Disclaimer: The information in this document is provided for your use only. It may not be cited, or reproduced for distribution without the expressed prior consent of the authors. The opinions contained in this document are the authors' and do not necessarily reflect government views or policies.
The authors have benefited from the assistance of many to produce this paper. Industry Canada's Statistics Team, headed by Andy Kormylo, worked together to design the BRITE statistics project. Statistics Canada personnel, Jamie Brunet and Dennis Cooley were helpful in guiding us through the survey design process and for generating the data output tables. The Canadian Association of Internet Providers (CAIP) provided much needed feedback on the ISP portion of the questionnaire and helped ensure a positive response among many in the ISP industrial sector. The authors have also benefitted from the assistance of summer students including Catherine Whilton, who helped assemble the list of companies and Alice Friesen who generated the charts and tables found in this document.
For all of their work and contribution to the success of this project, the authors are grateful.
Internet Service Providers, or ISPs, are a relatively new part of the rapidly developing information technology industry. Non-existent only a few years ago, this dynamic class of companies emerged because of the innovative and entrepreneurial spirit of the Internet's early pioneers. True to their roots in many of the country's research and education institutions, ISPs are creative, experimental and highly adaptive enterprises. And they continue to grow in ways never imagined by the Internet's original creators. To expand and succeed, ISPs develop new business models, create new products and services and continuously deal with myriad challenges.
ISPs have successfully built an awareness and use of a revolutionary new information service and continue to deal with obstacles such as access to capital and an uncertain regulatory environment. They also help usher in the globalization process of Canada's economy. ISPs have challenged, and at the same time, educated policy makers to deal with issues such as intellectual property, the Broadcasting and Telecommunications Acts and the implications of electronic commerce. Because of the wide range of services and products they offer and would like to offer, ISPs are an integral part of the information revolution facing Canada's economy and society.
Despite the important role they play in Canada's growing information economy and industry, very little is actually known about how ISPs operate, what characterises successful firms and what unique issues and concerns face their sector. The federal government has begun dialogues with individual companies and associations and while progress is being made on many issues, there is still much more that needs to be determined. Industry Canada, as the government's flagship microeconomic department has taken on the task to learn more about the ISP sector. The department supports the development of the Internet and the applications and services it offers, be they electronic commerce, health and education applications, community networking or government services.
Many ISPs had their start by solely providing access to the Internet. While Internet access is still the core of most companies, many have evolved to include other services, such as web site hosting, Intranet and electronic commerce solutions. Many companies now sell bundles of complementary services together with Internet access. Only time and study will show which strategy or combination of strategies will be the most effective in the longer-term. This paper will share the results of a first attempt to gather new statistics on the ISP sector in Canada.
The first step of the project was to identify several key questions to be addressed. These included: are ISPs profitable; what do their financial records look like; what are the characteristics of successful firms; what is the range of products and services offered; and, finally, what are some of the important issues for the future growth and development of the sector.
Through Statistics Canada, Industry Canada began a large data collection project for the information technology industry. The project cleaned up existing databases, linked past survey results and records and, where gaps existed, surveyed firms. This project, for the first time, tried to distinguish sub-sectors of the IT industry and tease out their unique characteristics. ISPs are only one sector in the Business Register of Information Technology Enterprises (BRITE) statistics project, but the one about which the least is known. Others sectors included in the BRITE project are IT hardware firms, multimedia companies, software developers and other content producers. The survey asked questions on basic financial and employment data, subscribers, connectivity and a few questions about business practices.
The information-gathering process for ISPs was more involved than other sectors. ISPs are particularly difficult to isolate in Statistics Canada data as the current industry classification system does not have a separate category for ISPs. They are found under various categories. To examine them as a group, a list needed to be developed. Through a variety of sources, mostly online, over 570 potential ISP survey recipients were identified. These companies were then contacted individually to verify their appropriateness to the survey. By the time the survey went out, only 360 companies remained on the list. This drop in numbers is telling. While some of these companies may have been originally misidentified as ISPs, many more ceased to exist as individual firms. Many were bought up or merged with other providers while still others folded up operations. There is no comparable data for others years on the number of ISPs in Canada, but anecdotal evidence and the experience from this survey in list-making, indicates a high level of churn in the ISPs sector.
The BRITE survey, which included the section on ISPs, had over 1500 respondents. Of these, 225 reported at least some revenues from the provision of Internet services. Using these results, Statistics Canada performed a cluster analysis on the respondents to determine which of these could be considered ISPs for the purposes of this paper. The results showed that a total of 102 companies derived at least 75 percent or more of their revenues from providing Internet services. The bulk of the analysis in this paper will focus on this core group of companies.
The response rate for the ISP survey is disappointing at only 30 percent, based on the list of 360 companies. Some areas, most notably British Columbia, are under- represented in the results. As a result of the low response rate, the survey does not compromise a comprehensive census of the industry, nor should it be considered a truly representative sample. The paper will not, therefore, report many of the absolute values of the survey, such as the total number of subscribers or the total revenue of ISP firms. These numbers would give a misleading and incomplete picture of the ISP sector.
The survey and this paper, however, do provide a rich source of information on ISP practices and will discuss the characteristics of the highest revenue-generating firms and the most profitable firms. Where they exist, the paper will also highlight some regional differences. The authors recognise a potential bias in the results, a hazard of empirical work. The rest of the paper will go through these findings and others in greater detail.
The results were looked at in aggregate for Canada and were also broken down into revenue and profitability tertiles. It quickly became apparent that the aggregate data were skewed by the eight largest companies. The revenue and profitability breakdowns are needed, therefore, to get a more detailed picture of the ISP sector in Canada.
Based on the data, it is possible to describe some general characteristics of ISPs. The highest revenue firms are the largest and have the most subscribers, although there is no significant relationship between the number of subscribers and the amount of revenues. They charge the higher prices for their business clients but have proportionately fewer of them. These companies tend to offer more connection options to their customers and pay a lower percentage of their total operating expenditures on their own connectivity. Their total operating expenditures, however, are significantly higher than companies with lower revenues, however, making the high revenue firms also the least profitable.
ISPs were asked to report the number of their subscribers, business and residential, in each Canadian province.
As shown in Figure 1, subscriptions have the following regional breakdown:
As noted in the methodology section, these results are based on the responses to the survey, and so are not a comprehensive indicator of Internet uptake across the regions of Canada.
Across Canada, the majority of subscribers were residential (73.24 percent), rather than business (26.76 percent). There was some degree of regional variation in this proportion (see Fig. 2). In Quebec and Ontario, 36 percent and 33 percent, respectively, of the subscribers were businesses, compared to 18 percent in Atlantic Canada, and 19 percent in both the Prairies, and British Columbia and the Territories. This may reflect stronger economic activity in Ontario and Quebec, and/or may suggest stronger uptake among businesses. This conclusion, however, should be taken very carefully, given possible sample bias.
Grouping by tertiles, there appears to be a positive relationship between revenue and total subscribers, ie. the upper tertile has more subscribers than the middle, which subsequently has more subscribers than the lower (see Fig. 3). When the eight largest companies are removed from the upper tertile, however, this conclusion no longer holds. Thus, the hypothesis that revenues are related to number of subscribers is rejected by this sample. Several possible explanations for this exist: companies with lower numbers of subscribers provide a number of value-added services that justify a premium price; these smaller companies may similarly be targeting lucrative niche markets, or may be specializing only in the business market; the number of customers might be over-reported, if there is a high degree of turnover; and, this may be the result of a deliberate marketing strategy where Internet access itself is a loss leader.
Looking at profitability tertiles, the bottom tertile, ie. the least profitable third of ISPs, had the most subscribers, accounting for 67.27 percent of the total (see Fig. 4a and 4b). This suggests that the biggest companies are not profitable. Furthermore, this suggest that the more profitable companies specialize and/or provide value added services in areas where margins are higher. The bottom line is that simply providing access -- and only pursuing a strategy of getting more customers -- is inferior to a strategy based on provision of value-added services.
ISPs were asked to provide information on the price of their most commonly sold monthly subscription packages, for both residential and business customers, and the services included for that fee. Average package costs for the entire sample were $23.05 for residential customers and $70.69 for business customers. However, the business average price falls to $57.71 when one outlier is removed (that had a price of $1200 per month!).
For residential customers, 15.58 percent of the ISPs provided an unlimited amount of Internet access as part of the package, while the remaining 84.42 percent of ISPs included an average of 64.51 hours in the monthly subscription price. Price levels are fairly consistent across both revenue and profitability tertiles. Additional services provided by ISPs as part of the package:
Other options cited by ISPs in an "other" category include: domain registration service; virtual web or domain park service; free set-up; shareware; access to news groups; and free set-up.
When examined by revenue tertiles, the number of hours included in the subscription price increases with revenues, suggesting that offering more hours attracts more customers and thus provides more revenues (see Fig. 5). Interestingly, when profitability tertiles are examined, the opposite relationship turns up: the least profitable companies provided the most number of free hours. Thus, this may not indeed by a successful strategy, which accords with the previous section conclusion that access is not where the money is to be made in this industry.
Significantly more ISPs provided unlimited access to the Internet for business customers than for residential customers, which in part explains the price premium for a business subscription package (see Fig.6). For business customers, 39.77 percent of the ISPs provided an unlimited number of hours of Internet access. The remaining ISPs provided an average of 71.8 hours as part of the subscription package. By tertiles, the high revenue tertile charged a much higher average price of $131.96 for an average of 67.2 hours on access, with 50 percent providing unlimited access. Middle and lower tertile charged $41 and $46, with 30 and 36.7 percent, respectively, providing unlimited access.
Additional services provided by ISPs as part of the business package:
A number of other services were in an "other" category including: free set-up; directory listing; firewall services; Web site creation; and on-line hours tracking. This demonstrates a market segmentation approach by ISPs, where different classes of customers are offered different prices and packages for services.
ISPs also provide additional services on a pay-per-use basis. Across Canada:
14.71 percent of the ISPs surveyed offer a range of other services on a pay-per-use basis, including: application development; customized names and email addresses; dedicated connections; email paging services; firewalls; hard disk space for Web pages; database services; network design and installation; on-site technical support; programming services; and training.
The variety of additional services offered demonstrates the degree of economies of scope in the ISP industry. That is, average costs fall as ISPs provide additional, related services. This includes those services bundled together as part of a package, as well as those offered a la carte.
ISPs were asked to estimate the percentage of revenues derived from various sources. The results, given in Figure 7, show that the dominant revenue source was residential subscriptions, accounting for an average of 49.54 percent of the total. This was followed by business subscriptions at 30.56 percent. Thus, according to this sample of companies, the combined business and residential subscriptions account for over 80 percent of the industry's revenue. This highlights the relative importance of properly bundling services to offer packages that meet consumer needs.
Installation charges amounted to 5.63 percent of revenues, a relatively small part of the total, but not insignificant as this is a one-time-only charge. Fees for additional Internet access amounted to 2.88 percent of revenues. This can be partly explained by high costs of incremental, pay-per-minute access, which may be subverting additional usage (price elasticity information would be required to assess this explanation). Another factor may be that consumers prefer to have a flat rate for access.
Advertising amounted to a mere 2.55 percent of revenues, a surprisingly small figure. While advertising is becoming more and more prominent on the Internet, this figure suggests it has not penetrated down to the ISP level in a substantial way, and remains in the domain of Web sites. Advertising will likely become more prominent in the future, as local producers use the Internet to reach local customers, and as ISPs at all levels tap advertising as a source of revenue. A very small number of ISPs have attempted a business model that provides free (or very low cost) access to the Internet in exchange for advertising downloaded to the browser once logged in, and this may account in part for the 2.55 percent figure.
Fee services to non-subscribers account for 2.53 percent of revenues. Other sources account for 6.26 percent of revenues, which include consulting, donations, Web page design, licensing, government funding, training, and other services mentioned in the previous section.
Broken down into tertiles, proportion of residential revenues is negatively related to both revenue and profitability tertile, as shown in Figure 8. That is, the high revenue and high profit ISPs tend to derive a smaller proportion of the their revenues from the residential market and more from the business market. This suggests that the smaller players in the market rely more on the residential market, and that the business market, while more difficult to access, is more lucrative.
Given that penetration rates of the Internet into both residential and consumer markets are low relative to the population, there is much room for growth in the ISP market. Competitive tactics surrounding value-added services will play a major role as this market evolves.
The survey also asked ISPs to estimate revenue growth rates for this year and for next year (see Fig. 9). Of the 32 respondents1, average growth this year was 207.12 percent, with average projected growth to be 458.53 percent next year. Clearly, these ISPs are optimistic about the future of the Internet. Interestingly, the companies in the lowest revenue tertile estimated revenue growth of 339.5 percent this year and 697.58 percent next year (see Fig. 10). Similar results appear for the lowest profitability tertile, suggesting a discontinuity between optimism and ability to produce profits.
The survey asked ISPs to estimate operating expenditures. Of the major expenditures, salaries and wages accounted for 42.42 percent, advertising 11.9 percent and leased line charges 32.64 percent (see Fig. 11). As illustrated in Figure 12 which breaks down expenditures by profitability tertiles, salaries and wages were positively related to profitability, ie. they accounted for a relatively larger portion of expenditures, as profitability increased. The most profitable tertile averaged 51.80 percent of expenditures on salaries and wages, compared to 40.91 percent for the middle tertile and 26.91 percent for the bottom tertile. This may be a result of the need for skilled (and relatively expensive) labour to produce high quality Web content or to manage a complex network.
The proportion of expenditures on advertising was negatively related to profitability, with the lowest tertile averaging 20.75 percent of expenditures on advertising, compared to 7.99 percent of the middle tertile and 8.01 percent of the highest tertile.
High revenue companies and high profitability companies paid proportionately less for their leased line charges (see Figure 13). The lowest revenue tertile paid 56.9 percent of expenditures to leased line charges, while the middle tertile paid 23.31 percent and the upper tertile 12.84 percent. The cost of leased lines is clearly an important factor in the ISP business. These numbers suggest economies of scale in the provision of access due to the relative discounts available for leasing large amounts of bandwidth. Thus, leased line charges are a potentially large barrier to entry in the ISP business, particularly as the industry overall grows in scale.
Overall, expenditure figures suggest that as higher revenue companies pay significantly less for leased line charges, they are able to put this savings toward hiring skilled labour to provide technical support and value added services to customers.
The upper revenue tertile had average operating expenditures of $1,615,866, compared to the middle tertile average of $121,699 and the lowest tertile average of $44,492 (see Figure 14). This is indicative of the range of ISP operations across Canada, which go from the very small to the very large. These numbers, graphed in Figure 15, are roughly reversed for the profitability tertile, with the upper tertile having average operating expenditures of $1,669,110, the middle $229,264 and the bottom $153,221. Comparing these numbers to the revenue tertiles suggests that the biggest companies are also the least profitable, and thus there may be plenty of room for small but profitable, niche operators.
Comparing operating revenues to operating expenses, the high revenue tertile is losing money, with average expenditures of $1,615,866, compared to average revenues of $828,910. (Note that these are averages and do not mean that all high revenue ISPs are losing money). The middle revenue tertile is roughly breaking even, with average expenditures of $121,699, compared to average revenues of $129,061. The low revenue tertile is making money, with average operating expenditures of $44,492, compared to average revenues of $35,847. The net values for all these tertiles are shown in Figure 16. The latter numbers suggest that small companies are making a living, but not a killing.
Data on capital expenditures was also requested in the survey, but was unavailable at the time of printing.
ISPs were asked to what extent that various factors impeded the growth of their Internet services. These were marked on a scale of 1 to 5, with 1 being the lowest and 5 the highest. Generally, there was little difference in the results across the revenue or profitability tertiles. However, as might be expected, the bottom revenue tertile was slightly more concerned than the overall averages for every category (except "cost of dial-up lines"). The summary results are given in Table 1.
| Barriers to Growth - Factors | Average Score Out of 5 | Percent Responding 4 or 5 |
|---|---|---|
| Lack of Qualified Staff | 2.87 | 27.1% |
| Cost of Leased Lines | 4.21 | 64.1% |
| Threat of Litigation | 1.85 | 5.8% |
| Cost of Dial-Up Lines | 3.86 | 56.9% |
| Regulatory Environment | 3.28 | 36.5% |
| Competition* | 4.78 | 100% |
| Access to Capital* | 4.78 | 100% |
| Business Management* | 4.78 | 100% |
*Note: These numbers reflect the overall averages from the "Other" category. No more detailed information is available.
Of the five options provided, the factor causing the most concern was "cost of leased lines". The average score for this factor was 4.21, and 64.1 percent of the respondents answered 4 or 5. At the time of the survey, the issue of whether ISPs would be required to pay contribution charges (which would increase the cost of leased lines) had not yet been resolved, so this may be reflected in the responses. Importantly, as noted above, a large proportion of ISP expenditures revolves around this cost, and given the profitability status of most ISPs, a rise in this input cost could be very damaging, particularly for smaller ISPs.
The second highest factor was the "cost of dial-up lines". This factor had an average score of 3.86, with 56.9 percent of respondents indicating this as 4 or 5. Thus, the two largest areas of concern for ISPs revolve around costs at both ends of the ISP business: their links to consumers; and their links to the Internet. Notably, with some degree of variation, there is limited or no competition in these markets, and although prices are regulated, ISPs are somewhat captive. This indicates that government policy should ensure that it helps create a level-playing field that allows a diverse range of companies to offer Internet services. This increases consumer choice and may help foster the greater future growth of the ISP industry in Canada.
The factor "regulatory environment" refers to general uncertainties regarding how a changing regulatory framework, or decisions made therein, could affect the ISP business. This factor received an average score of 3.28, with 36.5 percent giving an answer of 4 or 5. This indicates that ISPs are concerned about this factor, but not overly so, at this point in time. However, as the regulatory environment affects the prices of upstream providers, ISPs would certainly be concerned, as noted above. This goes to the core of the ISP business: providing a value-added service that is largely dependent on the infrastructure offered by bigger telecommunications companies, who may indeed be competitors in the Internet access market.
The factor "lack of qualified staff" received an average score of 2.87, with 27.1 percent of the respondents marking a 4 or 5. The ability to draw highly skilled personnel at reasonable cost is a major issue in the Information and Communications Technologies (ICT) industry generally, though these results indicate that ISPs are concerned but not overly worried about this factor. Looking back to the results of the previous section, presence of skilled personnel does seem to be a factor in providing the value added services that lead to profitability, so the extent that this is a problem for smaller ISPs will affect their growth.
The factor "threat of litigation" refers to fears that the company could be taken to court for possessing or providing access to materials that are considered illegal under either the Criminal Code (eg. child pornography, hate literature) or the Copyright Act (intellectual property, trademarks, copyright). The average response was 1.85, with 5.8 percent indicating a 4 or 5 response, both surprisingly low figures that indicate a general lack of concern on the part of ISPs. As noted in the Industry Canada publication, The Cyberspace is not a No-law Land, there is a lack of case law on such matters, but there are instances where ISPs could be held liable. Actions by trade associations, such as the Canadian Association of Internet Providers, to adopt a model code of conduct for its members, express industry concerns about this issue. Several factors can explain the low results: ignorance about potential liability regarding illegal materials; a feeling of confidence, that since these matters have not been tested in Canadian courts as the have in the US, that they will not be prosecuted; and/or, a reflection of the libertarian, frontier mentality of Internet culture.
The category "other" received an average score of 4.78, indicating that items written in here were of major concern. 36 respondents coalesced around four additional factors that can be drawn out. The biggest of these factors was "competition", cited by 12 of the 36. This refers to the arrival of the large telecommunications companies into the market for Internet access and services. This reflects a fear that, while smaller ISPs are more nimble and able to provide a more customized service, pricing practices of the big players could drive smaller ISPs out of the market. In addition, 8 respondents each cited "access to capital" and "business management", and 3 respondents cited "user issues".
As part of the general survey, respondents were asked to rate the ability of their company to compete in ten areas related to the production and marketing of IT products and services. Self-ratings, from highest to lowest are presented in Table 2 below.
| Areas of Competition | Average Rating out of Five |
|---|---|
| Customer service | 4.39 |
| Quality of IT products and services | 4.34 |
| Labour relations | 4.29 |
| Skill level of employees | 4.16 |
| Price of products and services | 3.91|
| R&D | 3.78 |
| Cost of production | 3.66 |
| Ability to retain qualified personnel | 3.52 |
| Frequency of introducing new products and services | 3.42 |
| Ability to attract qualified personnel | 3.42 |
ISPs generally rated themselves favourably for each of the ten areas.
ISPs were asked whether they had received customer complaints regarding offensive content. Across Canada, 32.4 percent stated that they had received complaints. For the upper revenue tertile, illustrated in Figure 17, 42.9 percent had received complaints, compared to 37.5 percent for the middle tertile and 18.2 percent for the bottom tertile. This is consistent with the hypothesis that the more customers an ISP has, the more likely they are to have had complaints.
Those that had received complaints were asked a subsequent set of questions regarding what their business practices were in response to offensive content complaints. Overall, 18.2 percent of this group said that they had no practices. However, none of the lowest revenue tertile respondents said that they had no practices, indicating a proactive strategy. Of the various practices:
In an "other"category, ISPs also said that they would "refer customer to blocking software", "investigate", and "advise customer of their responsibilities."
This section provides two additional explanations for why ISPs do not consider litigation to be a threat, as noted in the Barriers to Growth Section. In part, many had not received complaints about offensive or illegal materials, and notably, this number amounts to 71.8 percent of the smallest third of companies. In addition, those that had received complaints are engaging in a variety of proactive approaches to combat potential problems. This can explained as a function of protecting their interests through "good business practices."
ISPs were asked to assess what matters most to their customers for a variety of factors. The results are shown in Figure 18. Responses were ranked on a scale of 1 to 5, with 1 being the least important and 5 being the most important. The top response was "customer support", with an average response of 4.43, and 87 percent responding with a 4 or 5. This was followed by two quality of service oriented criteria, "no busy signals", with an average score of 4.27 and 75 percent responding with a 4 or 5, and "no dropped lines", with an average score of 3.99 and 70.3 percent responding with a 4 or 5. The result that customers care most about these critical dimensions of service is consistent with the fact that the ISP business is largely about providing the service of Internet access.
The middle range of respondents to the question includes "value-added services", with an average rating of 3.45 and 54.9 percent providing a 4 or 5 response, and "quality of on-line content", with an average rating of 3.03 and 37.1 percent providing a 4 or 5 response. Of the available choices, the lowest rated was "screening for offensive materials", which had an average rating of only 1.90 and 5.8 percent providing a 4 or 5 response. Thus, while customers are concerned about offensive materials, as noted in the section above, it does not follow that customers want their ISP to act as censor.
An "other" option was provided. 12 of 26 respondents to this option cited "price" or "cost" as a key factor that matters to customers. Other responses included: speed of access or transmission; local content; financial strength; ease of access; and expertise and overall quality of service.
These numbers are generally consistent with demand side information from another source commissioned by Industry Canada, as part of the AC/ Nielsen Canadian Internet Survey 1996. The results are given below in Table 3. Provided with a list of common problems cited by Internet users, households were asked to respond whether they had encountered any of them. In this survey, the most common problem was "busy signals", cited by 37 percent of those surveyed, followed by "interrupted connections", cited by 29 percent, and "poor customer service", cited by 9 percent. Only 3 percent cited inadequate screening of content. Notably, 39 percent said that they had not experienced any problems with their service provider.
| Problems Encountered With Service Provider | Percent of Households |
| Busy signals | 37% |
| Interrupted connections | 29% |
| Poor customer service | 9% |
| Inadequate screening of content | 3% |
| No problems experienced | 39% |
ISPs were asked if they offer regular dial-up service and 83.3 percent responded that they did. The difference is accounted for by the fact that some ISPs in the survey were specialized, offering, for example, only Web site hosting. Figure 19 shows that there was a modest, but positive, relationship between profitability tertile and dial-up service, with 84.4 percent of respondents in the upper tertile responding positively, compared to 82.1 percent in the middle tertile and 76.7 percent in the bottom tertile. This may simply confirm that access is a core component of any Internet services bundle.
ISPs were then asked to identify the connection options they provided to their customers. The results are as follows:
An additional 11.8% (15 responses) offer "other" options. These include: T1 connections (5 responses); wireless (5 responses); ADSL (2 responses); and, ATM (2 responses). Note that these numbers do not add up to 100 percent, as many ISPs provide more than one option.
Sorting by revenue tertiles, for each connection option (see Fig. 20), there was a positive relationship between the tertile and offering the connection option, ie. the upper tertile had a consistently greater percentage of positive responses to each connection option, relative to the middle. This suggests the presence of economies of scope in offering different connection options. However, offering different connection options is not necessarily very profitable. For example, the low profitability tertile had a higher percentage offering 1-800 or PDN connections. This suggests that the incremental cost of a 1-800 or PDN line may not really be necessary for the ISP to take on, relative to the benefits derived. Overall, for profitability, there was a negative relationship between tertile and each connection option.
Finally, ISPs were asked from whom they purchased their trunk lines. Responses are given below in Table 4.
| Trunk Line Suppliers | Percent Purchased |
| Canadian carrier (Stentor, Sprint Canada, AT&T Canada, etc.) | 41.8% |
| Canadian-based ISP (iSTAR, Hook-up, etc) | 31.37% |
| Direct connection to US Internet | 5.88% |
| Cable company* | 4.9% |
| Other sources (CA*net, regional networks, etc) | 19.61% |
* regionally, all 6 respondents from Atlantic Canada responded that they purchase from cable companies.
Internet service providers are a diverse group of companies. While providing Internet access is still a core activity and residential customers the largest overall source of revenue, other activities are more profitable. The paper concludes that simply providing access and only pursuing a strategy of getting more customers is inferior to a strategy based on providing value-added services.
One successful strategy appears to be one that segments the residential and business market by providing different classes of service. Classes of service are distinguished by the hours of access offered and the bundle of value-added services offered. If customers are placed in the appropriate class of service, ISPs are able to capture as much of the Internet population as possible -- from light to heavy users. This helps generate as much revenue as possible for the companies, while best meeting customer needs.
The variety of service classes and the range of products offered -- from database services to training -- demonstrate the degree of economies of scope in the ISP industry. ISPs are willing to expand into new service areas as long as average costs fall with each additional or complementary services. ISPs continue to introduce new services and products, showing that the full economies of scope may not be yet fully realised.
Many of the small players have not yet realised the benefits of economies of scope. The study of revenues indicates that the smaller players rely more heavily on the residential market and that the business market, while more difficult to access, is more lucrative. Smaller firms may have difficulty accessing the more lucrative business market. Nevertheless, given that the penetration rates of the Internet into both residential and consumer markets are low relative to the population, there is much room for growth in the ISP market.
In addition to economies of scope, economies of scale play an important role in the profitability status of Canada's ISPs. The above analysis suggests that economies of scale in the provision of Internet access exist due to the relative discounts available to companies leasing large amounts of bandwidth. As a result, leased line charges are a potentially large barrier to entry in the ISP business, particularly as the industry overall grows in scale. The expenditure figures obtained from the survey suggest that as higher revenue companies pay significantly less for leased line charges, they are able to put this savings toward hiring skilled labour to provide technical support and value-added services to customers.
The expenditure figures, when examined by revenue tertiles also suggest that the biggest companies are also the least profitable. This result does not support the commonly discussed scenario that the ISP market will collapse into an oligopoly where only a few large companies exist. The survey results suggest that there may be plenty of room for small but profitable niche operators.
Small companies, however, may be very sensitive to the price of their factors of production -- items such as leased lines, dial-up lines and personnel issues. As noted in the analysis above, a large proportion of ISP expenditures revolves around the cost of leased lines. Given the profitability status of most ISPs, a rise in this input cost could be very damaging, particularly for smaller ISPs. Dial-up lines is another source of cost concern, though these services are, to a certain extent, regulated.
Competition ranks as one of the chief concerns of many of ISPs, particularly competition resulting from the arrival of the large telecommunications companies into the ISP market. This reflects the fear that, while smaller ISPs are often more nimble and able to provide a more customised service, the pricing practices of the big players could drive the smaller ISPs out of the market. The concern over competition issues is much greater than other potential threats to the ISP industry, such as the threat of litigation. Although government policy is to increase the amount of competition in this area, policy should ensure a level playing field with a diversity of companies offering Internet services to maximise consumer choice. An outcome that allows for significant increases in the cost of lines will pose a danger to the growth and success of smaller ISPs.
This study has been able to draw out the characteristics of ISP industry in Canada. By examining the details of the responding ISPs, the paper has shown that the ISP industry is robust and offers a spectrum of services often bundled to capture market segments. As the competition for Internet access drives down the profits in that area, firms developed a wide variety of business strategies, company structures and, most importantly, value-added products and services to better compete along other margins. While the large companies are able to realise economies of scale, smaller companies are more nimble and profitable. The success of these smaller firms, however is often fragile and is potentially very sensitive to competitive threats, such as the price of lines.
Ultimately, this study is a snapshot of a dynamic, evolving industry that is at the core of an economy based on information, knowledge and networks. Policy should continue to foster the growth of innovative new companies offering an increasing variety of new services to Canadians. In doing so, Canada will keep its lead as the knowledge economy unfolds. Footnote 1: This represents about a third of the sample. Seemingly, many companies were uncomfortable providing this information.