What I think made PPC so big, is the fact that publishers only have to place a few lines of javascript code in a page and everything gets taken care of.
With CPI, publishers would have to search for good advertisements themselfs, which is something I think most people don't want to spend time on.

• Posted by: Bart N.
• August 26, 2004 11:54 AM

Sounds like an affiliate program.

• Posted by: pb
• August 26, 2004 12:07 PM

I agree with PB, it sounds very much like what Linkshare does. The difference being that Linkshare is the source for the ad rather than the ad being available for copying from any site. Seems like you could add that feature onto an affiliate service like Linkshare.

• Posted by: ed
• August 26, 2004 12:27 PM

Yes, it is like affiliates, but more viral, and ideally, more...high end in its inventory as the ads come from the source.

• Posted by: John Battelle
• August 26, 2004 12:42 PM

Automation is key. Publishers (especially small ones) want to focus on creating content, not manually picking ads.

However, you can imagine a hybrid approach where the publishing tools such as Blogger suggesting a handful of ad links (along with price/performance info) that might be relevant to the content and the author makes the final decision. You can imagine this only taking an additional 30 seconds per post.

• Posted by: Kevin
• August 26, 2004 12:44 PM

All of the above.

Automation is valued for some, much less so for others. The hybrid concept sounds useful too. Maybe a variation on that: you give the add broker your blog URL, and it gives you suggestions or refines your search in some way.

John, I like the idea.

• Posted by: Gabe
• August 26, 2004 1:41 PM

I like where this is going... From an ideal UX perspective I see a publisher embedding code once, then "subscribing" to ads as they find them on other sites. "Subscribed" ads then get added to the rotation on the publisher's site.

The mechanism enabling the subscribe could be a small meta tag attached to the ad itself, much like the way trackback and rss subs work.

• Posted by: ed
• August 26, 2004 2:06 PM

Is there any way to build in elements like branding to this system? And what about advertisers whose ads tend to work well, but not right away (eg, the ad's for a hotel package, but the visitor compares some deals before returning to the advertiser to buy).

In general I like the populist concept of it, but if this were to come to fruition and catch on, some advertisers could lose out.

There are of course a lot of ifs involved, and that advertiser will have other options, but this is just to show that some of the good advertisers could lose out.

• Posted by: David Berkowitz
• August 26, 2004 2:33 PM

This reminds me of something I'd imagined we might see, more than a decade ago (and posted to com-priv, though I don't think there's an archive of it back to the early 90s, when this was)... CU-See-Me was the phenomenon du jour, and my speculation was that we'd see the rise of amateur p0rn streams; one of the revenue aspects was being able to select and include video ads, i.e., Jane (or John) Entrepreneur could point the camera at the bed, fire up the outbound video feed, and have the system mix in ads fetched from some service.

Dang... wish there was a source for decades-old com-priv postings, so I can see what I'd actually written.

I'm also reminded of "The Space Merchants," a wonderful dystopic future SF novel, where nation states have been supplanted by Madison Avenue...

[fyi... I had to edit this post, as MT seems to disallow the "P-word."]

• Posted by: Ross Stapleton-Gray
• August 26, 2004 3:11 PM

Ok, having read the post….here is how I managed to initially dissect this little nugget of a post.

1. Ad Relevancy is a moving target

Take two different people with “apparently” similar interests, let’s use Bill Gates & one of his Product Managers. Thinking of how advertisements could user demographics to differentiate adverts to suit both persons (e.g. a search for “cheap flights”, might actually mean quite different price range for both.)

So with additional Meta information combined with an advert, a closer match is achievable, now the only problem here is do you know your audience?

Ask yourself the question John, do you know my profile? Can the publisher targeting actually be useful for the advertiser? Now if AOL for example we’re to use demographics with Google – then it’s a different ball game.

2. Ad syndication / distribution methods

Now you’ve mentioned a central clearing house system recently in your article “Crediting The Blogosphere.” Now this kind of makes me think of a UDDI based registry acting as a “Yellow pages” for adverts, would probably work well in theory.

What I like about this is advertisers can “upload” different formats (e.g. TV, Audio, Text, Image) – so the publisher could actually select / display an advert depending upon various “session variables” such as a broadband user etc.

3. The Auction effect

In terms of this new model could we actually propose a new bidding system, not based on price, or price + CTR, on demographics + user profile (time to check the Patent Office.) This would help the destination site or broker achieve the value per advert that CPC models would.

4. Walk like an Egyptian

Now pyramid models are not new – will they work in this fashion. Well as long as when each advert is able to ping the clearing house each time its pasted into a page, to check whether its still live – then I guess its has the traditional fuel for growth.

So I guess my take on this is…..I think it can work, but which parties are going to drive it forward?

• Posted by: ID:entity
• August 26, 2004 3:13 PM

Easily the most interesting concept I have read on your blog. As a blogger I will accept ads I wish to display. In a sense it is similar to the original idea of website advertising. You put up a site and court local businesses to pay to put their ad on your site. Of course this is automated and contextually more in sync with your sites intent.
First to market with this scheme will become popular very fast.

• Posted by: Mike
• August 26, 2004 4:20 PM

Easily the most interesting concept I have read on your blog. As a blogger I will accept ads I wish to display. In a sense it is similar to the original idea of website advertising. You put up a site and court local businesses to pay to put their ad on your site. Of course this is automated and contextually more in sync with your sites intent.
First to market with this scheme will become popular very fast.

• Posted by: Mike
• August 26, 2004 4:21 PM

I swear I only pressed once. LOL
Sorry about that.

• Posted by: Mike
• August 26, 2004 4:24 PM

Targeting is the core of what makes Internet advertising work and the process of publisher inventory/placement management efficient. This model seems to assume that the publishers will be able to easily make good targeting decisions (this ad should be shown on this page, etc.) Unless the pricing model is CPA (cost per action/sale/lead/whatever) why would the advertiser want take that risk? Nothing could be worse then a publisher putting an ad on their site, having the ads generate clicks but no sales/conversions. Under that scenario the advertiser's budget would quickly get depleted on a CPC basis due to bad targeting on behalf of the publisher.

This is what makes contextual targeting so appealing - as it provides a level playing field of ad targeting quality across a very disperate set of publishers.

I could, however, see a model where the publisher bids for the right to place the ad on their site, in a sense reversing the CPC auction. That would at least keep the publishers honest, but again without standardized quality of targeting this would have to be a CPA pricing model.

• Posted by: Peter Adams
• August 26, 2004 4:47 PM

I think the problem with this model is that advertisers do not have the luxury of *hoping* that publishers will pick up their ads, no matter how strong their brand, creative, or offer is.

Advertisers have to spend ad dollars in specific timeframes and with a goal of awareness or ROAS within that timeframe, because they have products that need to be sold before they go bad/rotten/out of style, and in conjunction with the requirements of just in time SCM and lean inventories.

There's a reason affliate programs constitue 15% of online advertising and no more - most advertising efforts are tied to a required outcome in a required timeframe.

• Posted by: Chris Zaharias
• August 26, 2004 9:19 PM

To Chris' point: We may need a Buy Side model:

http://www.corante.com/many/archives/2004/08/26/does_sell_side_advertising_need_a_buy_side.php

And this may be less of a fit for depreciating inventory. The whole idea is Ads we appreciate.

However, if Ad Buyers are buying with the constraint of time (budgets are capped in the Sell Side model) it capps their thier options.

Affiliate programs and sponsorship are only one-hop commodities (not viral) compared to the ruthlessly automated commodity markets of keywords. Together they used to account for just under half of online advertising, when more people were influencing their results (however inefficiently). Tomorrow we will see, and what we move to influence.

• Posted by: Ross Mayfield
• August 26, 2004 10:00 PM

As I understand it, this doesn't make sense for advertisers.

Say Nike is paying Om $X per click-through. You paste Om's Nike ad onto your site and then you get $Y per click-through while Om gets another $Z per click-through of "juice" money.

If $Y < $X then you should've pasted the ad from whatever site Om got it from!

If $Y = $X then the advertiser would rather you pasted the ad from whatever site Om got it from. That would save them $Z per click-through! The advertiser doesn't want a "tall pyramid" of affiliates if every layer above the payout costs them.

If $Y > $X then... well that's crazy.

• Posted by: Andrew Bunner
• August 26, 2004 10:53 PM

As I understand it, this doesn't make sense for advertisers.

Say Nike is paying Om $X per click-through. You paste Om's Nike ad onto your site and then you get $Y per click-through while Om gets another $Z per click-through of "juice" money.

If $Y < $X then you should've pasted the ad from whatever site Om got it from!

If $Y = $X then the advertiser would rather you pasted the ad from whatever site Om got it from. That would save them $Z per click-through! The advertiser doesn't want a "tall pyramid" of affiliates if every layer above the payout costs them.

If $Y > $X then... well that's crazy.

• Posted by: Andrew Bunner
• August 26, 2004 10:53 PM

If there is no $X then there is no $Y. But when $X influences $Y, you get not only $Z, but $A for Advertisers.

When $X = $Y, its still two placements.

Mostly $Y

If $Y > $X, then someone jumped the power law by influencing the right people, which gives them a sizable $Z under CPI. This is a very good thing.

• Posted by: Ross Mayfield
• August 26, 2004 11:31 PM

Now someone put me on track here if I'm wrong - commenting on Andrew's & Ross's last posts specifically - does CPI require a form of a "PageRank" algorithm underpinning the CPI model? It just sounds very similar as to the original concept of link based content assessment.

If yes, then....well lets just wait and see, but it does make you think about asset management - well depending who you work for!!

• Posted by: ID:entity
• August 27, 2004 4:25 AM

This was an idea that was incorporated into AOL's Digital Cities interface during the late 1990s and was one of the most popular aspects of the personal and local services Digital Cities offered (it was not blogging, but the idea was that people could post information about their hobbies and form local groups, e.g., around jogging or dog training). Users could set their personal page advertisements, specifying what ads from among an inventory could be displayed on their pages—it was organized as categories you could opt out of, so that, for example, I could say "no alcohol ads on my home page." A positive system, where people chose from ads, is even more powerful, because it is an implicit product endorsement to say "show the Nike ad on my site."

I worked on this project with John Borthwick of AOL. You should chat him up about it, because I know they thought it worked. The problem was a lack of traffic within the Digital Cities service. There was a lot of positive feedback—users writing how much they liked that feature—asking for more ads to choose from, but without inventory the system couldn't blossom. With the critical mass of blogging, this is a service that certainly would be viable.

• Posted by: Mitch Ratcliffe
• August 27, 2004 4:57 PM

Very interesting Mitch. We'll chat. Main differences would be the metric that focuses less on impression and the transitory nature of the ads.

• Posted by: Ross Mayfield
• August 27, 2004 8:08 PM

What they learned from the project may be the glue that ties the transitory ad to the advertiser's perception of value: The fact that an ad was an implicit endorsement may have multiplied the price the advertisers were willing to pay.

The transitory idea is a cool twist, but for an advertiser culture built on CPM a viral ad model isn't going to make the sale. It will probably scare most advertisers, who would think they weren't in control of their inventory. If instead it is pitched as a viral endorsement system, where the number of sites that have picked up an ad—because the publisher/blogger likes the product, likes the creative or is purely interested in the cash—becomes a bragging point that conveys the market's loyalty/support for a product, then we're onto something.

AOL's problem was too little traffic and too little inventory, but the selling metrics were apparently excellent from the point of view of everyone involved.

On the remunerated networks of influence, this too is a powerful selling point for everyone involved. I may have just the tools for measuring how an advertiser would quanitify the value of their total placements ("current mojo"?) at any moment and over time.

• Posted by: Mitch Ratcliffe
• August 27, 2004 8:24 PM

John: Brilliant post. I added to it here by suggesting that the really ballsy thing is to let consumers not just place ads but also create them.

• Posted by: Jeff Jarvis
• August 28, 2004 11:17 AM

Killer stuff. I have been thinking for a while about evolutionary models and this is pretty close, esp with Jeff's contribution. Very cool!

• Posted by: umair
• August 29, 2004 6:23 AM

This is exactly what independent publishers like me have been asking, seeking and expecting until now. This is no revolution. It is long-awaited evolution.

To me there some aspects, that may be as important as the ones so beautifully described here that would need to be considered too.

This is a true paradigm shift in terms of how advertising is carried out and as you correctly say it turns traditional advertising on its head.
That can't happen partially. Either you turn it completely upside down or in my view it won't work.

So there are issues about conttrol, personalization, voice and transparency that would seem to require at least as much attention.

The Cluetrain Manifesto spelled this change out beautifully, and in my view, that vision still holds true.

It is in that direction that I would humbly suggest further questioning and refinement of this concept.

• Posted by: Robin Good
• August 29, 2004 2:01 PM

I wonder how one would do with income tax?

And would the payment be via systems such as PayPal?

Legally is it viable? I purchased and downloaded a few items via internet. But these aspects of buy and sell always puzzled me.

And of course to take things a step further is about services-provided and wanted? If I take a job in a foreign country I would need work-permit If I take an assignment from a company located in the US, on-line, how would I be paid? Would I still need a work-permit?

Cindy

• Posted by: Cindy
• August 30, 2004 3:00 AM

What a PITA. Thanks to Google, I added a few lines of code and I'm done. I know when to expect my check and for how much. This model would totally alienate publishers and the ads would never get displayed. It's a good theory in concept, in practice, it's a misserable failure.

• Posted by: mopacfan
• August 30, 2004 9:21 AM

Not sure how this alienates publishers. I think it empowers them (especially micro-publishers) and makes advertising part of the conversations they are engaging in.

On a separate note, I see this taking off more as an extension of AdSense and/or Overture. AdSense adds a component to their advertising allowing participants in the AdSense program to select ads to appear in their AdSense rotation. CTR over time then would govern which ads stay and which ones go.

• Posted by: Scott Knowles
• August 30, 2004 9:54 AM

Perhaps another way to think of how this might come together is through the confluence of a few things. First, having an entity or ad network like a DoubleClick that already aggregates ads make these accessible to publishers.

Second, incorporate a tool or ASP service that functions like AdSense (a tech similar to Applied Semantics' which is what underlies AdSense) to facilitate the publisher's ability to locate contextually relevant ads based on their content and pick these for their site. Where they might be comfortable w/all of the ads that came back fm such a search, then like AdSense, this process could be automated to have the ads fm the contextual query result automatically appear on their site w/o having to constantly do the manual contextual matching search each time.

Third, add the Linkshare affilitate hook.

Finally, there's one other piece of technology which has recently received some interesting patents, but this piece would satisfy the ability to embed a full application into the ad graphic (if we're leveraging banner ads) which could communicate w/a Linkshare affiliate system as the ad is copied on to other publishers' sites.

The pieces are there, it's just a matter of putting all together. The business models are there to support this as well, though perhaps a new intermediary for the advertisers to provide their creative to, would be required since their biz model would need to be different than DoubleClick's existing one, since DOubleClick may not be game for such a use of their inventory.

• Posted by: Pierre Wolff
• August 31, 2004 8:59 AM

All this stuff is extremely interesting for a product or service that has a predominantly online quotient.

I can't, for eg., see it being used to peddle soaps or shampoo or any one of the mundane things that conventional advertising sells. Can you see any self-respecting blogger deign to put up a soap ad on his blog?

I also have my concerns about the initial seeding of the ads. We need a far more robust system than what has currently been outlined.

That said, the idea of an ad campaign virally propogated is profound. It has legs but where they fit at the moment needs thought

• Posted by: Venu Gopal Nair
• August 31, 2004 9:45 AM

My primary concern is that, as an advertiser, I already know the majority of the sites that will reach my target audience. I would therefore be spending rather a lot of redundant time defining sites for which my ad would be suitable, and still more defining sites for which my ad would not be suitable. It seems that, in the end, my energy would be better spent just focusing on that small set of sites I already know to be a good match for my ads, rather than wiggling around hoping that my ads land somewhere worthwhile.

Of course, this is in addition to the obvious technical hurdles such as mechanisms for preventing the unscrupulous from posting my ads on their site and getting large numbers of unqualified people to click through in order to inflate the pay they would receive.

• Posted by: Sam Romero
• September 1, 2004 11:23 AM

This is sort of like affiliates except that many affiliates decide which publishers to let into their program. This model gives publishers less control.

• Posted by: James Trotta
• September 12, 2004 2:09 AM

Seems like this thread is winding down a bit. Perhaps in the main because of Sam's very valid "advertiser's view". Of course if the publisher knows the psycho-demo-graphic profile of his/her audience/readers he may well be able to provide advertisers with valueable clicks. This all reminds me of conversations long ago about consumers selling their eye-balls. The better your profile matches the target the more you get paid to view/read/hear the message. The idea also links in to that of targeting key influencers - those whose purchasing habits (and views) others "ape". To some extent many bloggers are of this type.

• Posted by: Marcus Kirby
• September 14, 2004 3:04 AM

Also reminds me a bit of Weed, which uses a similar model with music filesharing.

http://weedwarehouse.com/index.php

Marcus, I don't know how much I trust this "valid advertiser's" view. What Sam is saying is "I know all the sites that would be worth advertising on, don't make me think any more"

But it seems Sell-side ads are a dynamic, self-adjusting system. As the audience shifts from site to site with the vagueries of fashion, these ads will "track" that. New sites that suddenly get popular will see they have an audience, and will bring the ads to their site.

Sam thinks his "expertise" in tracking audience demographics will outperform such a system. But such self-styled experts pitting themselves against the wisdom of the crowd, seem to be getting themselves knocked over every day.

• Posted by: phil jones
• September 20, 2004 11:30 AM

worldwide distribution of electronic components with emphasis on obsolete and hard to find devices

• Posted by: esrmtech
• October 9, 2005 8:51 AM

Hello,

I am uday.I represent a start up company called Intellygence inc.I just went through your web site, and have liked it.I would like to have test link on your front page which leads to content based folder hosted from your side.Our folder contains content on a variety of
topics.If you find it agreeable we would like to do business with you.Please let me know waiting for your responce.

Regards,

uday.

• Posted by: uday
• November 11, 2005 4:45 AM

Many thanks for creating this blog site. this is really helpful for me and my friends!

• Posted by: Business for sale
• December 28, 2005 2:09 PM

Marcus, I don't know how much I trust this "valid advertiser's" view. What Sam is saying is "I know all the sites that would be worth advertising on, don't make me think any more"

But it seems Sell-side ads are a dynamic, self-adjusting system. As the audience shifts from site to site with the vagueries of fashion, these ads will "track" that. New sites that suddenly get popular will see they have an audience, and will bring the ads to their site.

Sam thinks his "expertise" in tracking audience demographics will outperform such a system. But such self-styled experts pitting themselves against the wisdom of the crowd, seem to be getting themselves knocked over every day.

• Posted by: Sernak Plywood
• July 20, 2006 12:44 AM

Here is Sell Side Advertising web service - www.adfeedr.net

• Posted by: Abdul Jaleel
• March 14, 2007 5:37 AM

And would the payment be via systems such as PayPal?

• Posted by: fatih
• April 25, 2007 9:41 AM

thanks

• Posted by: komedi
• April 25, 2007 1:02 PM

Great work ! I really enjoyed browsing through this site. I will recommend it to my friends. Greetings

• Posted by: Darmowe gry
• September 23, 2007 2:00 AM