http://www.adexchanger.com/data-driven-thinking/real-costs-real-time-bidding-rtb/

For people interested in just playing around with the figures in this simple example themselves, here is a  spreadsheet you can input your own values: Excel 2007 Spreadsheet

We would certainly welcome comments and thoughts: there is a lot of real value that is and will be created here, but there needs to be further thought given to how it will all hang together.

Google has bought a Silicon Valley startup that customizes the online billboards known as display advertising.

Online billboards indeed! This deal was known to be happening fairly broadly, though the acquisition price I had heard rumored was a high one, which seemed too high. Will be great to understand some of the thinking behind the deal, but hard to go wrong with Google probably as an acquirer. Congrats to the Teracent team!

There are various pros and cons of Iframes (see here for a good overview some of which I summarize here), some pros include: Iframes don’t delay the serving of the page, typically loading in parallel with the other page content, there is no chance of name clashes with Javascript variable names if they are inside an iframe.

Some pretty big cons of iframes would include losing the referrer information of the page, which in turn means you can’t do contextual analysis of the page and more importantly you don’t really know where your ad has been running. Ads can’t communicate between each other if they are on the same page, so the coordination for wrap arounds or other things can be lost. You can’t run expandable ads (though some rich media creative companies like OggiFinoggi have come up with ways to “fake” expansion ads with iframes that is kind of interesting), which some users would say is a great benefit of these, but it complicates further the adserving ecosystem adding more heterogeneity to inventory which is now a mix of iframes and javascript tags and it is then unclear what can serve where.

As a demand-side platform helping advertisers buy optimized media, CPM Advisors (CPMa) runs billions of ad impressions per month across various advertising exchanges and we look at a LOT of impression logs. A lot of lower-cost inventory on some of these sources we have seen is hidden behind iframes, meaning most often the referring URL is that of the ad network serving the ads. As a whole, inventory that is served in iframes has a substantially lower price and underlying value than inventory that appears in javascript tags. The extent of the problem on any particular platform including Right Media is unclear, though I would guess that, excluding Yahoo! on the exchange, iframes might be 70 percent by volume if a fair bit less by number of publishers. There are obvious reasons for this, including that companies who would do nefarious things and generate large volumes of traffic often would hide their referring URLs behind iframes (and often, many nested layers of iframes running in parallel), but I do not believe the fairly young ad market yet correctly takes into account this trade-off.

Now there are exceptions. Some publishers like Amazon.com that work with ad networks grudgingly, purposefully put their ads in two layers of iframes to hide their origin AND to stop unscrupulous partners from “stealing” shopper-related data from their pages presumably – but as my other writings would attest I believe that the “sell the same stuff for different prices” world of online publishing will come to an end in the next several years and then this kind of good-quality-site-hiding-behind-frames behavior should be reduced – with the caveat that data use practices in a non-iframe world will also have to be more closely looked at, which I think will happen regardless. Are impressions a commodity? No, not yet but that wouldn’t be a bad thing if we had a real market to buy and sell impressions or “audience”. Let me put it this way – platinum and lead are both commodities but one sells for $1400 an ounce and the other for $1 a pound.

In tomorrow’s market-driven ad world, transparency will be more the norm and anonymous inventory will have its place, but on average will be discounted in price. Thus we may see what one might call the “iframe tax” – less measurability and accountability means more trust is needed which means more risk which means less money for networks and publishers that use them.

CRS is for small to mid-sized publishers looking for insider tips on how to maximize AdSense revenue, increase revenue opportunities via contextual optimization technologies, advertising network exchanges and the latest affiliate/lead generation alternatives.

Display Advertising for Search Marketers: Banners Aren’t Just for Branding Anymore
MODERATOR:
- Dax Hamman, VP, Display Media, iCrossing

PANELISTS:
- Roy de Souza, CEO, Zedo
- Rob Leathern, Founder and CEO, CPM Advisors
- Div Bhansali, Director, Self-Service Products, AOL Advertising

Hours of television per year for US Adults: 230,117,876 US Adults per the Census Bureau (18+) x 120.8 hours of television per month (Veronis Suhler Stevenson) = 27,798,239,421 adult-tv-hours per month, multiply by 12 = 333,578,873,050 television hours.

Television advertising dollars: $66,997,840,000 = added a 4% increase over a 2007 figure from TNS Media Intelligence.

Television access costs: $78,886,000,000 for all cable television revenue from SNL Kagan (referenced in the US Census Statistical Abstract 2009, Table 1105). You could probably make an argument about Premium cable and non-cable delivered premium services but then you get into online premium services etc. so I chose to focus on this figure instead.

The ad revenue per user television hour is thus the TV ad dollars / the Hours of TV per year = $0.20 per user hour

Access costs for television is Television access costs / hours of TV per year = $0.24 per user hour

Internet hours for 2008 for the US = 65,163,488,083 which figure was from comScore, adding up the monthly millions of minutes for each month in 2008.

Internet ad revenue for 2008 = $23,448,000,000 from IAB/PwC report, as is the split between search and non-search (45% is search for 2008 according to the report). From looking at these two numbers you get the online ad revenue dollars per user hour = $0.36

Based on comScore and Nielsen figures on the split between time spent with search (5%), we end up with 3,258,174,404 user hours for the year for search, and 45% of the revenue ($10,551,600,000) leading to $3.24 of search ad revenue per user hour spent searching.

Internet access fees were a little more difficult to get which is why my numbers were a bit more uncertain. I got from Pyramid Research a figure of $32 billion for broadband access but the dial-up and other access proved a bit more elusive so I guesstimated another $12 billion – so if you take $44 billion in access fees you end up with $0.68 per user hour for internet access. The figure is probably more if you take into account the costs of work access and so on, but we figured a $0.50 to $0.70 range is probably reasonable and conservative.

Anyone who has build forecasts and models knows that there is always uncertainty about figures even at this scale, but hopefully the analysis is interesting to people and draws attention to the job we have to do better as online advertising practitioners – driving user engagement with brands and products and making online advertising more effective AND useful.

Here is an interview with Rob Leathern, CPMa CEO on AdExchanger.com.

One point buried in there is useful to reiterate:

CPMa’s goal is to get every online marketer who is buying search to at least do some display advertising in the form of retargeting. We’ve created a quick self-service way for them to do it with a credit card at http://cpmatic.com that allows them to create their pixels, set up their campaign and everything without having to talk to a salesperson or fax an IO – the ROI is fantastic as you know, but volume can be low so reducing overhead is key.

We believe that setting up a retargeting campaign on CPMatic.com is simple – (just chose “retargeting” as the campaign type and create your pixel) but try it and let us know what we can improve. More info about retargeting here. ]]> http://www.cpmadvisors.com/2009/07/18/cpma-ceo-interviewed-by-adexchanger-com/feed/ 0 http://www.cpmadvisors.com/2009/05/31/cleaning-up-online-advertising-3-themes/ http://www.cpmadvisors.com/2009/05/31/cleaning-up-online-advertising-3-themes/#comments Sun, 31 May 2009 20:57:23 +0000 admin http://www.cpmadvisors.com/blog/?p=77

Online advertising feels like it has fundamentally changed very little over the last ten years. Perhaps indeed the “air got dirty and the sex got clean”, but it feels like things in the online ad world are smoggier than ever. Search advertising has scaled up nicely in no small part due to the dominance of our friends in Mountain View. Apart from that, though, for graphical advertising it seems as if the ad revenue engine supposed to monetize many of the new and useful applications and sites on the Web, has been stuck in second gear. Quo vadis, online display advertising?

There are a lot of things one could point to here, but I believe three change themes in the online display advertising business worth focusing on are Integration, Automation and Innovation.

Integration: The approach of publishers and networks as islands unto themselves will not continue to work. There need to be new, efficient onramps to publisher and network inventory. One reason an ad network is attractive to many advertisers, is because it aggregates inventory from a whole host of publishers. Likewise, an aggregator like the Rubicon Project who aggregates publishers based on optimizing their ad network relationships and in turn allows ad network buyers to buy inventory across groups of sites is providing a service of some value.

It’s just too difficult to buy from even large publishers – while many use IAB/AAAA standard terms, it’s often still a crapshoot on terms and conditions with many publishers. Then you get to the reporting, which can vary widely and needs to be reconciled, creative needs to be distributed etc. etc.

Buyers and sellers need to be more integrated, period. Whether that means one or more central marketplaces that clear inventory (by the way, I believe that publishers can still maintain premium pricing AND realize operational efficiencies in a market, but more about that some other time…) or ad-server plugins to allow them to list portions of their inventory… or even a standardized open CRM interface for sharing insertion orders between buyers and sellers, there is not going to be a single solution here, but the “connection index” between the companies with the eyeballs and the advertisers with the money needs to improve radically.

Related point: Unfortunately many “advertising as financial markets” analogies turn out to be wrong – impressions are NOT commodities that can be traded and sold across sites because of the complexities of placement and audience – so perhaps a still-flawed but much better way to think of where publishers fit in is to see publishers as stocks. (or really, placements at publishers – but you get the idea) Ad networks may hold a certain combination in a portfolio, and have a smart way to trade/buy/sell those stocks to help the investor (advertiser) make money. The idea here is that there is a central marketplace where buying and selling takes place, where values are established and where everyone can benefit. Not everyone has the size to be interesting to the market so networks themselves can also create stocks that are a combination of various smaller publishers and trade as a single entity. By taking steps to improve the quality of response that advertisers get (e.g. reducing the number of placements on a page, switching to better sizes), the value of the publisher inventory increases and so does their stock price.

Automation: This is not quite the same as the integration point, but hopefully goes hand-in-hand. By automation, I mean that we should automate our internal processes, and also move to the kinds of automated buying strategies and budget optimization that is possible once the two ends meet in the middle and there is a better chained set of data to accompany the inventory. It’s ‘If then else’ on steroids – look at the market conditions and take action, and take the unnecessary people out of the equation wherever possible. This is not to say take the creative people out of the loop, or never have eyes looking over algorithms – there is no such thing as complete automation, unfortunately. Just automate what you can in your process and always be looking for more to automate or eliminate if it is not really necessary. As you may guess I’ll have a lot to say on this subject here in the coming months since this is part of our core goal in what we are building. This also means less spreadsheet work, more built-in data analysis applications etc.

Finally, Innovation – trying new stuff and building real technologies around this industry, whether that is innovating within creatives (with a current average clickthrough rate of somewhere between 0.1 and 0.2 percent, or a 99.8 percent ignore rate!), or even just improving the current weak-but-somewhat-effective bucket-based behavioral targeting systems. Many of the sales-driven ad networks out there offer very little but aggregation and some economies of scale on adserving and the use of data, but true innovation escapes them. I like what some of the players in the social space are doing in trying to bring friend-of-friends data to adserving and build technology around that… as long as that’s not just retargeting by another name, that kind of new stuff is worth testing and supporting. CPMadvisors as a company is VERY MUCH open to partnering and testing new technologies, and we are always always testing new approaches and new tech partners’ offerings… innovation relies upon us as much learning from one another as properly capturing learnings from our own efforts.

I see encouraging signs here that bigger minds are being drawn to the online advertising space, and perhaps Wall Street’s recent troubles may help pull more people over. I don’t think the trickle will become a flood until there are real markets that we are working within: the current exchange environment when dug into just below the service proves fragile and winsome. The industry needs more smart minds building stuff, to push forward the underlying systems and infrastructure which are still very much in development or not even yet conceived of. As a friend of mine who has been designing proprietary trading systems on Wall Street for some time said in email today:

I’m curious to see if the progression of the ad exchange follows the typical innovation/commoditization process I’ve seen in the financial markets. It does seem like the wild west – no regulation, no formal exchanges, no standardized instruments/contracts. the natural evolution of real-time ad auctions is a fully fledged electronic market. once it is removed from one-to-one over the counter type trades and becomes standardized, the arrival of electronic market makers can explode the volume and profit opportunity.

David Rosenblatt (former head of DCLK) is leaving Google. The article has a good quote from him:

“We have an opportunity to radically change the economics of this industry,” he told me. But he added: “I don’t think we know 100% what the answer is yet.”

But it also ends in saying, “With others such as Yahoo’s Right Media and OpenX’s Market exchange, already moving quickly on that front, Google can’t afford to wait very long to come up with that answer.” Setting aside the need to discuss in depth what Google’s strategy around display is (another topic for later), it prompted something which has come up in discussion recently, namely whether display advertising in particular is a winner-take-all market either on the publisher side, the advertiser side or both. An interesting question to ponder – especially with the rise of ad exchanges – and I believe the answer to this is quite different from what we have so far seen with search, where Google is hugely dominant.

Will there be a hugely dominant central marketplace for display advertising and media? Even in market subsegments, should there be a single dominant provider of “ad network optimization” such as Rubicon Project or Pubmatic attempt to provide. I believe the answer to this is no. Search, in the case of Google, because they hold several aspects of the creative/inventory mix constant, creates a constrained environment that is easier to test against for advertisers (set format for text creatives, same display of the ads, etc.). When you start talking about graphical ads on third-party websites, it’s a whole new ballgame….

There is so much “untargeted” display inventory out there, much of it of dubious value, growing at a fast rate. The costs to trying out all this inventory, however, is going to be very high for advertisers or ad networks looking to benchmark it. The notion of seeing ad inventory as a tradeable commodity is also flawed; not least because it continues to grow in supply but price doesn’t fall in lockstep because of the testing cost problem, it’s very very different site by site and the complexity of trying to normalize for all these factors is huge. Given these, the diversity and creativity of website publishers and people innovating around inventory and ways of engaging users, it’s going to be hard for any single company to effectively boil down the differences.

The market for consumer information and targeting may be more winner-takes-most (more on that later), but for managing, optimizing and running media for publishers and advertisers, I would struggle to come up with a scenario where one player could effectively control most of that. Online non-serach advertising is really difficult, it can’t be solved by just technology (though technology is very under-penetrated in the display ad market, make no mistake), and yet the opportunity for advances and the ability to scale make this (with apologies) a very INefficient frontier where promise and profit will feed a whole host of companies for the forseeable future, even those whose names don’t end in -soft, -hoo, or -oogle.

We’ve been tracking goings-on in the social media advertising space for some time, and as such have also been tracking the growth of audience at large sites like Facebook. Facebook’s audience growth numbers are quite impressive, and as predicted, much of their US growth has come from the older adult audience. Growth since we measured things on 2/28/08 among users aged 13 to 23 has been just 33% (just!), which is minor compared to the growth rates in other areas:

While this larger audience creates increased targeting opportunities for marketers, and there have been some improvements in the Facebook advertising system (including a quality mechanism that turns off underperforming ads more quickly), the ad system still very much lags behind the innovation present in many other areas of the Facebook platform. The obvious consumer-side issue is that most of the ads are not that relevant. This is understandable – all social networks face this issue, in that the typical keywords on users’ profiles are diverse, center around interests, and mostly not that good for targeting. At LinkedIn, we were able to do a lot better than the average social network, with Google Adsense to generate valuable ads since the average user’s profile has higher-value keywords that can be used for targeting. So as a result, people target the kinds of things that ARE available, which is why when I saw this comment “I can’t wait to get married so I stop getting facebook ads for wedding planning” I laughed.

Facebook was under the impression initially that its ad platform would be used by brand marketers. An early set of guidelines banned companies from advertising there if they had a form that was gathering data as the landing page – the typical practice amongst most lead generation companies. The platform recently has seen a lot of use by direct marketers, with a variety of offers from skincare to quizzes (very popular) to car insurance; and yet Facebook offers middling reporting and no conversion tracking or other kinds of services to assist these types of marketers. Hopefully these things will change. I think that Facebook as an ad platform has a lot of promise and we are going to continue to use it for some of our clients, mostly in testing mode right now. One thing we have also gleaned from our experiments is that the text/graphical hybrid ads sold via self-service are certainly out-performing the banner ads on the Facebook system.

Herewith, our data on the growth rates by age group for the US audience, all taken from the Facebook ad system. 137 percent growth in a year and a quarter is not bad at all! Data at this link. ]]> http://www.cpmadvisors.com/2009/04/16/facebook-is-mainstream-advertising-not-quite-there-yet/feed/ 0