular and simple form of retargeting is not particularly accurate or effective.
My first Skechers ad on Yahoo! arrived courtesy of Blue Lithium, but it might easily have been associated with other sources identified on the Skechers site on July 1, 2009, including Acerno, AudienceScience, Burst, Fastclick, Fetchback, Interclick, Trafficmp, Tribal Fusion.
4. Will the big dog hunt?
On the web, Google is like the world’s tallest dog, Gibson, measuring more than seven feet tall standing up. He appears to be friendly, but you can imagine how fearsome a presence this towering Great Dane makes, and how he must garner equal measures of respect and terror even among dog lovers.
Google introduced its flavor of interest-based targeting several months ago, so I promptly signed up for over a dozen different categories of business and consumer interests. Ho-hum … I have noticed exactly ONE ad that may have been a result of this program.
Interest-based targeting on Google: not interesting
If the big dog keeps up this performance, you watchdogs can move on to a new fire hydrant in someone else’s industry.
Several years ago I started a list of alternative terms for online ad inventory that I used in presentations when we first introduced the Right Media Exchange. It was part of a larger effort to persuade publishers to think about the importance of having a real strategy for non-premium inventory.
It was simple when the world of website ad inventory was divided between premium and non-premium. The language has evolved tortuously to fulfill the industry’s impulse to redefine anything simple so that it appears to be much more complex and, therefore, presumably, more valuable.
The online advertising ecosystem now includes ad exchanges and marketplaces, yield optimizers and creative optimizers, buy-side platforms and data exchanges, and numerous intermediaries who serve other intermediaries. Not content to feed from the existing language pool, each new organism in the ecosystem spawns its own new language, better to describe why it deserves greater prominence.
A notable coinage used by William Morrison and Robert Coolbrith of ThinkEquity in their excellent report, “The Opportunity in Non-Premium DisplayAdvertising,” inserts a new layer of website ad inventory called “secondary premium.” This strikes me as a rate card distinction, not some previously undiscovered form of inventory found living in the cracks between premium and non-premium. Here’s the ThinkEquity view:
By any name, there’s no denying that publishers are creating more and more inventory that they can’t monetize easily without a lot of help from networks, ad exchanges, and yield optimizers. One of the latter group, PubMatic. has introduced “2nd Channel,” which I believe nicely encompasses and describes anything that isn’t in the premium sales team’s sweet spot. (Disclosure: I’ve consulted for PubMatic.) Here’s the rest of the lineup, organized roughly according to the context in which the terms are typically used:
A non-premium display ad by any other name? 2nd Channel
This is pretty subjective, of course. I’d love to see a neat, descriptive term such as “2nd Channel” replace the mish-mash of differences without distinction represented by the list in the right-hand column. Keep it simple, and focus on the goals of improving user experience, advertiser performance, and publisher yield.
Nice post by Andrew Chen about the proliferation of ad networks and the surprising success of several that have sold for very high multiples. He summarizes it this way:
My overall lesson from all of this is that a lot of times, people view things as “winner take all” and sometimes it is that way – but in this case:
mature industry + real revenue + adjacent space heating up
= huge outcomes for everyone
I think this is true, but it’s probably cyclical and there’s bound to be a lot of consolidation ahead. True, thanks largely to MySpace, YouTube, Facebook, et al, total ad inventory available has skyrocketed. True, no single network or handful of networks can meet 100% of this growing demand. And true, multiple pricing models and targeting technologies help to make this a market for the multitudes.
But consider some of the pressures working against “huge outcomes for everyone.” Various estimates put the total number of operating online ad networks in the hundreds. Yahoo’s Right Media Exchange and nascent others can accommodate all of them – but only for as long as they contribute real value to the marketplace. Better targeting, better optimization, better something is required for establishing a niche, staying in business and succeeding.
Without the exchanges, there is almost literally no hope for dozens of these networks. Media buyers are humans, and they are in short supply as it is. There is simply no way for them to add another thirty phone calls and meetings to every day to give serious attention to every network that wants their business.
Ultimately, there can and will be “a number of” winners. That number will probably be closer to 20 than 200, however, due to basic Darwinian principles. Some reasons why this is so:
1. The largest contributors of inventory to ad networks already work with 10, 20, 30 or more networks. It’s a big, ugly, inefficient process in which networks pass ad impressions back and forth, up and down a daisy chain or waterfall, depending on your metaphoric predilections. As in a beauty contest, there can only be one true winner, and that’s the number one network in line. Ad impressions get stepped on many times, the way poppies get reduced to heroin and then to a street mix that’s cut time and again before it hits the “user.” (At last, an excuse to relate web users and drug users 
2. Conversely, the hundreds of ad networks are all selling the same media placements. They are all offering every known and quite a few imagined forms of targeting, but they all base their targeting decisions off the same limited data set – primarily context. This is true of contextual ad networks and of behavioral networks that rely on context to define behaviors or interests.
In the end, individual networks won’t win simply because inventory is growing and they’re bigger and badder in getting to the best of it first. Media placement, context-based targeting or some new spin on optimization don’t matter that much. With apologies to my publishing colleagues, that’s the commodity end of the business.
Advertisers need a better, more predictive and accurate data source to drive much more value through the media value chain.
Not very useful, that’s what a hammer is without a nail. In a business environment where everyone has a great idea, but most ideas don’t address and solve real problems, it’s great to find a hammer that slams the ol’ nail on the head.
Last year, Pierre Grenier was an associate at SF-based Catamount Ventures when he was dispatched to help portfolio company Piczo with its fast growing social network for teens. He quickly saw that the site’s staff spent an inordinate amount of time reviewing submitted images and posts for porn, abusive speech and other bad content that could ruin a family-friendly site. In short order, Keibi Technologies was conceived to solve a very real problem shared by all sites with user-generated content that requires ongoing moderation.
Paul Remer came over from Piczo as CEO to lead a team that includes Johan Wikman, VP of Engineering, and Jon Wilks, VP of Sales & Business Development, both of whom had worked with him previously on successful startups. With Grenier on board as Founder and VP of Product Development, last month they announced the launch of the Keibi Moderation Suite for automated moderation and classification of user-generated content.
When I first talked to Remer about Keibi back in June, what got me excited was not simply the nuts and bolts problem Keibi squarely addresses, but its potential to help sites that have both edited and user content to provide advertisers with a new layer of content that has been moderated and classified as, for example, “brand safe.”
Say a large social network has a highly saleable home page, section fronts and additional edited pages below the section fronts. Perhaps ten percent of their ad inventory is on these pages and they monetize this effectively. The other 90 percent, however, is user-generated and therefore largely off-limits to brand advertisers. The site typically works with multiple ad networks to fill in with low-priced performance-based inventory. If the site can chip away at the 90 percent, adding a layer of safe content that has been moderated and classified using Keibi, it’s a big win for the site and for brand advertisers that want to reach prospects and customers where they spend their online time, but are fearful of doing so within unmoderated content.
While the sweet spot is among online communities, if you’re an ad network or an advertiser with this problem, or an edited site that has additional user content, Keibi is also worth a look.
Q: When does “sold out” mean “plenty of inventory available”?
A: Almost always.
Herein lies the dirty little secret that will continue to drive online ad prices down over time for all but a fraction of the highest value branded inventory. In February 2007, Right Media presented findings of an Insight Express survey conducted among about 100 small, medium and large publishers, with representation by directors, VPs and higher from sales, operations, administration, marketing and editorial.
This slide shows that, on average, only 52 percent of total inventory on respondent sites is sold by the site’s own sales staff. Another 25 percent goes straight to networks. Respondents also said they used more than one network, with 43 percent using four or more ad networks to monetize their inventory.
Download the deck here: How Publishers Think About, Manage & Monetize Non-Premium Inventory.
… ask what media can do for tech. So might JFK have put it were he inspiring legions of media publishers, rather than U.S. citizens, to step up and be counted on to make a difference.
How well do revenue-enhancing media technologies really perform for publishers? Henry Blodget takes a good stab at how much Tacoda might contribute to AOL’s network revenues. Like most industry observers, he misses some basic assumptions that result in overestimating the impact, at least near term.
When any ad network claims to reach X percent of web visitors, they’re typically describing potential reach, not the actual number of visitors to whom the network served an ad in the previous month.
Example: Network Z serves ads to Sites A, B and C, which have monthly unique visitors of 2, 4 and 8 million, or 14 million combined. After removing duplicates (N visitors to A also visit B, N+X visitors to A also visit B AND C, etc.), say we have an unduplicated total of 7 million visitors to all three sites, or 50% of the duplicated total. The network may claim a potential unduplicated reach of 7 million.
But on a typical site, 20 percent of visitors generate 80 percent of traffic. So Site A, with 2 million uniques, has fewer than a half-million visitors per month who generate most pages and ad impressions. These same heavy users consume the lion’s share of ads delivered by Network Z to the unduplicated total.
Back to Blodget. First, he is correct about how Tacoda, Blue Lithium, Valueclick and others take non-premium inventory that yields well under a dollar and then use audience targeting to convert some of this into inventory yielding around $3.
In his AOL calculation, he assumes that “25% of display inventory and 50% network inventory can be ‘enhanced’ and that the ‘enhancement’ might range from 50% to 150%.”
This is a well-considered discounting of how much enhanced targeting can do for a vast amount of inventory, but it doesn’t go far enough. Since most of AOL’s volume is in email, IM, and other hard to target inventory, the amount that needs enhancement is probably closer to 90% than 50%.
Most ad impressions are generated by a small group of heavy users. If frequency caps are in effect, and if heavy users don’t do a lot of web surfing on other sites in the behavioral network, then Tacoda will have the same trouble monetizing most of this traffic that AOL has today. Therefore, my guess is that increasing AOL’s run rate by $200 million (the bottom of Blodget’s 10% to 40% enhancement estimate) will take years and many new tricks.