WarrZone 2001

back to WarrZone

Science Portals on the Internet

Wendy Warr
Wendy Warr & Associates,
6 Berwick Court,
Holmes Chapel,
Cheshire, CW4 7HZ,
Tel and Fax +44 1477 533837
Email: wendy@warr.com
WWW: http://www.warr.com


An Internet portal is a Web site that provides an entry point to the Internet and offers value-added services such as directories, searching, news, and links to related Web sites, and perhaps “community” facilities such as e-mail, message boards, chat lines, forums and news groups. The name suggests that a portal is a doorway or gateway to other places. However, a recent survey carried out by Booz Allen & Hamilton and NetRatings (1) showed that Internet users now see portals as destinations in themselves rather than as gateways to the Web: users spend an average of 4.5 hours at portals each month, three times more than they spend at shopping or entertainment sites. The defining characteristics of a portal are content, communication, sometimes a community, and often, commerce. The better portals aim to offer not just content but context (interlinking, multimedia, statistics), and many of them are now emphasizing “personalization”, i.e., a facility in which a user can customize a version of the Web site to best match his own interests.

Personalization encourages a user to make repeat visits. Web site owners refer to this characteristic of a site as “stickiness” or “stickability”, or hope that their visitors are “locked in”. The site HealthandAge.com, which has many senior citizens amongst its visitors, lets a user choose font size and, when he revisits, the home page appears in the font size he chose on his last visit. Other examples of personalization are pre-filling application forms, not displaying advertisements for products the visitor has already bought, and when a visitor logs in for a second time, taking him to the page he was at when he last left. In the past users have resented being asked to register. If personalization offers sufficient advantages to users, and is not be too much effort, they can be persuaded to part with personal details that are of benefit to the portal owner.


Portals can be divided into various categories (2):

  • search engines
  • home pages of Internet Service Providers
  • vertical portals, or “vortals” e.g., chemistry.org, ChemWeb.com
  • subject-based gateways (also vortals) e.g., ChemIndustry.com
  • the sites of intelligent agents or bots, e.g., shopping bots
  • intranet portals
  • extranet portals

Commercial portals are playing the “new game” adopted by dot.com companies. This game is the business model adopted by Amazon.com, AOL.com, Yahoo and eBay, for example, where it replaces the manufacturing model. The new game has a “viral” model, one of explosive growth, rather than a “make and sell” model and it makes no mention of profits: profit is an “old economy” metric. The dot.com company grows its installed user base, gains mindshare, attracts linkages, grows the network, and thus builds brand equity, which leads to further growth in the user base. Funding comes from the capital markets with anticipation of long term profits after brands have been built and user loyalty has been gained. However, the Harvard Business Review stopped using the term “new economy” in its editorials in 1999. There never was a new economy. The economy is the same; it is the technology that is new (3).

A “sense of belonging” brings visitors back to a virtual community. According to PricewaterhouseCoopers, the sense of belonging comes from precisely tailored content, identification with the brand, awareness of other like-minded users, ability to interact with other like-minded users, opportunity to shape Web site development, and mutual benefits of participation (4). A site owner can derive value, but not necessarily income from visitors. The majority of community-building activities do not generate revenue, they create an environment that can stimulate income generation. However, value does relate to income. Whereas suppliers sell features, customers buy benefits. Customers do not want what suppliers sell, they want to benefit from what happens when they use what suppliers sell. “Value” is the economic consequences the customer receives from the benefits (5).

Having built a critical mass of users, a portal may well turn to e-commerce. At this stage it must realize that e-commerce is not (only) about transactions: it is more about interactions (6). Great brands get the customer experience right every time, over time. Good advice is to know your users, analyze your users, and spend a lot on research. Dot.coms have traditionally concentrated much more on advertising than on research. Not only does the dot.com need a sound understanding of its users, but also its partners and advertisers need this information. Partnerships are a very important factor in the growth and success of portals.


Since attracting large numbers of visitors is so important in the new game, it is worth looking at some measures of popularity but before we do so, we should consider the dangers inherent in statistics of Web usage. One problem is “caching”: a user views a Web page but when he views it on another occasion he does so by automatically reading a version of it stored on his hard disk, or a disk on a local or wider network. Thus he may read the page multiple times but the statistics show that he has only read it once (7). Another problem is the so-called “invisible Web”: pages not easily accessible to search engines or other counters. Search engines also vary in the number of pages they index (8).

Snyder and Rosenbaum (9) state:

“The limitations which are inherent in commercial search engines should cause researchers to have reservations about any conclusions that rely on these tools as primary data-gathering instruments”.

Search engines give highly variable results, have limited functionality, poorly documented functions and opaque search logic, and change functions over time (9).

Willett and co-workers (10) have also discussed the pitfalls involved in “webometric analysis” and they list many useful references.

Despite these reservations, in July 2001 we expanded on an analysis of chemistry sites on the Web originally carried out by Heller (11) in April 2001. He used a software product, Alexa, which is incorporated in Netscape’s “What’s Related” feature and which can be downloaded free for use with Microsoft Internet Explorer. This tool can be used to determine the “popularity” of a Web site and the number of other sites on the Web which link to it. The numerical values from Internet Explorer and Netscape differ but the rankings are very similar so we report here only the Netscape figures. Heller chose sites that are well known to chemists and we used much the same set, whether or not they should be classed as portals (see Table 1).

In Table 2 we show how many other sites link to our chemistry sites. The absolute values should be treated with caution for some of the reasons already mentioned, and it should also be noted that there is not necessarily a good correlation between the “backlinks” as recorded on a site’s own log and the links counted by search engines (12). Relatively, however, the figures seem significant: it is not surprising that there are millions of links to Yahoo yet only 200 cite e-chemicals.

After Yahoo, patent information sites are the most cited, and after that come publishers. Various software vendors and vortals follow. Highly focused sites gravitate to the bottom. TheScientificWorld is much newer than most of the sites in the Table: perhaps few subject gateways have noted it yet. Given the choice would you advertise on Combichem.net, or on Yahoo? You might surmise from this Table that advertising on Yahoo would produce more “click-throughs”. Life is not that simple. (Incidentally, warr.com does not carry advertisements or intend to make any direct revenues.)

It seemed sensible next to try counting “links-to” using one of the five or more search engines that offer that facility. We tried Google. The absolute numbers output were lower (e.g., there are apparently only 1,100,000 links to Yahoo, not 4,807,434) but the ranking was not very different. The bottom 7 sites were the same but ordered differently. The top 9 sites were ranked differently but were the same sites as those that Netscape found, with two exceptions: Sheffield Chemdex appeared much higher (at 7 not 15) and Delphion was much lower (at 13 rather than 2). Perhaps Delphion was lower than expected because it was recently renamed. We experimented briefly with eliminating self-citations (10) but this did not seem to affect the results very much.

After this we tried out “popularity” (Table 3). Here the lower the number the greater the popularity, since Netscape is stating, for example, that Yahoo is the fourth most popular site on the Web. It is worth noting that there are more than 1000 million Web sites (8) and many thriving companies rank below 2 million. We are concerned here only with portal-like companies, which tend to rank higher.

The popularity ranking of Table 3 is not too different from the ranking by links-in (i.e., general portals at the top, patents after that, then publishers, then software vendors) but note that ChemIndustry is very much higher. When Heller did his analysis, ChemIndustry was below ChemWeb.com. It could well be significant that Chemical Abstracts Service has recently started offering ChemIndustry in its eScience service on STN, but a source close to ChemIndustry has denied that their statistics show the influence of CAS. Another point worth noting is that the “popularity” results change faster than the “links” results. In a few months time, the rise of ChemIndustry may be reflected by the number of sites linking to it.


Ranking has very little influence on profitability. Combichem.net (number 28 in popularity) makes a profit, while, as far as we can tell, ChemIndustry (number 3) does not. Cheminfo and Chemdex are not-for-profit ventures. Companies who wish to make revenues from a Web venture have just three options: advertising, sales (e-commerce), and commission on someone else’s sales. ChemIndustry makes its money from a model like that of goto.com i.e., by giving precedence in search listings to companies who pay to be ranked higher. This is akin to advertising in our argument.

Is advertising a viable revenue model in the long term? Opinions have always varied. E-commerce companies used inflated share valuations to subsidize the costs of achieving an advertising base, hoping to keep themselves afloat until e-commerce transactions produced profits. For the dot.coms, this business model failed when the bottom fell out of Internet stocks. Newspapers and television companies are also struggling with lower advertising revenues. Yet it could be that a few selected Web sites will continue to attract significant advertising revenue, and survive in conjunction with e-commerce.

The Booz Allen and Hamilton study (1) concludes that portals need to adopt new revenue sources such as paid content or services, in order to survive. It was found that only 1% of visitors to portals click on banners (graphic images as advertisements, often spread across the width of a Web page). The study advises advertisers to engage in co-sponsorships with portals rather than buying banners. However, as the banner advertisement share of the advertisement market falls, newer advertising formats are appearing and new revenue calculations based on cost-per-click and cost-per-sale are coming in alongside the common advertising metric of cost per thousand page impressions.

The downturn in advertising has led to a trend away from free information content on the Web. Free content only makes business sense if you have the same revenue model as a narcotics dealer. Having built up a big community of readers, companies such as Delphion and Moreover.com have started to charge subscriptions for content that was previously free. Publishers have long been offering tables of contents free in order to sell articles on a pay-per-view basis but they much prefer subscribers, a revenue model which offers better margins, and with which they are familiar and comfortable.

Portal users expect content must be very up-to-date, relevant, and of good quality. They are deterred by “dead” hyperlinks. Content publishers are adopting the XML publishing standard to enable them to reuse information (amongst other reasons). They make partnerships and rebrand partner content as their own. Branding and creating a trusted identity are important. Names such as chemistry.org, and ChemWeb.com trip easily off the tongue; highthroughputexperimentation.com says what it does but it is more tiresome to pronounce and type.

Partnerships are critical to a science portal if it is to make commission on selling someone else’s products. Partners must be chosen with care: an alliance with one company may prevent later deals with its competitors. ChemWeb.com is owned by one publisher (Elsevier Science) but aggregates content from many. Launched in April 1997, it has acquired hundreds of thousands of members but has never achieved profitability. Despite this, the American Chemical Society (a not-for-profit organization) continues to invest in chemistry.org which it appears to view as a competitor to ChemWeb.com. Any sort of partnership between these two is unlikely.


Newer bidders for chemists’ attention continue to enter the market. We will compare just two as examples. The first is combichem.net, an interesting and unusual example because the owner, Technology Networks, is not only profitable but makes its profits largely from online advertising. Technology Networks runs 5 or 6 portals, including some new ones e.g., highthroughputexperimentation.com. Labonachip.com is currently the most popular. The company is profitable, making 30-35% gross profit on turnover, after paying 4-5 staff and all development costs, hardware etc., although the company’s founder is not yet taking his cut. Another interesting feature is that about 90-95% of the expected audience was reached within the first 6 months. In contrast, ChemWeb.com, which already has about 300,000 members, is still appealing for more.

Combichem.net is an example of what can be achieved by maintaining a tight focus (combinatorial chemistry). This focus is attractive to advertisers, who provide most of the revenues. Email newsletters attract more traffic and also make money. The site has a strong emphasis on meetings and co-promotes meetings in a partnership with ACS Prospectives. It also offers daily news (it emphasizes currency) and in-depth coverage of topics. The site is in its third redesign.

A second example is TheScientificWorld. This site is aimed at research scientists in the life and environmental sciences. It provides them with information services such as alerting, verification from several abstracting services and full text delivery to the desktop. In addition, it offers a digital publishing environment through a proprietary, multidisciplinary publication, TheScientificWorldJOURNAL. Through a relationship with Alchematrix, a wholly owned subsidiary of Fisher Scientific, it also enables scientists to purchase scientific equipment and supplies from TheScientificWorld’s Web site.

The site’s focus is on providing the end-user scientist with a comprehensive information service at the point of decision making. Pricing is flexible: customers can buy site licenses, open accounts for pay-per-view purchases or buy by credit card. The revenue earners are the company’s proprietary publications, the document delivery service and revenues from the sale of equipment, supplies and biological resources. Some advertising and sponsorship revenues are anticipated. Alerting and verification are provided free at this time. WorldMEET, a database of forthcoming meetings and conferences, and NewsLAB, a customizable news service, are free.

The company believes that since information content cannot be created free of any cost, content cannot be given away totally free. All its proprietary content is peer reviewed in the belief that quality control is essential to the company’s future. The rejection rate on TheScientificWorldJOURNAL (which has over 300 well-known scientists on its editorial board) is over 50%. Advertising is a small line item in the company’s revenue forecasts: less than 10%. Direct sales and commission sales are forecast at being equal, i.e., 45%. The company anticipates reaching profitability in the final quarter of 2002. It will be interesting to see if it succeeds.


When a revolution in technology first occurs, it is frequently discussed using the familiar terminology of an earlier technological era. The “horseless carriage” was the predecessor of the automobile. What will be the successor to the “portal”? Web sites have evolved from simple home pages and “brochureware” to e-markets, communities, and portals but the end is not in sight. Very few portals make money; all are experimenting with business models. No big winners have yet emerged. Portals with established “bricks and mortar” companies (or learned societies) behind them have a more stable financial environment in which to develop but even they will have to make profits or otherwise justify their existence soon if world economies show no signs of recovery. Mainstream companies which fail to integrate the new technology into their activities are just as likely to fail as the pure-play dot.coms.


  1. Quoted at http://www.ft.com on February 23, 2001.
  2. J. Rowley; Managing Information 1/2 62 (2000)
  3. D. Junger; InPharm World Review, Issue 34, 21 March 2001, http://www.inpharm.com/worldreview/210301.html#view visited September 17, 2001.
  4. J. Old, Revenue Models for Virtual Communities. Paper given at the Fourth International Conference on Virtual Communities, London, 20-21 June 2001.
  5. G. E. Cressman; Chimica Oggi 6 65 (2001).
  6. M. Wild; Building Sustainable Regional Communities Online: Lessons from Western Australia. Paper given at the Fourth International Conference on Virtual Communities, London, 20-21 June 2001.
  7. J. Goldberg; Why Web Usage Statistics are Worse than Meaningless.
    http://www.goldmark.org/netrants/webstats visited August 9, 2001.
  8. Two well-known sites maintain statistics on search engines: http://searchenginewatch.com/ and http://www.notess.com/search, visited September 16, 2001.
  9. H. Snyder, H. Rosenbaum; J. Doc., 55 375 (1999).
  10. O. Thomas, Willett, P; J. Inf. Sci., 26 421 (2000) and references 10-17 and 20 therein.
  11. S. R. Heller; Chemical Information and the Web: Past, Present, and Future. Paper given at the 221st National Meeting of the American Chemical Society, San Diego, April 1-5, 2001.
  12. M. Thelwall; Aslib Proc. 53 217 (2001).

Table 1. A selection of Web sites

Site Location
Accelrys msi.com,later accelrys.com
Advanced Chemistry Development acdlabs.com
American Chemical Society (ACS) chemistry.org
CambridgeSoft camsoft.com
ChemConnect chemconnect.com
ChemFinder chemfinder.com
ChemGuide chemguide.de
Chemical Abstracts Service (CAS) cas.org
ChemIndustry chemindustry.com
ChemWeb.com chemweb.com
Claessen Website claessen.net
Combichem.net combichem.net
Daylight daylight.com
Delphion, formerly IBM Patents delphion.com
Derwent derwent.com
E-chemicals e-chemicals.com
Elsevier Science elsevier.com
Gary Wiggins' Cheminfo indiana.edu/~cheminfo/
International Union of Pure and Applied Chemistry (IUPAC) iupac.org
John Wiley & Sons wiley.com
MDL Information Systems mdli.com
Royal Society of Chemistry (RSC) ChemSoc chemsoc.org
SciQuest sciquest.com
Sheffield University's Chemdex chemdex.org
TheScientificWorld thescientificworld.com
US Patents uspto.gov
WebElements webelements.com
Wendy Warr & Associates warr.com
Yahoo yahoo.com

Table 2. Links into Web sites from other sites (Netscape version, July 2001)

Site Links to Site
Yahoo 4807434
Delphion 35030*
US Patents 32775*
Elsevier 31192
Wiley 17709
Accelrys 14906*
ACS 14103
RSC ChemSoc 10274
WebElements 9151
CAS 8791
MDL 7726
ChemWeb.com 6705
CambridgeSoft 5115
ChemFinder 3980*
Sheffield Chemdex 3776
SciQuest 2950
IUPAC 2583
ACD Labs 1932
Derwent 1864
ChemConnect 1720
Daylight 1422
Gary Wiggins' Cheminfo 1248
Wendy Warr & Associates 585
Claessen Website 489
ChemIndustry 429
E-chemicals 207
Combichem.net 183
ChemGuide 8*
TheScientificWorld 1

* April figure since Netscape would not produce a score for the site in July.

Table 3. "Popularity" according to Netscape (July 2001)

Site Netscape Popularity
Yahoo 4
Delphion 2700
ChemIndustry 3282
US Patents 5102
Elsevier 7367
WebElements 13147
Wiley 21501
ChemWeb.com 22530
ChemFinder 29753*
Gary Wiggins' Cheminfo 42675
CAS 45227
ACS 49417
SciQuest 57030
RSC ChemSoc 63294
ACD Labs 88551
Accelrys 94995*
MDL 102143
ChemConnect 108790
TheScientificWorld 138261
E-chemicals 206377
Derwent 290631
CambridgeSoft 293523
Sheffield Chemdex 293990
Daylight 328399
IUPAC 364676
Claessen Website 520126
ChemGuide 580810*
Combichem.net 718107
Wendy Warr & Associates 1228849

*April figure since Netscape would not produce a score for the site in July.