Vendors and Consortium Clash Over Who Will Drive the Future of AdsML

  Ideally, a technical standard serves its function so well that it appears to have been created out of whole cloth - a logical progression of technology, transparent to its beneficiaries. In fact, creating an international technical standard is only slightly less complicated than, say, writing a constitution and deciding whether it will guide a federation or a republic. In our second installment on AdsML, we explore the development of the standard, the key players involved and the impediments to its adoption.


  The AdsML Consortium, chartered in Darmstadt, Germany, described itself in a May press release, jointly issued with Adobe Inc., as "an international standards-writing organization for the advertising industry" composed of a growing number of participants in the advertising supply chain: software manufacturers and suppliers, trade associations and consultants. It is only one of a growing list of efforts on which the consortium and Ifra have collaborated. So far, the effort seems to have gained a much stronger foothold in Europe than in the U.S.


  Once the AdsML framework, a standard that employs XML metadata to describe advertising messages and data, was exposed to the rest of the world, unforeseen issues became major concerns, sending the standard's creators back to the drawing board to rethink implementation, though not the concept itself.


  All consortium members, suppliers and other interested organizations, including those who engage with advertising at various points in its life cycle, have strong ideas about what they want and need, and even stronger ideas about what they don't want and need. For the most part, U.S. publishers have been more reluctant to join in the fracas than their European counterparts. Nonetheless, some pioneers have ventured into the void to create the software that will transmit messages and data using the best practices and guidelines the consortium has labored so hard to create.


  Although the consortium began its work in 2002, a large portion of its efforts have remained all but invisible to those who have not attended the groundwork-laying meetings from which everything else in the standard will be developed. Building the Framework


  The consortium acknowledges that its chartered tasks - breaking down the life cycle of an advertisement into the smallest addressable increments and designing a component that will address both call and response for each modicum of data along the way - are "huge" and likely to "take many years to complete."


  According to the consortium, "The first order of business is to establish the foundation upon which everything else will rest." An initial step was to create the documents that define the AdsML consortium's charter and govern the behavior of those who develop it. To address antitrust issues and other unfair competition and restraint-of-trade considerations that might arise from a consortium of competitors, the group wrote a "code of conduct," which is revisited, at least in theory, at each meeting and even during conference calls, according to the AdsML Framework Overview. "It is important to realize that an action that may seem to make 'good business sense' can still injure competition and be prohibited under antitrust or unfair competition laws," the overview says.


  Thus, a fly on the wall would presumably not overhear discussions of rates, fares, surcharges, conditions, terms or prices of services, allocating or sharing of customers, or refusals to deal with a particular supplier or class of suppliers. Nor would it hear discussions of members' marketing, pricing or service plans. The document (perhaps reflecting concerns regarding recorded phone conversations made during Enron's heyday) warns members that "neither serious nor flippant remarks about such subjects will be permitted." Any member who feels that the conversation is moving in a direction that makes them uncomfortable is encouraged to "say so promptly."


  The first document that described the conceptual framework and contained an approved specification is a management-level overview, released in May 2004. (It can be downloaded as a PDF from AdsML.org if this report does not fully satisfy your need to know. But trust me, it will.) The primary authors responsible for the technical working group's document included RivCom Ltd.'s Tony Stewart, John Iobst of the Newspaper Association of America (who provided the basis for portions of the document), CCI Europe's Christian Ratenburg and Jay Cousins (also of RivCom), although others contributed and reviewed it. Once the governing elements were established, someone had to do the work. The consortium has three working groups: process, technical and marketing/communications, as well as an option for a Project Office.


  With the organizational foundation more or less in place, the consortium established its first set of business goals:


  * Digital e-commerce;


  * Automating the sending and receiving of content;


  * Processing and transmitting classified ads;


  * Providing guidelines and best practices, as well as a technical framework to allow developers to add AdsML message exchange capabilities to their systems.


  In its first round of development, the consortium intends to address print media, print inserts and Web sites. The industry players most likely to be affected in this round are advertising buyers and sellers and those who deliver ad materials to publishers, said Sue Sherrard Fine, manager of technology communication for the NAA and chair of the marcomm working group. The Matrix: Cooperation and Controversy
From July 2003 through 2004, the consortium posted a questionnaire on its Web site requesting information about interfaces that would fill out a conceptual matrix of interfaces at each intersection of an ad's life cycle (http://195.52.248.218/WebSite/adsml.nsf/HTML/Index.htm). When only eight vendors responded, Stewart found himself facing a rather intractable group of U.S. vendors belonging to the Newspaper Vendor Group.


  The Newspaper Vendor Group (NVG) is an extension of the evening bull sessions that reflected the Seybold Seminars in their original format some years ago, and is similar to the Newspaper Systems Group, itself a gathering of senior information technology professionals. The NVG has met once or twice a year for about 16 years, usually in the winter at a warm weather location to restore the spirits of those vendors huddled in snowy New England.


  Industry consultant Bill Solimeno (who wrote for The Seybold Report for years) convened the gathering of senior executives at vendor companies to talk about issues important to them, although senior newspaper people - from publishers to senior prepress and IT folks, as well as the NAA - are also included. This year, America East will also be present, said Solimeno. The vendors also make suggestions about who else should be invited. "It's not elitist, it's people who can speak about the issues and their company view," Solimeno said. Usually about 30 people attend the meetings, and there's only one person per company. The format is a round table, "and you can't do that if the group gets much bigger," he said. "I ask them what they want to talk about and write it on the board, and that's the agenda."


  "While Bill Solimeno hosts the Newspaper Vendor Group, neither he nor I claim leadership or to speak for the members of the group. It is a free association whose only agenda is to share concerns that affect all vendors to the newspaper business," said Richard Cichelli, president and co-owner of Software Computing Services (SCS) of Nazareth, Pa.


  The vendor group's most recent meetings followed this year's major newspaper conferences, the Dallas Nexpo in March and the Hershey, Pa., America East meeting shortly after.


  The subject of AdsML was raised at both meetings this year. "There was a backlash reaction and some very strong feelings to the tune of 'here goes another management-by-committee project' that won't get off the ground but will cost a lot of money," said Solimeno. He pointed out that a lot of the vendors who will potentially be affected by the AdsML initiative can't afford to join or to travel around the world to attend meetings. "Small vendors don't have that kind of money."


  Another impediment, he said, is the size of the prospectus; it takes a lot of time just to read it (I can personally count a number of evenings in which the various documents comprising the Framework completely covered the floor and every piece of furniture in my living room).


  "The primary issues were the perceived complexity and the cost of implementation," Iobst said of the America East meeting. "The results of the discussion were: it is not as complicated as it appears at first glance. Understanding the AdsML specs requires knowing something about the implementation philosophy and goals. Because AdsML is a general solution for advertising, it has more functionality and flexibility than a specific solution," he said.


  Open Source Message Transfer


  At the meeting in Hershey, Cichelli proposed open-source message-transfer software based on the AdsML message protocol, a suggestion that was well received. "It cost less money and they don't think it will slow down their systems when implemented at customer sites," said Cichelli.


  Old History, Current Concerns


  "Neither the NVG nor Richard [Cichelli] is against the NAA," said Solimeno. In fact, he said, "Everyone, especially myself, feels that the concept of AdsML is very good. Many are a bit frustrated since there is no easy way (due to the expense and the constraints of time) to pay enough attention to the passing train and still stay in business." It doesn't serve anyone well, especially newspapers, if their existing vendors simply disintegrate due to major changes to their systems, he said.


  Money is a particular concern. "The vendors would have to pay to join and to travel," Solimeno reiterated. "It's not the days of Atex, DEC and Kodak anymore." AdsML can potentially change the entire cast of players based on who can afford to jump in at this time, he said. Add to that the parallel cost of changes to their advertising systems, and it could cost the industry a lot. "A lot of very good players are trying to accommodate existing customers while making significant changes to move forward - which they will have to do," he said, adding that they don't necessarily want to do it "per some other group's spec," even if it is "elegant."


  Solimeno said the consortium, as well as the industry, would benefit if more people who are currently installing and maintaining the majority of ad systems in existence participated in the design and creation of the AdsML standard. Another concern was how slowly the consortium has moved since it was started in 2002. "Look how long it has been going on and what has it accomplished?" asked Solimeno.


  Cichelli, on the other hand, has been working in the U.S. newspaper industry since about 1973, when he spent 10 years as the research manager for Computer Applications with the American Newspaper Publishers Association Research Institute (now the NAA), where two of his projects were the display ad dummying system, Layout-80, and the spell checker, Micromark. He joined his wife, Martha, the founder of Software Consulting Services LLC, in 1983 as co-owner and president. The company now has 16 major applications running at more than 300 newspapers in 17 countries.


  The vendor group agreed that it would prefer to "bite off something we can implement by July - immediately. Everyone, including Gannett, thought it was a great idea," said Solimeno. "The standard Cichelli has written is open source and he will give it to anybody." It will take an ad from its source, parse it into AdsML and forward it to five or six newspaper systems, said Solimeno.


  Most of the work at this point will focus on the interchange between business and ad order, layout and ad tracking systems.


  At the NVG meetings, most suppliers were in favor of a level, open playing field. "I would have preferred the NAA to have spent its time writing code and releasing it as open source than creating such a huge document," said Cichelli. "I think it tries to be all things to all interested parties without being of much use to any. I felt there was strong support of my open source offering and much agreement about the same concerns that I had," said Cichelli, who hastens to clarify that he is not the NVG's leader. "I am too old to be a rebel. I am only a pragmatist."


  The pragmatic Cichelli said, "Most newspapers would like to expand their markets by cross-selling with friend and foe alike. I also suspect that when AdsML is presented, this is expected to be a benefit. After you have answered "why" for AdsML, the remaining questions are still: who, what, where, when, how and how much."
While Cichelli said he understands the virtues of having a standard, he was troubled by the ambiguity of the answers to those questions he'd heard from the consortium. "So I came up with some of my own," he said.


  "We have chosen to provide an open-source solution, Catapult, that addresses the transmission issue," said Cichelli. "We believe that it will provide a platform-independent, vendor-neutral secure transport level for sending messages (any digital data) among separate newspaper systems. We think this is a prerequisite to any message interchanging, so it should be of good utility," he said.


  Cichelli hopes that his first step will be "a simple, consistent means to transfer data among newspapers to be supplied [at least initially] by newspaper vendors using the Internet. The data could be insertion orders, ads, etc. Content will follow communication." He said he hopes that others will want to enhance his software, and he sees it as a community effort for use anywhere, distributed initially to NVG members who agree to test and provide feedback on cross system communication.  The software is free under GPL.


  In the next phase, Cichelli and his group will gauge the "interest in supporting cooperative cross-selling" and will choose a simple messaging format, perhaps based on known proposed or existing formats. That allows insertion orders and ads to be exchanged with the goal of making it as easy to do as SCS currently does among the sites running our advertising systems: fully automatic in all respects.


  "We will try to engineer a solution that is effective, focused, platform independent and designed for quick implementation, easy use and sharing," Cichelli said.


  A Modular Approach As Stewart sees it, Cichelli, on behalf of the vendor group, is "developing some tools to help people implement AdsML more easily." Iobst, who has his own long biography of newspaper-related achievements, said implementers are most likely to implement subsets of the standard to match the needs of their customer base. The consortium's approach to this has been to design a modular version of the specification, with numerous shared components. "This will help implementers manage their cost of implementation," said Iobst.


  Stewart said that small parts of the process of generating and receiving ads could be "widgets" that everyone could use and could be written by the consortium as open source. "But we are chartered to write standards, not code," he added. Unless vendors decide to adopt Cichelli's work or look into the other products that are coming to market, they will have to write their own code, since AdsML is not software, Stewart said. It is like XML, he said, except that "it goes further because it's dealing with particular kinds of documents that require their own rules and standards," said Stewart, adding that "both Java and C can be used to write XML standards."


  Neither AdsML nor Catapult alone is sufficient for exchanging mutually understandable messages that are about or are ads, said Cichelli. "AdsML is a means of describing content; Catapult is a means for transferring content. With Catapult, you could transfer any message, including AdsML-compliant ones, because the mechanism for doing so is common and shared. AdsML describes how some advertising messages could be shared. The two technologies could be complementary," he said.


  But he also sees some philosophical differences between his product, Catapult, and AdsML. "There are basic differences in the approaches taken. One (AdsML) describes an 'open specification.' The other (Catapult) is open source. One might say one is large and complex, based on the idea that every vendor will have to do a conforming implementation. The other says there should be just one shared project, incrementally developed by a community. Shared compliance and early useful deployment are assured because there is just one project.


  "One is a major development project for an entire industry's vendor community, each working on its own. The other is a targeted development effort, designed to establish proof of concept and offer immediate ROI. It will be controlled by the vendor and newspaper developers spending the R&D dollars writing deployable software," he said.


  "When we look at AdsML and our own in-production interfaces, we see some commonality and many differences," said Cichelli. "Many of these give rise to questions of 'how will AdsML support X?'"


  Cichelli's concerns about the unwieldy bulk of the AdsML specification haven't been assuaged by the consortium's talk of modularity. "Whenever we think of what AdsML doesn't handle, we also look at how big the specification has gotten already. Putting all of it into a system might cost over $500,000, and there would be no assurance that any other vendor would actually have built in compatibility you could depend on," he said.


  Cichelli is also concerned about interfacing to dummying, press management and classified pagination systems, credit card processing and e-tearsheets. "We have worked and are working with a number of vendors that prefer XML-style interfaces, including CCI Europe and Harland-Simon. While XML is used in these interfaces, none are AdsML," he said.


  "Many NVG members serve midmarket newspapers," Cichelli said. "They've all said that no customer or prospect has requested or has made it a requirement condition to support AdsML. We concluded that if AdsML were a good thing, newspapers surely don't know it. For success, AdsML must facilitate interchange of advertising data among disparate systems. No vendor was willing to commit to this without reservation. Doing so might not be to their competitive advantage."


  For their part, the vendors had a laundry list of problems with AdsML, too. Not enough vendors wanted to pay for developing it and there is no guarantee that separate implementations will work together. Vendors were concerned that there will be requests for fees to get e-mail of current documents, said Cichelli.


  "Our goal is to provide the transport mechanism that can be attached to any advertising order entry system," Cichelli wrote in an e-mail to Solimeno following the Hershey meeting. "It will allow different systems to provide order entry data to other systems. It might be used among applications at a given site or among sites with the same or different applications. Any advertising application that uses order entry data might use it. By making the code available as open source, we hope to assure compatibility and allow agile, rapid, shared, open, incremental development while avoiding an expensive, risky competition among ourselves to meet a third party's specification."


  Solimeno said that no one is averse to AdsML from a conceptual standpoint and the vendors support a standard for communicating among disparate systems. But it needs to be simple and easy, and it needs to be done by a group that's not in it for the profit, he said. (Catapult is now available on www.newspapersystems.com/products/catapult/index.php.)


  Adobe Cichelli's Catapult is not the only open-source product to take into account the new AdsML standard. In mid-March, the consortium announced plans to collaborate with Adobe Systems and the International Press Telecommunications Council (IPTC) to extend the AdsML Framework suite of standards. They plan to design "custom panels," exposed through Adobe's Creative Suite software, for advertising metadata stored in Adobe's XMP (extensible metadata platform), which will "link" with the AdsML workflow. In this case, of course, open-source does not mean free, as it does with Catapult.


  According to Mark Hilton, senior director of creative professional products at Adobe, "The collaboration with the AdsML Consortium will strengthen electronic advertising workflows throughout the industry by taking advantage of the extensibility of XMP to specify a set of custom panels to display and embed AdsML-compliant information." Adobe won't necessarily do the programming since the specifications, along with the XMP software developers toolkit (SDK), are freely available under open source license.

 
  This effort will build on another new public specification, the IPTC Core Schema for XMP, which was developed to make metadata more flexible and accessible for media and photographer groups by expanding the scope of information and data captured within each file across the workflow. The panels in Creative Suite will serve a similar purpose, according to the joint statement released by the consortium from Darmstadt.

 
  "XMP is a schema-neutral method for handling file metadata, which allows users to include particular metadata of relevance to them or the organization they work for in the XMP metadata," said Bob Schaffel, senior product manager for XMP and Version Cue at Adobe. "To make use of this capability, a user needs a way to visualize and edit such custom metadata. In Adobe applications, one way XMP is exposed to a user is a File Info dialog which is organized into panels that can be selected for viewing (and editing, if enabled) by a user."


  Schaffel explained how the work will proceed. "The specifications for custom panel creation include the definition of the user interface. At the time a custom panel is created, decisions can be made about such things as the visible form (such as popup list or text input field) or whether a field is even user editable (not editing camera EXIF metadata, for example), as well as labeling, provision of instructional text, etc.," he said.


  "A custom user interface is implicit in the creation of an XMP custom panel," Schaffel continued. "The underlying structure (programming) of a panel is similar to XML, and the creation of the panel and its user interface is a matter of creating this according to the specifications. The amount of time involved depends on the complexity of the panel (or in the case of IPTC, for example, how many different panels are required)."
This panel will perform two functions, he said. First, by its presence, the custom panel adds the custom metadata fields to the XMP. Second, it makes the metadata visible in the File Info panel user interface so that a user can view and/or edit the metadata. Once the XMP metadata has been input, it is embedded in the file, thus making the file a more "intelligent" carrier of rich metadata. Any software or digital asset management system that has XMP capabilities can read the metadata and expose it for visualization and action.


  Adobe declined to comment on when this capability will be incorporated into Creative Suite or on its expectations regarding revenue return.


  Schaffel said he sees broad support for the AdsML-XMP working group and expects it to be adopted by a broad range of users involved in advertising workflows, including advertising agencies, magazines, newspapers and so forth. There are currently more than 40 members of the working group (this is a different group than the AdsML consortium membership).


  This is not Adobe's first foray into developing standards with the newspaper industry. Schaffel pointed to the EXIF camera metadata standard, which is "probably being used more and more by newspapers as photojournalism with digital cameras continues to supplant traditional film-based methods. The ability to handle EXIF has been present in Adobe Photoshop for the previous couple of releases."


  What Adobe brings to the AdsML table, Schaffel said, is its ability "to flexibly handle custom metadata" and a "standard way to have the content files themselves carry that metadata." Without that, lots of money would have to be spent on "a lot of expensive custom programming, along with dedicated systems to help keep track of it all. It was also almost impossible to handle someone else's metadata."


  The strength of Adobe XMP, he said, is that "users and organizations now not only have a way to include their own metadata, but they can also make use of other groups' metadata. The files themselves are the carriers and are not dependent on the presence of a dedicated system to hold the metadata. Since our Adobe applications are all XMP-capable, such metadata can be brought right into the workflows themselves. The XMP is exposed not only in the creative applications, but are also available in Adobe Bridge, as well as Adobe Version Cue asset management system in Adobe Creative Suite."


  Schaffel also pointed out that because many digital asset management vendors have adopted XMP technology, the files can carry the metadata into the DAMs, too. "The excitement, therefore, is in having a standard way to handle custom metadata, having a standard way that it is embedded into the content files, and being able to bring the benefits of it directly into the workflow. XMP has gained a lot of traction in other areas besides newspapers because of this, particularly in the DAM arena," he said.


  Across the Atlantic: Cyber UK and CNet


  At least one company in the U.K. is developing AdsML compatibles. Cyber UK, which was involved with the standards project from the beginning, teamed up with Swedish company CNet Svenska AB, another early participant, which has been selling its Visual Net application server product in the Scandinavian media market for several years. CNet, said Brian Lacey, CEO of CyberUK and a former head of Atex, had also provided technical advice for Ifra's AdConnexion (which, he noted, "is being prudently subsumed into AdsML," adding that "it is this process of using what is around that seems to be positive and progressive").


  Ad-e-Transact is Cyber/CNet's implementation of AdsML, with a first installation already in place at an undisclosed U.K. ad agency. And FLT Media of Stockholm has engaged the duo to deploy Ad-e-Transact integration and message server. According to Lacey, FLT's internally developed booking system has been enabled to export and import bookings data, which Ad-e-Transact transforms into AdConnexion XML format using the AdsML Envelope 1.0, then transmits to an exchange Web portal, where the newspapers can download the bookings for import into their own booking system.
Rune Johnson, IT director for FLT, chose to base the pilot on Ifra AdConnexion rather than AdsML Bookings due to the widespread support for Ifra AdConnexion import in Scandinavian newspapers' booking systems. But he anticipates choosing AdsML Bookings in the future, thanks to its support for package bookings and better price-description capability. In both cases, he said, AdsML Envelope provides a reliable and secure messaging infrastructure.


  The Swedish pilot project has been running since April 2005. "Engaging in an e-commerce pilot like this allows companies to prepare their booking systems and business processes for an e-commerce future," said Lacey. "It is a good strategy to first optimize the e-commerce processes with a limited set of partners before turning to a full integration project with many external companies."


  Ulf Wingstedt, a consultant at CNet and a delegate to the consortium, said the FLT Media case is a good example of how newspapers can quickly get in place automatic processes for order taking.


  Lacey believes that the booking capabilities of AdsML, as well as future stages of development, will yield both benefits and derive significant opportunities for participating organizations, primarily in "headcount and therefore cost, as well as increase sales time for their sales people." So far, the focus has been on bookings and insertions, but as important as that is to agencies and publishers, it's only one stage in a long relationship. There will be many other organizations in the ad food chain who will be increasingly involved as the standard takes hold.


  Lacey understands that members of the Newspaper Vendor Group have been less than impressed with what has been achieved. "Clearly, e-commerce in the advertising food chain, in all aspects of advertising from print through multimedia to broadcast, would offer the industry the level of cost saving that earlier evolutions or revolutions have delivered," Lacey said. "It seems inevitable that not all groups or industry participants (a definition for industry might be useful) are not going to be reached or involved at the same point. One needs to understand this, and when you do understand, you either contribute and collaborate or you compete."


  Conclusion

  Although this article is at its conclusion, it's likely that there will be a good deal more   to say about AdsML, particularly in the U.S., as publishers speak more clearly about their needs and developers decide which path they will take. AdsML seems to be taking root much more quickly and deeply in Europe than it has in the U.S., perhaps giving U.S. suppliers a pass for the short term, but putting them in the position of playing catch up or perhaps competing with more mature products from European suppliers.


  Whether it was in conversation with the NVG or with other groups or individuals the AdsML Consortium has been in touch with, the consortium seems to be listening. Most of the key players in this story have known each other for decades. And I'd bet my turkey sandwich - and cranberry sauce - that this isn't the first time they've come to loggerheads trying to work out both the philosophical and the practical implications of a grand scheme. And when you look at the AdsML schema, even its critics speak of its elegance.


  Like it or not, American publishers won't be able to "just sit this one out" - AdsML is not optional. It'll be around longer than videotex was and it won't take as long as pagination did. Figuring out how to get a machine to do something that people have been doing for years is one of the things newspapers do best. As more companies and "properties" have become publicly held and as new media has become an integral part of the newspaper industry's strategic thinking, two key thrusts guide the development of technology: the demand for ever-escalating return on investment has been the compass that steers the direction of companies, and the search for the magic that will allow us to remain viable in the face of increasing sources of advertising competition from radio and television to Craig's List and Google News on the Internet.


  Initially, publishers whose pages carry national and international advertising will feel the greatest urgency about seeing AdsML work. Implementing AdsML will become a requirement for keeping advertisers happy, and a necessity if, as I anticipate, newspapers face a a more globalized advertising market. But they'll also implement AdsML to reduce costs, and companies known for truly aggressive cost-cutting will probably be among the first to look and linger. Corporate history regarding technological innovation tends to follow a pattern, and some organizations stand out as having the kind of technical depth to feel comfortable being the first to leap. Others demand that every wrinkle is ironed out before they make a commitment, meaning that they might wait a very long time before making a move.


  At some point it seems inevitable that the AdsML standard will either embrace or interface with other media that agencies deal with. A big question is, how will this standard interface work with existing or future standards for handling advertising in other non-print media. Newspapers have to go head-to-head with these competitors, while sustaining the culture and traditions that still make newspapers the most trusted news source.


  That issue aside, this standard begins to address big problems that newspapers identified long ago, including advertisers unhappy about production quality, the accuracy of ad content, and the need for a viable way to do e-commerce.

[时间:2005-09-20  作者: L. Carol Christophe  来源:bisenet]

黄品青微站