A position in a search results list can look like an answer, even when somebody has paid to occupy it. That ambiguity became commercially important in 1996, when Open Text briefly offered “Preferred Listings” associated with particular keywords. The experiment belongs to the early history of paid search. Its significance lies in the decision to sell visibility inside a search experience, rather than in any claim that one company invented every component of modern search advertising (Fain and Pedersen, 2006).
The historical question is therefore more interesting than identifying a winner in a race to be first. What changes when a service designed to help people find information also sells access to the people asking the questions? Open Text exposed that issue at an early stage. Subsequent businesses developed more durable arrangements for pricing, distributing and purchasing such access. Those later arrangements should not be projected backwards onto a short experiment whose surviving documentation supports a narrower account.
Why a question could become an advertising opportunity
An advertisement in a newspaper reaches readers while they are doing something else. The publisher has assembled a context, and the advertiser purchases a presence within it. Search begins with an action by the audience. Someone types a word or phrase because they want to find something. The request may express curiosity, frustration, an attempt to reach a familiar website or an intention to buy. It provides a connection between a present activity and a possible offer.
This changed the nature of the inventory being sold. A search advertisement is not simply a small rectangle of space. Its value depends on the relationship between a query, a message and a position in the response. The same words may be useful in one search and distracting in another. The advertising opportunity is partly produced by what the user does, rather than being fixed entirely by a publication schedule or a page layout.
That is an interpretation of the business problem, not a claim that Open Text already possessed the complete machinery of later platforms. Paid placement, charging for a click, competitive bidding and an advertiser interface are distinct arrangements. They can be combined, but their eventual combination does not establish a single invention date. The early history becomes clearer when these components remain visible as separate choices.
The question also changes the advertiser’s task. Instead of buying only an audience category, a business can try to become visible in relation to an expressed problem. It still has to decide whether its offer is suitable. A query creates an opportunity to infer interest; it does not eliminate uncertainty about the person behind it. The appeal of search advertising rests on a closer connection to an immediate request, not on perfect knowledge of every potential customer.
What Open Text’s Preferred Listings establish
The historical survey by Daniel C. Fain and Jan O. Pedersen identifies Open Text’s 1996 paid keyword listings, while locating the later combination of bids, relevance review, matching, ranking and click billing at GoTo in 1998. Their chronology supports a distinction between an early paid-placement experiment and the more elaborate commercial systems that followed (Fain and Pedersen, 2006).
The careful description is consequently “an early experiment in paid search placement.” Calling it the invention of the modern keyword auction would require additional evidence about the actual rules. Similarly, a claim about precise prices or extensive advertiser adoption should rest on contemporary product or contractual records. Neither later familiarity with paid search nor repeated appearance in a timeline can fill a gap in that evidence.
The word “preferred” points to the central issue. A results list is not a neutral container in which every position communicates the same thing. An entry near the beginning may be understood as especially useful. If payment helps explain its prominence, that fact becomes part of the information needed to interpret the result. Selling a position therefore affects more than the seller’s revenue. It affects what the ordering of the list appears to mean.
This does not make a paid link inherently irrelevant. A supplier can purchase exposure for an offer that directly addresses a searcher’s problem. The question is whether the interface explains the commercial relationship. Is the user seeing a clearly identified advertisement, or might the business be borrowing the authority associated with an apparently independent selection? That distinction is central to understanding why a small alteration to a text list could matter so much.
The controversy concerned expectations about search
In a 2010 retrospective, search-industry observer Danny Sullivan recalled the brief appearance of Open Text’s paid listings in the summer of 1996 and their withdrawal following criticism. He described the advertisements as labelled. His account is useful testimony from someone who followed the industry closely. It is not an internal company record explaining every reason behind the decision (Sullivan, 2010).
The dispute can be understood as a conflict over the promise of the service. People had asked for help finding something. A payment influencing the response introduced an additional relationship into that exchange. Even if the advertised site was useful, users had reason to want to know why it appeared where it did. The issue was not exhausted by asking whether advertising in general belonged on the commercial web.
It would be too easy to describe the critics as people who simply failed to understand the future. Their objections concerned an enduring problem: an information service can earn money from a decision that shapes the information its audience sees. Later commercial success does not retroactively settle every question about the presentation of that decision. A profitable practice still depends on conventions that make its role intelligible.
The opposite account, in which payment destroyed an otherwise perfectly neutral search environment, is also inadequate. Every search system selects, excludes and orders material. The relevant distinction is not between an entirely unconstructed answer and a constructed one. It is between different influences on that construction. A commercial transaction adds an influence whose existence may matter to the person trying to assess the result.
Understanding the controversy in these terms preserves the historical agency of the audience. People were not merely waiting for a new technology to mature. They were helping establish what kind of commercial conduct would be acceptable inside a service they used. The development of paid search involved these expectations as well as code, prices and sales arrangements.
Open Text’s corporate history followed another path
Open Text’s 1996 annual report provides an important qualification. It emphasized intranet applications and the combination of search technology with document management, workflow and collaboration through Livelink. It also reported a January flotation that raised approximately US$61 million net. The company was positioning its capabilities for enterprise customers, not simply describing a business dependent on public search advertising (Open Text Corporation, 1996).
This document does not reveal the complete internal explanation for ending Preferred Listings. It does show why the advertising controversy should not be made to explain the entire subsequent history of the company. The fate of a public search product and the fate of the organization offering it are different subjects. A company can withdraw from one market while pursuing another use for related capabilities.
This distinction is especially valuable in internet history. A vanished homepage can create the impression that the underlying business vanished with it. Yet search technology can support different customer relationships. A public service may depend on attracting users and monetizing introductions. An enterprise application can be sold directly to an organization seeking to retrieve and manage its own information. Related technical skills do not make the business models interchangeable.
The evidence therefore supports a more careful reading of 1996. Open Text tested paid visibility in public web search while its corporate strategy also emphasized enterprise software. The short advertising experiment deserves attention, but treating its reception as the sole cause of the company’s direction would go beyond the sources. Chronological proximity is not sufficient evidence of a complete causal explanation.
It also changes the meaning of failure. A discontinued product may fail to become an enduring product without proving that its technical foundations or the organization behind it failed. Historical analysis should specify the unit being assessed. In this case, the limited life of Preferred Listings is a finding about an offering; it is not a verdict on every activity undertaken by Open Text.
A contemporary view of the change in 1998
Sullivan’s March 1998 Search Engine Report described the replacement of the general Open Text Index by Livelink Pinstripe on 16 March. The business-oriented service was also intended to attract prospective software customers. This contemporary account places the change in the context of a different audience and commercial purpose (Sullivan, 1998).
The distinction broadens the history of monetization. A search service does not have to charge for every introduction to serve an economic purpose. It can demonstrate the operator’s capabilities and bring an appropriate audience into contact with a different product. General advertising reach and a specialized showcase are distinct strategic possibilities. The contemporary report documents the latter possibility without proving its eventual financial success.
Read alongside the annual report, it provides a useful counterweight to compressed retrospective timelines. It allows the paid-listing experiment to be situated within a wider change in corporate positioning. The controversy remains important, but it cannot explain every decision about audience, index coverage and the relationship between public search and enterprise software.
SEM, SEO and paid search acquired contested meanings
Terminology can conceal the sequence of events. Sullivan’s 2010 discussion described a dispute between using “search engine marketing” as an umbrella for organic and paid activity and using SEM more narrowly for paid search. His testimony documents disagreement within the profession, rather than a timeless definition accepted by everyone (Sullivan, 2010).
For this account, search engine marketing means efforts to reach relevant visitors through search services. Search engine optimization concerns visibility in results that are not directly sold as advertising placements. Paid search concerns purchased advertising within the search experience. These are working distinctions for historical clarity. They do not imply that everyone in 1996 already used the same vocabulary or organized their work under today’s categories.
One practical consequence follows immediately. Paying for an advertisement does not, by itself, demonstrate that an advertiser has purchased a better position in the unpaid results. Another is equally important: unpaid placement does not mean that visibility requires no resources. Preparing material, maintaining a usable website and making information accessible can involve substantial work. The distinction concerns the immediate transaction with the search provider, not the existence of economic effort.
The two activities can also have different temporal characteristics. A purchased placement may depend on an active booking and a budget. The visibility of a document in an unpaid index depends on other selection processes. Combining their outcomes without distinction can make it difficult to understand what was bought, what was developed and what persists after an expenditure stops.
Historical vocabulary should therefore follow the evidence rather than dictate it. Referring retrospectively to Open Text as part of SEM history is useful. Suggesting that it launched an already settled profession called SEM in 1996 would be a different claim. A history of a practice and a history of the name eventually attached to that practice do not necessarily begin at the same moment.
Placement, inclusion and billing are different arrangements
A June 2002 Federal Trade Commission staff letter distinguished paid placement, which affects ranking, from paid inclusion, which affects entry into the searchable pool or the timing of that entry. It called for clear disclosure while explicitly withholding a final determination about the legality of the challenged practices. The letter is historical evidence of regulatory concerns, not a statement of present-day legal requirements (Federal Trade Commission, 2002).
A hypothetical archive illustrates the difference. Paying to have a document entered in its catalogue is not the same transaction as paying to have an existing entry moved to the beginning of a list. Both can change visibility. Only the second directly buys preferential ordering. This example explains the categories; it is not an additional documented description of Open Text’s contracts.
Billing introduces another dimension. A position might be sold for a period, for impressions or for clicks. Knowing the billing unit does not establish how the position was allocated. Likewise, charging per click does not automatically imply an auction. A provider could set a fixed click price, while an auction could allocate some other kind of advertising entitlement. The institutional choices need to be reconstructed separately.
This is why asking who “invented SEM” can produce unsatisfactory answers. The category contains several problems that different organizations addressed at different times. A more productive inquiry asks which business combined which components, under what rules and with what response. Open Text belongs to the early commercialization of search visibility. Later companies developed other combinations of pricing, distribution, review and advertiser access.
The distinctions also clarify what a source can prove. An announcement that businesses could purchase listings establishes an offer. It does not necessarily establish the scale of actual use. A billing description establishes what triggered a charge, not the advertiser’s profit. A ranking description establishes a rule, not the quality of every result. Preserving these boundaries makes the historical argument more useful and more reliable.
Search intent was never identical to purchase intent
Andrei Broder’s 2002 taxonomy distinguished informational, navigational and transactional searches. He also cautioned that intent could not be inferred with certainty from a query. Transactional searching included activities beyond purchasing. This later framework helps interpret the economic possibilities of search, but it is not evidence that Open Text used those categories internally in 1996 (Broder, 2002).
Consider three illustrative searches concerning a camera. One person wants its history, another wants the manufacturer’s familiar website and another needs a repair service. The same product name could appear in all three requests, but the commercial opportunities differ. Matching a word is not equivalent to understanding the task. Nor does recognizing a task reveal the person’s budget, location or willingness to act immediately.
This makes the apparent simplicity of keyword advertising deceptive. A term can be common and commercially disappointing. A narrower phrase can occur less often yet connect an offer to a more specific problem. Neither proposition supplies a universal performance rule. Both show why relevance involves more than counting the number of times a word appears in the audience’s activity.
The advertisement itself also helps form expectations. A message that attracts attention by promising something the destination does not provide may generate observable visits without useful business. Conversely, a clear description may discourage unsuitable visitors before they click. From an advertiser’s perspective, fewer interactions can sometimes be preferable if the interactions better match the service available. That is an analytical possibility, not a claimed result from a recovered 1996 campaign.
The historical attraction of search advertising is therefore best described as proximity to an expressed request. It offers a signal unavailable in some other advertising situations. The signal remains incomplete. Treating it as a complete account of the searcher would confuse an observable action with a fully understood individual.
GoTo turned a related idea into a more durable system
Overture’s report for the year ending in 2002 records incorporation in September 1997, the official service launch on 1 June 1998 and the change from GoTo to Overture in October 2001. It describes bid-ranked listings with advertisers paying for clicks. The official service-launch date should be distinguished from earlier demonstrations or public presentations (Overture Services, 2003).
The economic difference was not simply that a company requested payment. A repeatable market requires rules and procedures. Advertisers need to know what they can purchase, how changes are made and which event produces a charge. The provider needs to associate interactions with accounts and decide whether advertisements fit the relevant queries. These supporting activities are part of the product, even when the visible output is only a short text entry.
A bid can indicate what a business is willing to pay for a visit opportunity. It does not directly reveal which business offers the best answer to the searcher’s problem. Willingness to pay may reflect margins, expectations or commercial strategy. The platform must therefore decide how price relates to suitability. That decision is not solved merely by moving the transaction onto a computer.
Open Text and GoTo belong in the same history without being treated as identical systems. Both concern commercially purchased visibility associated with queries. The difference lies in how that relationship was organized and sustained. A short experiment and an operating market can share an underlying idea while differing substantially in the arrangements that make it usable.
This perspective also avoids a misleading account of innovation as a relay in which each participant merely enlarges a finished invention. Combining billing, review, advertiser control and distribution can change what the offering is. The later system does not simply contain more of the earlier product. It can establish a different relationship among the parties using it.
Advertising could travel through a network of search services
The October 2001 announcement marking the adoption of the Overture name described distribution relationships including AOL, MSN and AltaVista, together with editorial screening for relevance. As a company release, it documents the network being promoted and the company’s account of its procedures; it is not an independent evaluation of their effectiveness (GoTo.com, 2001).
This matters because the website visible to the user and the business supplying paid listings did not have to be the same organization. A search service could incorporate results from another provider. The market was therefore built partly through relationships among companies, not only through a direct relationship between an advertiser and a famous homepage. Counting visible search brands alone can conceal the infrastructure behind them.
For advertisers, a distribution network could simplify access to different audiences. Yet aggregation also complicates interpretation. A click occurs somewhere, in a particular interface and context. Combining many such events into one total does not make all the underlying opportunities equivalent. Distribution can expand reach without guaranteeing that every environment produces the same value.
The international implications followed the same logic. A technically accessible advertisement is not automatically an economically useful introduction. Language, delivery, payment and support affect whether a business can serve the person who sees it. The internet’s ability to carry a message across a border and a company’s ability to complete a transaction across that border are separate capabilities.
The history of search advertising is consequently also a history of intermediation. One business attracts the question, another may supply the commercial listings and a third hopes to convert the introduction into a useful relationship. The transaction looks immediate on the screen, but it can rest on several contracts and operational decisions behind it.
Google in 2000 and 2002: separate steps in the chronology
Google announced the AdWords self-service programme on 23 October 2000. Its release described keyword-related text advertisements, web-based campaign management and sponsored links separated from search results. Approximately 350 businesses and agencies had used the beta programme, according to the announcement. The advertiser’s ability to operate the buying interface was itself part of the offering (Google, 2000).
On 20 February 2002, Google announced AdWords Select with cost-per-click pricing. The release described ranking influenced by both the bid and click-through rate, while premium advertisements retained impression-based pricing. Those details distinguish the new offering from an account in which the same click-auction model supposedly existed unchanged from the 2000 launch (Google, 2002).
The comparison reveals several decisions rather than a single inevitable trajectory. A system can emphasize the price offered for a position or incorporate additional indicators of how advertisements perform. It can sell through negotiated arrangements or make more of the transaction accessible through an interface. These choices affect the advertiser’s work as well as the provider’s revenues.
Self-service is particularly important as an organizational change. The interface gives the advertiser a set of permitted actions and a way of seeing results. It embeds rules about how participation works. Buying advertising thus becomes partly a task performed through software. The product includes not only access to an audience but also a procedure for acquiring and managing that access.
These later dates provide context for Open Text rather than a complete account of Google’s advertising history. The point is the sequence: a 1996 paid-listing experiment, GoTo’s more extensively connected system in 1998, Google’s self-service offering in 2000 and its new click-priced variant in 2002. Keeping the steps separate makes it possible to see what actually changed.
What the click made visible, and what remained uncertain
Charging for a click connects expenditure with an observable action. It is a different promise from simply making space available for a message. Nevertheless, a click remains an intermediate event. It establishes that a link was activated, not that an additional profitable sale occurred. This limit follows from the thing being measured, rather than from any particular historical performance estimate.
An explicitly hypothetical calculation makes the point. If one hundred clicks each cost one unit of currency, the advertising bill is one hundred units. If five purchases follow, the arithmetic cost per observed purchase is twenty. The calculation does not show that all five purchases were caused by the advertisement. Some buyers might have reached the seller anyway. It also leaves out product costs, returns and service expenses. These are illustrative numbers, not recovered campaign results.
The historical importance of greater observability should not be understated. A recorded action can help a business compare messages and revise spending. But precision in counting does not automatically produce precision in explaining economic effects. An attractive ratio can conceal a mistaken causal interpretation if an observed purchase is treated as an entirely additional purchase.
The parties may also define success differently. A platform can earn revenue from a contact that proves unprofitable for the advertiser. A searcher can find a useful answer without buying anything. An advertiser can acquire a customer whose value emerges later. The business involves several relationships, so no single intermediate statistic captures every participant’s experience.
This does not make measurement futile. It makes the definition of the measurement important. The development of search advertising expanded what could be observed while leaving businesses to decide what those observations meant. That combination of improved visibility and continuing uncertainty is more historically informative than the simple claim that digital advertising finally made marketing fully measurable.
Auctions depend on designed rules
Search advertising eventually became a subject for formal economic analysis. Benjamin Edelman, Michael Ostrovsky and Michael Schwarz examined the generalized second-price auction in 2007, showing that its strategic properties differ from those of the Vickrey-Clarke-Groves mechanism. In particular, the apparent resemblance does not make truthful bidding a generally optimal strategy in the same way (Edelman, Ostrovsky and Schwarz, 2007).
For an account centred on 1996, the study is also a reminder about anachronism. A later mathematical analysis does not establish that Open Text used the mechanism being analysed. “Auction” is not a complete specification. The rules governing allocation and payment must be identified before meaningful comparisons can be made.
The broader historical lesson is that a platform market is constructed within a technical environment. Its operator determines important aspects of participation and pricing. Immediate revenue, useful results and a wide advertiser base need not always point in the same direction. An automated calculation can implement a rule, but the choice of rule remains an institutional decision.
An older advertising problem inside a new medium
In a 2015 CHARM paper, Lilly Anne Buchwitz compared radio and internet advertising through a periodization of technology, content, advertising and advertising becoming content. Her framework draws attention to the way a medium’s commercial form develops over time rather than arriving fully determined by its technical invention (Buchwitz, 2015).
Open Text’s Preferred Listings can be interpreted as one moment in such a negotiation. A result was both a guide to another location and a possible advertising vehicle. Companies had paid for visibility long before the web. What changed was the embedding of that payment in a machine-generated response to a request made at that moment.
Older business directories offer a useful comparison, although the analogy has limits. A directory organizes suppliers within a pre-established scheme. A search service constructs a response in interaction with a query. This ties the commercial opportunity more closely to a particular linguistic action. The distinction helps explain why paid search could resemble familiar advertising while also raising distinctive questions about the meaning of an answer.
The more closely an advertisement fits the task, the more readily it may be experienced as useful content. That fit can increase its commercial value. It also makes the explanation of its status important. Relevance does not remove the underlying economic interest. A helpful paid link is still a paid link, and understanding that relationship remains part of understanding the interface.
Open Text’s place in marketing history
The history of search engine marketing does not begin with the familiar appearance of later advertising platforms. Open Text preserves an earlier moment when preferred search visibility was offered commercially and quickly became contentious. Its short-lived programme was not a complete preview of every subsequent system. It shows that technical possibility and an accepted commercial institution are different achievements.
The defensible historical account is therefore specific. Open Text experimented with paid keyword listings in 1996. Later providers connected related opportunities with additional arrangements for pricing, measurement, purchasing and distribution. At the same time, Open Text’s corporate strategy cannot be reduced to this one public-search offering. The business report places it within a wider movement towards enterprise applications.
What remains is a consequential change in intermediation. Businesses could pay for the possibility of being encountered when someone formulated a question. The terms of that purchase shaped the relationship among searcher, advertiser and provider. Open Text’s brief experiment deserves attention because it exposed the central issue before the later systems became familiar: what role does money play in the ordering of an answer?
References
Broder, A. (2002) ‘A taxonomy of web search’, ACM SIGIR Forum, 36(2), pp. 3–10. https://doi.org/10.1145/792550.792552. Full text: https://sigir.hosting.acm.org/files/forum/F2002/broder.pdf.
Buchwitz, L.A. (2015) ‘A Model of Periodization of Radio and Internet Advertising History’, Proceedings of the Conference on Historical Analysis and Research in Marketing, 17, pp. 14–29. https://ojs.library.carleton.ca/index.php/pcharm/article/view/1465.
Edelman, B., Ostrovsky, M. and Schwarz, M. (2007) ‘Internet Advertising and the Generalized Second-Price Auction: Selling Billions of Dollars Worth of Keywords’, American Economic Review, 97(1), pp. 242–259. https://doi.org/10.1257/aer.97.1.242.
Fain, D.C. and Pedersen, J.O. (2006) ‘Sponsored search: A brief history’, Bulletin of the American Society for Information Science and Technology, 32(2), pp. 12–13. https://doi.org/10.1002/bult.1720320206.
Federal Trade Commission (2002) Commercial Alert Response Letter, 27 June. https://www.ftc.gov/sites/default/files/documents/closing_letters/commercial-alert-response-letter/commercialalertletter.pdf.
Google (2000) ‘Google Launches Self-Service Advertising Program’, company announcement, 23 October. https://googlepress.blogspot.com/2000/10/google-launches-self-service.html.
Google (2002) ‘Google Introduces New Pricing For Popular Self-Service Online Advertising Program’, company announcement, 20 February. https://googlepress.blogspot.com/2002/02/google-introduces-new-pricing-for.html.
GoTo.com (2001) ‘GoTo Officially Becomes Overture’, company announcement, 8 October, SEC Exhibit 99.1. https://www.sec.gov/Archives/edgar/data/1060439/000109581101505455/v76181ex99-1.htm.
Open Text Corporation (1996) Annual Report for 1996. https://www.annualreports.com/HostedData/AnnualReportArchive/o/NASDAQ_OTEX_1996.pdf.
Overture Services (2003) Form 10-K for the fiscal year ended December 31, 2002, especially Item 1: Business. https://www.sec.gov/Archives/edgar/data/1060439/000095014803000429/v88074e10vk.htm.
Sullivan, D. (1998) ‘The Search Engine Report March 31, 1998 – Number 17’, Search Engine Watch, issue dated 31 March; webpage dated 30 March. https://searchenginewatch.com/1998/03/30/the-search-engine-report-march-31-1998-number-17/.
Sullivan, D. (2010) ‘Does SEM = SEO + CPC Still Add Up?’, Search Engine Land, 4 March. https://searchengineland.com/does-sem-seo-cpc-still-add-up-37297.
