Figure 3.4 shows the way in which companies or home users connect to the Internet. The diagram is greatly simplified in that there are several tiers of ISPs. A user may connect to one ISP which will then transfer the request to another ISP which is connected to the main Internet backbone.
High- speed broadband is now the dominant home access method rather than the previ- ously popular dial-up connection.
However, companies should remember that there are significant numbers of Internet users who have the slower dial-up access. Ofcom (2010) reported that the proportion of individuals with access to broadband services in the UK is 71% (fixed broadband 65% and mobile broadband 15%).
Broadband commonly uses a technology known as ADSL or asymmetric digital sub- scriber line, which means that the traditional phone line can be used for digital data trans- fer. It is asymmetric since download speeds are typically higher than upload speeds. Small and medium businesses can also benefit from faster continuous access than was previously possible.
The higher speeds available through broadband together with a continuous always-on connection have already transformed use of the Internet. Information access is more rapid and it becomes more practical to access richer content such as digital video.
Issues in management of ISP and hosting relationships
The primary issue for businesses in managing ISPs and hosting providers is to ensure a satis- factory service quality at a reasonable price. As the customers and partners of organisations become more dependent on their web services, it is important that downtime be minimised.
But severe problems of downtime can occur, as shown in Box 3.4, and the consequences of these need to be avoided or managed.
Speed of access
A site or digital business service fails if it fails to deliver an acceptable download speed for users. In the broadband world this is still important as digital business applications become more complex and sites integrate rich media. But what is acceptable?
Research supported by Akamai (2006) suggested that content needs to load within 4 sec- onds, otherwise site experience suffers. The research also showed, however, that high prod- uct price and shipping costs and problems with shipping were considered more important than speed. However, for sites perceived to have poor performance, many shoppers said they would not be likely to visit the site again (64%) or buy from the e-retailer (62%).
Hosting provider A service provider that manages the server used to host an organisation website and its connection to the Internet backbones.
Dial-up connection Access to the Internet via phone lines using analogue modems.
Broadband connection
Access to the Internet via phone lines using a digital data transfer mechanism.
In 2010 Google introduced speed as a signal into its ranking algorithm, effectively penalis- ing slower sites by positioning them lower in its listings. The announcement suggested that this would only affect 1% of sites (Google, 2010). The announcement also summarised more recent research from Akamai which suggested that less than 2 seconds was now an accept- able download speed for e-commerce site users. With Google taking page download speed into account when ranking some particularly slow sites, it’s worth comparing your ‘page weight’ or bloat compared to other sites. The average page size is 320 Kb according to this compilation: http://code.google.com/speed/articles/web-metrics.html.
Box 3.4 How long before you become impatient?
Usability specialist Jakob Nielsen noted (Nielsen, 1994) that the basic advice for response times for human– computer interaction has been about the same for 30 years.
He describes these requirements for response:
● 0.1 second is about the limit for having the user feel that the system is reacting instantaneously, meaning that no special feedback is necessary except to display the result.
● 1.0 second is about the limit for the user’s flow of thought to stay uninterrupted, even though the user will notice the delay. Normally, no special feedback is neces- sary during delays of more than 0.1 but less than 1.0 second, but the user does lose the feeling of operating directly on the data.
● 10 seconds is about the limit for keeping the user’s attention focussed on the dia- logue. For longer delays, users will want to perform other tasks while waiting for the computer to finish, so they should be given feedback indicating when the com- puter expects to be done. Feedback during the delay is especially important if the response time is likely to be highly variable, since users will then not know what to expect.
Speed of access to services is determined by both the speed of the server and the speed of the network connection to the server. The speed of the site governs how fast the response is to a request for information from the end- user. This will be dependent on the speed of the server machine on which the website is hosted and how quickly the server processes the information. If there are only a small number of users accessing information on the server, then there will not be a noticeable delay on requests for pages. If, however, there are thou- sands of users requesting information at the same time then there may be a delay and it is important that the combination of web server software and hardware can cope. Web server software will not greatly affect the speed at which requests are answered. The speed of the server is mainly controlled by the amount of primary storage (for example, 1024 Mb RAM is faster than 512 Mb RAM) and the speed of the magnetic storage (hard disk). Many of the search- engine websites now store all their index data in RAM since this is faster than reading data from the hard disk. Companies will pay ISPs according to the capabilities of the server.
An important aspect of hosting selection is whether the server is dedicated or shared (co-located). Clearly, if content on a server is shared with other sites hosted on the same server then performance and downtime will be affected by demand loads on these other sites.
But a dedicated server package can cost 5 to 10 times the amount of a shared plan, so many small and medium businesses are better advised to adopt a shared plan, but take steps to minimise the risks with other sites going down.
For high- traffic sites, servers may be located across several computers with many pro- cessors to spread the demand load. New distributed methods of hosting content, summa- rised by Spinrad (1999), have been introduced to improve the speed of serving web pages for very large corporate sites by distributing content on servers around the globe, and the most
Dedicated server Server contains only content and applications for a single company.
widely used service is Akamai (www.akamai.com). These are used by companies such as Yahoo!, Apple and other ‘ hot- spot’ sites likely to receive many hits.
The speed is also governed by the speed of the network connection, commonly referred to as the network ‘bandwidth’. The bandwidth of a website’s connection to the Internet and the bandwidth of the customer’s connection to the Internet will affect the speed with which web pages and associated graphics load. The term is so called because of the width of range of electromagnetic frequencies an analogue or digital signal occupies for a given transmission medium.
As described in Box 3.5, bandwidth gives an indication of the speed at which data can be transferred from a web server along a particular medium such as a network cable or phone line. In simple terms bandwidth can be thought of as the size of a pipe along which informa- tion flows. The higher the bandwidth, the greater the diameter of the pipe, and the faster information is delivered to the user. Many ISPs have bandwidth caps, even on ‘unlimited’
Internet access plans, for users who consume high volumes of bandwidth.
Bandwidth Indicates the speed at which data are transferred using a particular network medium. It is measured in bits per second (bps).
Box 3.5 Bandwidth measures
Bandwidth measures are in bits per second where one character or digit, such as the number ‘1’, would be equivalent to 8 bits. So a modem operating at 57,600 bits per second (57.6 kbps) will transfer information at 7,200 characters per second (57,600 ÷ 8). When selecting an ISP or hosting provider it is important to consider the bandwidth of the connection between the ISP and the Internet. Choices may be:
● ISDN – 56 kbps up to 128 kbps
● Frame relay – 56 kbps up to a T1 communications channel (1.55 Mbps)
● Dedicated point-to-point – 56 kbps up to T3 (45 Mbps): connected to the Internet backbone.
kbps is one kilobit per second or 1,000 bps (a modem operates at up to 56.6 kbps) Mbps is one megabit per second or 1,000,000 bps (company networks operate at 10 or more Mbps)
Gbps is one gigabit per second or 1,000,000,000 bps ( fibre- optic or satellite links operate at Gbps).
A major factor for a company to consider when choosing an ISP is whether the server is dedicated to one company or whether content from several companies is located on the same server. A dedicated server is best, but it will attract a premium price.
Availability
The availability of a website is an indication of how easy it is for a user to connect to it. In theory this figure should be 100 per cent, but sometimes, for technical reasons such as fail- ures in the server hardware or upgrades to software, the figure can drop substantially below this. Box 3.6 illustrates some of the potential problems and how companies can evaluate and address them.
Box 3.6 Preventing wobbly shopping carts
The extent of the problem of e-commerce service levels was indicated by The Register (2004) in an article titled ‘Wobbly shopping carts blight UK e-commerce’. The research showed that failure of transactions once customers have decided to buy is
often a problem. As the article said, ‘UK E‑commerce sites are slapping customers in the face, rather than shaking them by the hand. Turning consumers away once they have made a decision to buy is commercial suicide.’ The research showed this level of problems:
● 20% of shopping carts did not function for 12 hours a month or more.
● 75% failed the standard service level availability of 99.9% uptime.
● 80% performed inconsistently with widely varying response times, time- outs and errors – leaving customers at best wondering what to do next and at worst unable to complete their purchases.
Similarly, SciVisum, a web- testing specialist, found that three- quarters of Internet marketing campaigns are impacted by website failures, with 14% of failures so severe that they prevented the campaign meeting its objectives. The company surveyed marketing professionals from 100 UK-based organisations across the retail, financial, travel and online gaming sectors. More than a third of failures were rated as ‘serious to severe’, with many customers complaining or unable to complete web transactions.
These are often seen by marketers as technology issues which are owned by others in the business, but marketers need to ask the right questions. The SciVisum (2005) research showed that nearly two- thirds of marketing professionals did not know how many users making transactions their websites could support, despite an average transaction value of £50 to £100, so they were not able to factor this into campaign plans. Thirty- seven per cent could not put a monetary value on losses caused by cus- tomers abandoning web transactions. A quarter of organisations experienced website overloads and crashes as a direct result of a lack of communication between the two departments.
SciVisum recommends that companies do the following:
1 Define the peak visitor throughput requirements for each customer journey on the site. For example, the site should be able to support at the same time: approxi- mately ten checkout journeys per second, thirty add-to-basket journeys per sec- ond, five registration journeys per second, two check-my- order- status journeys per second.
2 Service- level agreement. More detailed technical requirements need to be agreed for each of the transaction stages. Home- page delivery time and server uptime are insufficiently detailed.
3 Set up a monitoring programme that measures and reports on the agreed journeys 24/7.
Service level agreements
To ensure the best speed and availability a company should check the service level agree- ments (SLAs) carefully when outsourcing website hosting services. The SLA will define confirmed standards of availability and performance measured in terms of the latency or network delay when information is passed from one point to the next. The SLA also includes notification to the customer detailing when the web service becomes unavailable with rea- sons why and estimates of when the service will be restored.
Security
Security is another important issue in service quality. How to control security was referred to in the earlier section on firewalls and is considered in detail in ‘Focus on Security design for digital business’ (Chapter 11, p. 572).
Service level agreement (SLA) A contractual specification of service standards a contractor must meet.
Managing internal digital communications through intranets and extranets
In Chapter 1 , we introduced the concept of intranets and extranets.
Intranet applications
Intranets are used extensively for supporting sell- side e-commerce from within the mar- keting function. They are also used to support core supply chain management activities as described in the next section on extranets. Today, they are typically deployed as web- based services supplemented by messages and alerts delivered by email or when users login to a company network. A marketing intranet has the following advantages:
● Reduced product life cycles – as information on product development and marketing campaigns is rationalised we can get products to market faster.
● Reduced costs through higher productivity, and savings on hard copy.
● Better customer service – responsive and personalised support with staff accessing cus- tomers over the web.
● Distribution of information through remote offi ces nationally or globally.
Intranets are also used for internal marketing communications since they can include:
● Staff phone directories.
● Staff procedures or quality manuals.
● Information for agents such as product specifi cations, current list and discounted prices, competitor information, factory schedules and stocking levels, all of which normally have to be updated frequently and can be costly.
● Staff bulletin or newsletter.
● Training courses.
Intranets can be used for much more than publishing information, as shown in Box 3.7 . Web browsers also provide an access platform for business applications which were tradi- tionally accessed using separate software programs. This can help reduce the total cost of ownership (TCO) of delivering and managing information systems. Applications delivered through a web- based intranet or extranet can be cheaper to maintain since no installation is required on the end- user’s PC, upgrades are easier and there are fewer problems with users reconfiguring software. Applications include tools for workgroups to collaborate on projects, self- service human resources (e.g. to book a holiday or arrange a job review), financial mod- elling tools and a vehicle- build tracking system. Traditional information such as competitive intelligence, company news and manufacturing quality statistics can also be shared.
Managing internal digital communications through intranets and extranets
Total cost of ownership (TCO) The sum of all cost elements of managing information systems for end- users, including purchase, support and maintenance.
Box 3.7 Twelve ways to use your intranet to cut your costs
This guidance is from the Intranet Benchmarking Forum (IBF), the world’s leading intranet and portal benchmarking group.
1 Build bridges with internal customers. Intranet initiatives are driven from the business units that will benefit. They say: ‘Where intranets are achieving costsav- ings, the impetus often comes from business units or functions, not the central intranet team. From HR and finance to manufacturing units and customer service operations, it is these business areas that are best placed to identify inefficient pro- cesses and practices in their area, and then approach the intranet team for help.’
2 Research users’ needs. This is, of course, a prerequisite of any success- ful information systems project. The IBF advise: ‘The leaders in the field carry
out research with the aim of building a picture, for each of their main employee groups, of their working patterns, the processes they follow and where the frus- trations, blockages and inefficiencies lie, as well as finding out in detail about how they currently use the intranet and where they think it could help them work more efficiently.’
3 Implement or expand self- service. On re-engineering processes to enable self- service, the IBF say: ‘The most significant way intranets cut costs for organisations is by enabling administrative processes to be reengineered – particularly in the HR area – and migrated online via the intranet. This can make processes far more cost- efficient (and effective) for the organisation and individual users’. They give the example of how the British Airways intranet has achieved some impressive results following its re-launch as a self- service intranet:
● 100 per cent of internal recruitment is now carried out on the intranet
● 100 per cent of staff travel is booked on the intranet
● 33 per cent of staff training is delivered through the intranet
● 80 per cent of employees update their contact details on the intranet
● The most popular self- service application has been the relatively simple e-Pay tool where employees access their payslip. This alone delivered BA savings of
£90,000 per year.
4 Target further design, print and distribution savings. Reduction in physical and distribution costs through moving towards a ‘paperless office’.
5 Improve usability. Making it quicker to find information through improving informa- tion architecture and ‘findability’, i.e. better browsing and searching functionality.
6 Revamp HR content. As indicated by the examples given above, improvements to HR functionality often give the biggest benefits to the employees and the business.
7 Create content for customer- facing staff. The example is given of the UK-based insurance group Prudential which has used its intranet to provide content and tools that help contact centre staff respond rapidly to telephone, email and postal enquiries from customers.
8 Create internal helpdesk content. Costs of internal helpdesks, for example for IT, HR or Finance, can be delivered more efficiently via the intranet. The IBF suggests it costs about £8 to £10 to respond to each request for help by telephone, and about £5 to do so by email.
9 Enhance the employee directory. The IBF say: ‘A good people search can be a killer app: many intranet experts agree that, more than anything else, staff want to use the intranet to get in touch with one another.’
10 Put senior leaders online. Intranets make it easier and more cost- effective for senior leaders to communicate their ideas and ‘walk the virtual floor’ – for example through blogs that allow staff to comment on posts, or through a regular online webcast or chat Q& A sessions.
11 Leverage online meetings. This is web conferencing which, although not directly enabled by the intranet, should facilitate collaboration.
12 Measure savings. The IBF state that: ‘Few organisations have made progress in measuring the cost savings they can attribute to the intranet, or even to parts of it.’
This is partly because it is difficult to measure cause and effect. But the study does give some examples:
● Ford estimates that online training delivered via its portal will drive down training costs to an average of $0.21 per class, down from $ 300– $2,500 per class.
● Cisco cut the cost of processing employee expense reports from $50.69 with the previous forms- based system to $1.90 three years later. Total corporate savings by that third year were $7m. The average elapsed time for processing each expense report dropped from 21 to 4 days.