A call for an open-source ad tag

A call for an open-source ad tag

: I want to make a proposal for an open-source ad tag.

We need to enable advertising across citizens’ media, to support our new medium, and to serve marketers more efficiently than we possibly can today.

The problem today is that you have to pick the ad call you put on your site and that’s fine — except that let’s say an advertiser comes along wanting to buy ads on, say, the top 100 food blogs. Well, today, some of them will have BlogAds ad avails, others Burst, others Google, others Kanoodle, and so on. The advertiser can’t buy the best and largest network they want and they won’t bother going to each blog. So they lose the opportunity to reach this audience of smart influencers. And we miss the revenue.

It’s true that you can shift from one network to another: I recently spoke at the Burst company meeting and they explained that their publishers can set their tags so that, if Burst doesn’t have an ad to serve, the call can go to one of Burst’s competitors. But that still doesn’t really allow the advertiser to come in and create a customized network of the best sites to meet its goals.

I’ve been fretting about this for more than a year. One solution to this would be to create a citizens’ media ad network and infrastructure and work like hell to make it the ad network of choice. But that’s unlikely to work and it only presents another choice in a world where people are getting good revenue from BlogAds and Burst and Google.

So I started thinking about this idea of an open-source ad tag. I’m sure I’ll mess this up and others who know a helluva lot more about this will come in correct me — that’s what we do, eh? But try this on for size:

: You put the open-source ad tag on your site and it pings some central server (or is scraped by a spider) to announce itself, to make itself available to the network.

: Perhaps the ad can be associated not just with a site or a page but with a post and tags describing its content. That is targeting nirvana.

: You post data about your site with standard tags: topics, author, the kinds of ads you won’t take, etc.

: You also agree to put on your site a standard identifier so traffic, audience, and demographics of your audience can be collected (you can opt out, but you’ll sell less). You serve standard cookies to enable this.

: This standard identifier is also used with other data services so advertisers can put together ad hoc networks according to an amazing array of requirements — more than in any other medium: With the traffic data and with, say, Technorati, they could determine the 100 top food blogs as defined by (a) audience, (b) traffic, (c) influence, as defined by incoming links, (d) meme starting, as defined by your spot on the progress of stories. Or they could choose to buy blogs not by the demographics of the audience but instead by the demographics of the authors (sell a new album on blogs written by teenage girls, for example). This kind of targeting will be a key differentiator vs. other media.

: An ad agency or publisher or another organization can put together “quality networks” of selected sites, vetting them for quality, content, safety, whatever. So an agency can find the most influential sports blogs. Or a publisher can find compatible blogs to extend their reach or to bring in influencers for joint sales to advertisers. Or a church could vet blogs that are OK for Bible publishers or kids. This solves the problem of advertisers not wanting to be on “unsafe” content.

: You can accept or reject any ad. You can set your tag to take ads from the highest bidder. Or you can choose to approve every ad. You can set a time period to review ads so you can accept the best bidder. Or you can automatically accept ads from sources you trust. Or you can accept only ads that pay over a certain rate. I’m not sure how the bidding process works: Perhaps advertisers can start with a low price and increase it only if they don’t get takers.

: You can be paid on various models: sponsorship (cost per time period); CPM (cost per audience); performance (cost per click or per sale).

If this is done and if it becomes ubiquitous, then anyone who wants to sell or buy advertising can. The universe of sites for advertising — via networks or not — increases. The pool of money coming into advertising on citizens’ media explodes. The marketplace is efficient for buyer and seller and gives each maximum control of what matters to them.

I have been saying for a year that advertising on citizens’ media will not grow to its potential until we have three things:

: Metrics: How big, who, how they behave, etc.

: Serving: The ability to put together ad hoc targeted networks and place ads on them with no effort and high reliablity.

: ROI: The ability to measure return on investment and tune ad campaigns accordingly.

Again, one could build a network that does all that but it only competes with — rather than helps expand — the incumbents. And at the end of the day, as one VC friend said, you may end up with the DoubleClick of citizens’ media… except, in his words, “DoubleClicks’ business sucks.”

So I had my ‘ding!’ moment and decided that this should be an open-source solution. The goal is to get the ad call on any site that wants it and to collect the data advertisers want and to enable efficient buying and selling.

The net result of this — the good reason to do all this — is that it supports citizens’ media, bringing more revenue to it, allowing more people to be able to afford more quality media and I think that’s a good thing for those who want it.

I’m eager for feedback on how to make this happen, on whether it’s a good idea, on how to improve it. Pass the word….

  • Jonas Cord

    Had all the online ad networks agreed upon some standards, what you’re describing would be far easier to achieve. Because the inevitable consequence of creating what you’ve described is that it would be yet another “fork” – even more balkanization of the online advertising world – where some were in your aggregated camp and some were putzing around with propriatary tech.
    So, to tell you the truth, I think the technical aspects are probably the easiest part. It’s going to be the social pressure to get everybody at the table together, and then agree to use it.

  • Every form of voluntary tagging, irrespective of form and goal, has always foundered on the problem of spamming.
    They all are based on a fundamental assumption of good faith. They all require that everyone use the tags as they are intended.
    And every time in the past that such tags have been tried, it’s always been revealed that some unscrupulous individuals abuse them.
    That’s why meta-tags are no longer worth bothering with, for example.

  • I agee with the goal and hope that a spam-proof means to do it can be developed. Steven is correct, however. I’m already seeing spammers tag deli.cio.us entries so that my newsreader tuned into certain feeds related to tags I follow are unfortunately beginning to look like my spam-filled e-mail inbox.

  • Dave Morgan

    Great points jeff. I think that the open ad tagging concept is very consistent with the the idea of Sell Side Advertising that has been blooged and written about lately by you, Ross Mayfield, and John Batelle.
    I fully agree with the notion that the online advertising market is shifting to from a page-centric one to an audience-centric one. Like you, I believe that the future will enable more power to audience owners aand audience buyers. In my mind, what will drive the notion of Open Ad Tagging and Sell Side Advertising will happen when commercial ad sales and buying entities — the ad networks, SEMS, and interactive ad agencies — learn how to deploy, support, monetize, and optimize the tags and their data. It will be a very fluid and dynamic market and will certainly outpace anything comparable offline.

  • Rex
    When things like this occur, I always wonder why there can’t be a slashdot effect, creating a self regulating community where the value of a submission is determined by voting.
    Slashdot, I think, has stalled, because it is a single community, with particular bent, so, in a larger context, like deli.cio.us, it seems like there may be a need to form parties that can rank them.
    Is there also a means to route a message using a social network? Following the talkbacks, so that a message that responds well on one blog, travels to similar blogs? Are they doing this already?

  • Dang — you guys beat me to the punch.
    Self-created meta information — “tagging” –falls apart a soon as there is an economic reason to lie.
    I went off on the kumbaya around tagging at my talk at SES last week, and Danny Sullivan noted in a conversation I had with him later that there are hundreds of millions of “tagged” web sites right now, and none of that ever worked very well, after 1996. He was kind of chuckling over the parallels from the blogging community and the early search industry.
    The Open Directory (where I ran the community from 1998-2001) was all about creating a peer reviewed system to handle conflict in listing web sites (which is an analogous problem). And, still, major issues around SPAM.
    Also, there are no reliable, public, figures around traffic and audience — Alexa is about as good as it gets, and for most of the samll sites, the figures are in the noise…you could run some of the site reporting tools that Nick does — but those are kind of heavyweight for a small site.
    Anyway, I think its a laudable goal to make the influencer stuff work in an economic way, but you’re likely going to need an EBay or epinions-like reputation management system in place, and in all likelihood, it’s got to be at an Ebay or epinions like company to work.

  • Has anyone considered having an ad tag be an extension of RSS itself ((as in a markup tag). Implementing ad tags as an RSS 2.0 module would provide a standardized environment for implementing widely accepted “ad tags”.
    As an extension tag, you keep the contents of the tag under tighter control. Publishing software would produce the tag, RSS readers would decide how to handle/display the tag and feed advertisers would still produce the ads needed.
    I’ve dealt with a fair amount of spam on my site (comments, trackback) but not once has a spambot altered my feed. And with a decently secure server, it never could.
    As far as I know, feed advertisers just insert images into the content area. With a finer grain of control (you’re still “tagging” the ad), you can provide better tracking for all involved.
    I know this isn’t completely along the lines of what was proposed but it’s just a thought.

  • What about a referenced system? One where there are anchors of trust, like Jarvis, and he references credible sites? You can look at a FOAF file, or he can tag his links to other blogs, in the same way one would do for Amazon. If we trust Jeff Jarvis, then we make him an arbiter of trust. This way we track the tag back to its source. There is accountability. If some online [foo*] sets up a bogus blog, with bogus keys, that fine. If they are not rated a trusted site, then their bogus tags are ignored.
    Here the model is to exchange references. Forget about ads, this a product placement system. A means of trading in credibility.
    [*foo] The example I offered was blocked by Jeff’s blog software as questionable content.

  • Keep in mind too, that an organization that holds onto the rankings need not be terribly expensive to run. It would simply record, increment and decrement, a ranking, and serve it. The organic algorithm would be an exchange of trust links.
    Oh, that’s the name. Trust links.
    It might be something like del.icio.us that is external, where one could see a list of links in their feed, and rank them.

  • (disclaimer: I didn’t read the whole article, so this is more a reaction than a thoughtful response)
    I’ve installed an adserver that is able (like many others) to rotate different ads in a “zone” (think of a zone as your front-page-upper-right-ad-space). You could set it up in a way that rotates Google, BlogAds, Kanoodle and you-name-it evenly in that zone, so you would receive revenue from all the relevant campaigns advertisers place on them.
    If ad networks agree in a special way to identify non relevant ads (like Google’s public service ads), it is possible to adapt the ad server to recognize it and switch to the next ad network.
    This is a non-optimal solution, but you gave me a grrreat idea :)
    I can open an account for you at any time.

  • Dan

    What you’re describing is basially just google’s adsense, except instead of letting you tag your site for health insurance ads for their revenue, google spiders all the text on your blog and uses that as your tags to determine which ads to serve.
    I’d prefer a more open option as well though, and I hope eventually it happens.

  • Mark

    Like Andres I use an adserver on a commercial site of mine, the open source PHPAdsNew. You need to be able to install some software (PHP and the MySQL database), but if you use WordPress for your blog, you already have that. I’m able to rotate ads from various advertisers registered with LinkShare and Commission Junction, as well as AdSense. BlogAds are probably the most non-standard, and can’t really be rotated with others since the sizes are oddball.
    PHPAdsNew is quite powerful (you can place ads based on search keywords, or the geolocation of the visitor, for instance), but there is a definite learning curve involved, and online support support is spotty, although the documentation is excellent. But there are simpler, more rudimentary adservers available.
    As an occasional advertiser myself, I know that advertisers want to qualify the sites that they are advertising on. This would be something that an open initiative would have to deal with.

  • I agree that this is what Google does already. I think I noted this.
    I’m intrigued by the notion that an social network could perform the same sort of ranking and classification as Google, especially if there was a profit motive, revenue generated from references.
    How does one enforce collection?
    Also, Jeff recently noted how Craig’s List is destroying advertising revenue (was that you Jeff?). What will be the Craig’s List of GoogleAds?

  • Dan (not the Dan above)

    Yes, but there’s no reason that this tag should be ad-specific. It’s really a “topic” tag for the site. I actually write a decent number of tech specs in the Web space. I’ll try to write something up and circulate it (I’ve been thinking about something of the sort for about a year).
    It does take a little bit to formulate something that works well (a bad spec will be worse than no spec; witness the wasted effort of the RSS format wars).

  • No. Not ad specific. Didn’t I convey that?
    I’d imagine it would be something like deli.cio.us, but tied to a piece of real estate, like a blog. The links with a blog entry can be assigned a ranking, based on a scale of the networks choosing. I’m going to have to write a blog entry on this myself.

  • Kevin Kelly

    Great idea, Jeff. And buried in your first draft is the ideal name for this open source effort: AD HOC.
    May it live long and prosper.

  • The problem with the “community of trust” concept is that it’s unstable. All you need is one person who is good-hearted but also gullible, who improperly introduces one scoundrel into the community of trust, and it all falls apart. At that point it’s infected by scoundrels and can never recover.
    I know of cases like this in the past where there was an informal “community of trust” which fell apart in exactly this fashion.
    Michael Shipp, you miss the point. It isn’t that a scoundrel will corrupt your particular RSS feed, it’s that they’ll create their own which is bogus. The danger isn’t on the send end, it’s on the receive end. Someone out there is looking at the RSS feeds, and once a system like this is in place an increasingly large number of the RSS feeds will be dubious ones created by scoundrels.
    Movable Type’s automated trackback system is exactly the kind of thing we’re talking about, and after an initial period of exuberance about how cool it was, most sites have been forced to disable it because the vast majority of trackbacks they received were bogus ones indicating that they’d “been linked” by online cas inos or online drugstores or…
    Chris Tolles is correct: this kind of system fails catastrophically and permanently as soon as there is an economic incentive for anyone to lie.
    The biggest mistake that the guys at Bell Labs made in the early design of Unix was to assume that email senders would always tell the truth about their originating addresses. Thus they did not include an enforced authentication mechanism. That’s why upwards of 2/3rds of the email now is spam.
    All Communities of Trust fail because it isn’t possible to avoid the Prisoner’s Dilemma (and its derivations: the Freerider problem, and Tragedy of the Commons).
    Any system which relies on all participants to act in good faith is doomed. The only systems which are robust enough to survive in practical use are those designed under the assumption that a significant number of people are malicious and will seek to subvert or destroy them.
    Andy Grove had it right: “Only the paranoid survive.” You must assume that there are thousands of greedy and unscrupulous people who will try to abuse your system design to make it serve their purposes instead of yours. Design your system so that they cannot hijack it, because if you don’t design it that way, they will surely do so.