• Welcome guest (Login)
• Register
• Shopping Basket

Home > Broadband Money Makers > Tools and Applications Grid > Video-based applications

Video on Demand

1 Overview

Video on Demand (VOD) remains a very important differentiator that distinguishes cable and DSL operators from satellite providers. (Satellite is unbale to provide ‘true’ unicast VOD). This profile provides technical details of video-on-demand services, as well as historical information on VOD deployments to 2004. Subsequent developments are covered in the Video content over broadband profile, which covers ‘broadcast’ channels as well as unicast VOD. Nearly every operator includes broadcast channels in their TV services, with the current exception of T-Online in Germany, which provides the T-Online Vision service via a very expensive set-top box. However, this profile provides a useful record of the early stages of current telco TV/IPTV/VOD deployments.

VOD to 2004: a summary

During the 1990s, telcos tried to develop VOD-only services. But these failed to anticipate the growth in multiple digital broadcast channels. The services were also not technically capable of meeting the considerable challenges of delivering tens or hundreds of thousands of simultaneous video streeams. And the network infrastructure was not reliable enough for VOD.

But after the launch of successful broadband networks in the late 1990s, operators were ready to start looking at VOD again. By May 2004, Point Topic estimated that there were at least 230,000 DSL and fibre subscribers worldwide with access to true VOD.

Reasons for the improved outlook included:

• Sharp falls in the initial capex per end-user required to set up VOD service
• Improved DSLAM and network technology allowing more efficient and cost-effective distribution of video content
• Lower backhaul bandwidth costs
• Positive feedback from the users of existing services, who prove to like VOD, are tending to increase their use of it, and are les likely to churn to other service providers when they already have VOD.

VOD over cable raises different issues, both because the cable industry is already built upon video entertainment and because of the different technical characteristics of broadband distribution over cable networks. True VOD or high-performance near-VOD over digital cable is beginning to see a large scale commercial service on a district by district basis. Many cable companies already offer near video on demand, but widespread true VOD will require major network investment. It will therefore happen as an evolutionary rather than revolutionary development of existing services.

2 Key features

2.1 Typical experience

When a TV customer selects the video on demand service from their TV service (usually from an on-screen Electronic Program Guide), the first thing that happens is the On Demand Application (ODA) client program is started. This program is either stored on the hard drive of the set top box (STB), or is downloaded to the STB when the ODA is launched.

The ODA presents the customer with a menu of movies and other content available on-demand (an electronic programme guide or EPG). Customers can use the menu to navigate around this range of content, sorting by title, genre, actor or any other metadata category defined by the VOD service provider.

Each title usually has additional information, for example pricing or images, which the customer can look at before deciding to purchase. When the customer decides to buy, the ODA communicates a session setup request to the VOD system.

VOD providers can regulate access to titles by ratings (U, PG, 18 etc) or by PIN number. Different members of a household can have different PIN numbers allowing them to view content of a specified rating, for example. Access can be denied if the PIN is wrong or does not cover a particular rating.

The session setup request tells the VOD server that a particular video file must be streamed to a specific set top box that is connected to a particular hub, connector and video card in the distribution network (whether cable, Gigabit Ethernet or DSL).

The video server controller uses an algorithm such as dynamic channel allocation (DCA) to create a unique video session, by identifying a complete path through the network, and allocating bandwidth to the session. The controller then assigns an MPEG program number to the video stream, and instructs the video server to start streaming the video.

The video server finds the required file, usually stored on a low-cost high-capacity disk farm. It also sends a session setup confirmation message to the client ODA, which tells the set top box what tuning settings it needs. The streaming can then begin, and the subscriber can watch the video.

During the streaming, the customer can usually pause, rewind or fast forward the video. The ODA communicates a pause request, for example, directly to the video server, which can then change the video stream as required.

The charge for the video is then added to the subscriber’s monthly bill by the ‘back office’ subscriber management system.

2.2 Customer appeal

‘Watch whatever you want – now’

The key selling points of video on demand services are convenience and choice. The viewer is able to access what they want, when they want it, without having to move from their armchair.

In theory, VOD makes it possible to deliver tailored, niche content, for example giving access to an online archive of films and programming. Ultimately, this ‘video dialtone’ would enable the viewer to find quickly any piece of recorded video, TV or film, in any language, on any subject. In practice, this will take years to achieve. Current deployments typically offer several hundred hours of programming or movies per year.

2.3 Variations

Near VOD (NVOD) is a simpler offering, where the subscriber has a more limited choice of programming, and where programme start times are fixed. For example, the top 20 movies might be available starting every 15 minutes throughout the day. This method is used because of limitations in server or network performance in streaming multiple simultaneous streams of the same content. Many cable TV companies that advertise VOD services are in fact using NVOD. This is a natural extension of their established pay TV business. Some people describe NVOD as “just pay TV but with more frequent showings.”

The end-user has no immediate influence on what programmes are shown. NVOD is usually achieved by simply using multiple broadcast-type channels. Channel 1 shows film 1 starting on the hour, channel 2 shows film 1 starting 15 minutes later, channel 3 has a delay of 30 minutes, and so on. In this way, 8 channels are needed to provide a 2 hour film on an NVOD basis and 160 channels would be needed to offer a choice of 20 NVOD films. Capacity of this order is now available on digital satellite and cable networks, although it is questionable whether it is economic to allocate such a high proportion of resources to NVOD.

Although it may seem very similar from the point of view of a user who is satisfied with a limited choice of current movies, NVOD is very different technically and in its potential market positioning. Providing a limited set of video streams which are defined and scheduled in advance is quite different from allowing users to select any one of a large number of videos at a time of their own choosing. And while NVOD can be offered by digital satellite or cable, true VOD is only possible over a service which can provide the end-user with their own dedicated broadband channel, all the way from the video server to the set-top box. This is potentially a unique selling point for broadband in the long term, but it does not have much force at the moment when the infrastructure needed to support it has yet to be rolled out.

Offline video delivery is another alternative version of video-on-demand which does not include the option of viewing programmes immediately, in real-time. Instead, the video requested is delivered as a large stand-alone file, using communications capacity as it becomes available. This is already happening on a large scale as the usually illegal peer-to-peer swapping of video files between individual Internet users. It is also likely to develop as a legal video service, on the lines of the emerging music services, offering access to large libraries of special-interest videos. The very size of a video file - of the order of a gigabyte or more - ensures that this type of application will still require substantial broadband capacity.

Subscription VOD (sVOD) eliminates the pay-per-view aspect of VOD and gives the subscriber unlimited access to a video library for a monthly fee. The title list changes every month.

TV over IP is being deployed by some telcos to provision TV over DSL (or fibre). This technology is generally first used to provide multi-channel TV, similar to a basic cable TV offering. Once this technology is proven, the operator can then deploy a VOD service to complement the TV offering.

PC TV is an alternative way of delivering VOD. The mass market will want to watch VOD programming on their TV sets but there is some demand for viewing via the PC. Combined with offline video delivery, this is likely to be particularly important in driving the availability of VOD for niche special-interest material such as archive films or educational videos. It may also appeal to telcos and other operators who are not allowed to enter the TV market for regulatory reasons.

VOD applications can be seen as a subset of On Demand Applications (ODA), where the viewer sends information back to the central server. Examples of other ODAs include music and computer game download services, education-on-demand, retailing or the type of television and information-on-demand systems used in hotel rooms.

Postal ‘video-on-demand’ services allow users to browse and order videos or DVDs via a website. These are then physically dispatched via the mail, along with a pre-pay return envelope. After viewing, the user posts the DVD back to the rental company. Some services have flexible returns policies that are cheaper than those of a video rental store. For occasional users, this type of service offers some of the benefits of VOD (not having to leave home) without the need to have a broadband-enabled set-top box. It is therefore an important interim technology to consider when developing VOD strategy.

2.4 Leading examples

See section 6 for a listing of more of the main deployments worldwide.

BB Cable TV, Japan

BB Cable TV is the VOD and TV service operated by Broadmedia, a division of the Japanese technology conglomerate Softbank, using the network of Yahoo! Japan, also owned by Softbank. A pilot scheme was launched in March 2003, in Tokyo, aimed at the subscribers to Yahoo’s ADSL Internet access service. Commercial launch was in September 2003, with about 1000 initial subscribers, and the service was available to about 35% of Yahoo! Japan’s homes passed. By May 2004 BB Cable TV was reported to be adding about 10,000 subscribers per month. Current head-end capacity was sufficient for only 120,000 subscribers in total, but it was planned to scale this up to 1 million using server clusters. In mid-2003, Softbank's CEO, Taro Hashimoto, was aiming to get 10% of Yahoo!'s subscriber base signed up to VOD by the end of 2004, driven by a massive marketing campaign. Since then the company has been rather quiet about the progress of its VOD venture.

Yahoo! is an aggressively marketed broadband service that had 3.55 million Internet subscribers at the end of November 2003. Just over 3.3 million of these subscribers also took advantage of Yahoo’s very low cost IP Telephony service, which significantly undercuts the charges of the incumbent operator NTT. Yahoo has deployed a widespread IP network to cover the major cities of Japan, and this enables costs to be minimised compared to a traditional telephone company. Local loop access to each home is provided either over fibre within apartment blocks, or copper local loops rented from NTT.

BB Cable TV offers over 30 broadcast television channels and video-on-demand with over 4000 titles, using parent company Softbank’s IP network with MPEG2 compression.

Pilot viewers on average buy about 2.5 movies or clips per month with an average length per clip of about 80 to 90 minutes. Prices range from $2.50 to $6.50 a clip, depending on the content. The Installation fee is approximately $80, with a monthly fee of $21.

FastWeb, Italy

Available in 6 cities in northern Italy. FastWeb delivers TV and VOD, plus broadband and telephony. By end 2003, 3.1 million homes were passed, and 330,600 residential and business customers were connected. With over 100,000 subscribers the TV and VOD service had achieved more than 30% penetration. Although its initial plans concentrated on a fibre build-out, FastWeb now says that the economics are usually in favour of DSL. It expects to provision only around 25% of its planned 10 million homes coverage by 2010 using fibre. The rest will be DSL. The capex costs are around $700 per subscriber for fibre, versus $500 for DSL.

TV has grown strongly during 2003. The contribution of video to ARPU, averaged over all FastWeb’s subscribers, was 17 Euros per year in the 12 months to Q4 2002. A year later, it was 91 Euros. But not all subscribers take video services. Amongst video customers, ARPU for video services was 341 Euros in the 12 months to December 2003. 209 Euros of this was from Pay TV subscriptions, with 132 Euros from Video on Demand services.

A basic TV subscription is 10 Euros per month for mainly free channels, but there is a menu of extra channels and VOD options. Many subscribers take the TV service as part of a triple bundle of TV, Internet, video conferencing and telephone, which costs 110 Euros per month. Without TV and video communication, the service is 85 Euros/month. New films on the VOD service cost 6 Euros per 24 hours, older movies cost 4 Euros per 24 hours.

Kingston Interactive Television, UK

KIT is a long-running trial and service in Hull, using local city incumbent Kingston Communication’s ADSL network. Originally launched in 1999, it is believed to have had an initial set-up cost of about $2000 per user, and this high cost structure has been a burden on its economics and growth ever since. KIT had about 9,000 subscribers in early 2003, although the number of paying subscribers seems to have declined since then. Its VOD service runs on nCube servers capable of providing 2,500 video streams.

The national broadcaster, the BBC, is using Kingston’s deployment as a testing ground for its interactive services, and has already tested interactive coverage of the football cup final. It has also introduced more locally produced content, using digital technology to reduce production costs.

2.5 Suppliers

A video-on-demand system requires a wide range of hardware, software and network assets. These must all be integrated together extremely reliably to deliver the service levels that customers expect from television and video. Integration is critical in making VOD systems work reliably without lengthy installation times, and explains VOD providers’ strong interest in standards.

An important trend is to reduce the cost per stream for VOD providers. As of mid 2002, an approximate cost per stream of $250 was the US list price from Silicon Graphics for video servers. Transport costs are reducing faster than the server costs, which remain the key expense for provisioning VOD, although using off-the-shelf hardware is beginning to bring costs down.

Leading suppliers of the components for VOD systems are listed below. The requirements for different components are set out in more detail in section 4.1.

Components and suppliers for VOD systems

3 Marketing

3.1 Target markets

VOD over both cable and DSL is still in the early adopter phase, but subscriber numbers are rising, especially for cable. We estimate that the total number of subscribers to all types of TV and video over DSL and fibre broadband services (but not cable) was over 520,000 as of May 2004. However, at least 290,000 of these did not have VOD available as yet. Of the remaining 230,000 with VOD access not all use it, but usage rates and ARPU are usually quoted as an average over the whole subscriber base with access.

Currently VOD operators are targeting heavy users of video rental services for their movies on demand services. These users watch a large amount of TV programming. As TV on demand, rather than films on demand, begins to become available, users of the TiVO personal video recorder will begin to be a target audience. These users are generally more interested in being in control of their viewing, and value the ability to watch programming at their convenience.

3.2 Prices

Pricing for TV and video services is either on a monthly subscription or a one-off payment basis. ‘On-demand’ applications are typically charged on a ‘per-use’ basis. For each program, event or movie that is ordered, the fee is either deducted from a pre-paid or agreed credit account, or added to the monthly customer invoice.

On demand services are usually available only as part of an existing TV package. This includes other content, and packages differ widely between operators. So direct comparisons of the cost per film or program are difficult if the monthly subscription fees are taken into account. But the one-off cost per film is relatively stable, at around $2.50 for older titles, and $4 to $7 for recent releases.

Examples of VOD charges

Typical prices for Kingston Interactive TV packages

3.3 Promotional techniques

Customers seem to love VOD once they have it, and enthuse about it to their friends. (Current VOD users will tend to be self-selecting TV fans, so this is not surprising.) For example, Italian operator FastWeb doubled its VOD revenue during 2002, from around $5/month to approximately $10/month. This would equate to 2 new release or 3-4 library titles rented per month. Rogers Cable in Canada reported an average of 3.4 buys per user per month in January 2003 for its Toronto VOD service.

Churn rates are also greatly improved, down from around 5% without VOD to under 1% with VOD. This means that customer retention costs are reduced.

Pay less to view later, pay more to view now

Pricing tiers will enable operators to manage traffic levels on the network, and can be used to emphasise the convenience of VOD. For example, a new film release could cost $10 for immediate viewing, $7.50 for near-video-on demand viewing or $5 for downloading overnight for watching the next day. As early adopters of advanced television services tend to be short of time, the ability to control when they view programmes is important

‘Up-sell’ to existing DSL subscribers

Existing DSL subscribers already have a relationship with the operator. The DSL line has already been qualified, and they are interested in new technology. They are therefore the first prospects for marketing VOD over DSL.

Develop bundles

Cable companies will be strong competitors in the video-on-demand business. They already have a good understanding of how to bundle together Internet access, telephony and television services as a complete package with discounts. Most already have experience with pay per view , and they are also starting to offer video on demand. Operators entering the video-on-demand market need to understand what subscribers are used to paying for TV and video services, in order to set tariffs at an appropriate level.

Develop innovative services

New services, enabled by set-top and head end technology, are an important part of service differentiation. For example, FastWeb markets the VideoRec service, which allows users to record programmes, with fast forward and rewind functions, without having to own a video recorder. The recording can be set on the TV or from anywhere via the Internet.

Geographic focus

FastWeb in Italy focuses on local city-based campaigns, avoiding the expense of national coverage, and reflecting its presence in key cities. This focus is important. Attempting a simultaneous national rollout of VOD over DSL would be technically very difficult and extremely expensive. Customers have high expectations of video service reliability and ease of use.

Evolution, not revolution

Broadcast TV will remain important for most viewers for the medium term at least. VOD and other interactive services must therefore work together with the TV channels.

3.4 Marketing issues

Dominance of the Hollywood studios

Current business models for VOD depend on the availability of Hollywood blockbusters to drive mass take-up and generate an attractive ARPU. On the other hand, the Hollywood studios are taking a cautious approach towards a new medium which may threaten their control over distribution channels. The studios have already seen the home video channel (videos and DVDs) grow to at least 35% of the $20 billion in annual revenues from movie sales, more than is produced by cinema showings. VOD is a direct alternative to this channel and could damage the studios’ control over their intellectual property by making films available for peer-to-peer file swapping over the Internet.

These concerns, and the way that a small number of big studios dominate the film market, has given the aggregator a special role in developing VOD. Just eight studios (Warner Brothers/New Line, Fox, Sony, Disney/Miramax, Universal, Paramount, MGM/UA, and Dreamworks) account for 90% of Hollywood’s movie revenue. Having attractive VOD content depends on having agreements with as many of them as possible – no one studio produces enough blockbusters in a year to keep VOD customers buying.

But the studios are unwilling even to talk to small operators directly, and they lack the experience needed to negotiate good terms in any case. So the aggregator has emerged as an essential intermediary. The aggregator brings together demand from a large number of end users on the one side, and agreements with multiple studios and content providers on the other.

Even so it has taken aggregators years to achieve a good set of content agreements, and it demands the impressive credentials which come from strong financial backing and the promise of high-volume distribution. Key issues which have to be negotiated include:

• The assurance of quality distribution with strong digital rights management and the best possible protection against piracy. The nightmare for the studios is that their precious products will leak out and become freely available through peer-to-peer distribution over the Internet.
• The revenue share which the aggregator, and the other players downstream of him in the value chain, will be able to keep. At least one aggregator which stuck out for a 50:50 split on Hollywood movies has had to close down its operations. In fact 60:40 in favour of the studio is often what aggregators will have to accept. Better terms, often much better, will be available on less highly-prized content.
• The time window when movies will be available for VOD. The studios manage film distribution with a well-established series of time windows, starting with theatrical (cinema) release when a film is new, then moving through first video and DVD release, to pay TV and finally free-to-air TV channels. It is vital for the success of VOD that a reasonably early window is negotiated but to start with the studios were offering only a fifth window, after free-to-air TV – which can be as much as three years after the film is first released. Aggregators are trying to negotiate that VOD should share the video and DVD window, but often they have to be content with sharing release dates with pay-TV.
• All these factors combine to ensure that the aggregator will play an essential role as far as the studios are concerned. The structure and economics of the emerging VOD industry also suggest that the service providers will also prefer to work through independent aggregators. Although some are trying to be their own aggregators today those who work with specialist aggregators are likely to have a competitive advantage simply because they are able to offer a better range of programming.

A number of entrepreneurs saw the opportunity and set up multimedia and video aggregator companies in the early days of VOD. But companies such as Intertainer in the USA and YesTV in Hong Kong and the UK were ahead of their time and are no longer playing a significant role. On the other hand, players with a base in the cable TV industry, where content acquisition is well understood, are emerging as potentially important. They include Arrivo, part of the UPC group, and ODG (On-Demand Group) which is a partnership between the two big UK cable companies, NTL and Telewest. Anytime, formerly Intertainer’s Asian partner, is aiming to be a major player in the Asia-Pacific region.

Other marketing issues

Quality of service

Evidence from Yahoo in Japan shows that service quality and the range of available programming are the two most important concerns to VOD subscribers. But there can be problems with delivering this.

Theoretically 1.5 Mbps is good enough for VHS-quality video. In practice, this bandwidth can give poor quality results. The quality of service depends on a large number of factors across the network. For video on demand, buffering can be used to prevent interruption in the film, due to jitter or latency in transmission. Buffering is less acceptable in live television, when a delay of even a second or two might be too much.

Range of services

One of the problems with current VOD implementations is that the choice of programming frequently consists mainly of older titles. Content owners are reluctant to cannibalise revenues from video/DVD sales and pay per view channels. One example of the caution of content owners is the PCCW TV over broadband service in Hong Kong. The set top box does not allow home taping of programmes (to watch at a later date), due to content owner demands.

Personal video recorders

In the short term, the personal video-recorder (PVR) competes with part of the market for VOD. Therefore marketing messages for VOD services should emphasise that video recorder functionality comes as standard. PVRs are being deployed by direct broadcast satellite (DBS) operators such as DirecTV and Sky because satellite broadcast architecture cannot support VOD. A PVR with a large hard disk is the satellite company’s way of competing with VOD from cable and DSL operators. But as video server technology develops, the cost of updating PVR hardware to provide comparable functionality could prove too expensive for satellite operators in the longer term.

4 Technical requirements

4.1 Hardware and software

Customer equipment

The aim for all VOD providers must be to keep the CPE as simple to use as possible. A set-top box, with a built in hard-drive and a DSL modem, will be required for DSL systems.

Service provider equipment

VOD services rely on a major investment in a chain of components. The main components are identified below.

VOD system components

4.2 Communications link

The VOD requirement

The key communications requirement for video-on-demand is to provide a dedicated, high-performance, end-to-end communications channel from the video server to the end-user.

The communications channel has to be high performance because it must have the following features to deliver an attractive real-time viewing experience:

• high bandwidth; most VOD implementations are based on delivering at least 1.5 Mbps while 3.8 Mbps may be needed for good picture quality
• continuous bit-rate; the video stream must offer uninterrupted, continuous quality - it must not vary in bandwidth because it is sharing a channel with other users
• low and consistent latency; this is particularly important for the audio component of programming but it may also be important for video, for example in the real-time viewing of a sports event.

Some degree of buffering can be used to compromise on these requirements. They can also be relaxed considerably for offline delivery of video for later viewing. But they will be essential for a VOD service which wants to offer its customers immediate access and premium pay-per-view events.

As noted above, at least 1.5 Mbps is needed for VOD, and this will deliver best-efforts VHS-quality images. For many viewers, now using DVDs and widescreen TVs, and possibly used to high definition TV (HDTV), VHS quality is not good enough.

The development of compression algorithms could ease the pressure on bandwidth. But for the moment, 3.8 Mbps is a widely used compromise transmission speed that gives good picture quality.

An alternative method of providing near video on demand is downloading a film overnight to the set top box hard drive. This method is used by satellite broadcasters, for example, where true VOD is beyond the capabilities of the network. Using the Divx codec, a film would require about 1000-1500 Mb of memory. Cable or DSL VOD providers can also use this method to deliver content during quiet periods on the network. Customers could opt to pay less to view later.

BT's VideoStream wholesale product

BT's VideoStream product provides a good example of how these requirements can be achieved using established technology. VideoStream was developed as a wholesale product, and is used by Video Networks for its HomeChoice service offered in the North London area. It is based on ADSL access lines and ATM routing over BT's SDH (synchronous digital hierarchy) optical fibre core network. Newer solutions are more likely to be implemented with IP rather than ATM but the underlying requirements remain the same.

BT VideoStream's customers are the suppliers of VOD services, such as Video Networks, while the VOD supplier's customers are identified as end-users. The original version of the product required the customer to make a rather large commitment to buying a minimum of 15,000 end-user access connections in each VideoStream footprint area for a minimum of three years. A newer version, VideoStream Plus, launched in 2001, provided a more flexible and cheaper solution, and the price was further reduced in May 2002.

Video over ADSL using the BT Videostream product

The components of a VideoStream service are:

• End-user access lines, providing an ADSL connection at 2.3Mbps downstream
• End-user data paths (EUDPs) each carrying the end-to-end video stream for an individual user over a virtual circuit
• "Virtual paths" which group multiple EUDPs from different local exchanges (central offices) for transport to BT's VideoStream serving centre
• An access link which provides transport to the VideoStream serving centre for the video services provided by the customer.

The figure below shows how these components are priced and calculates a costing for 3,000 end-user accesses over BT VideoStream. This costing provides enough virtual path capacity to allow about 1 in 3 end-users to access the service simultaneously, which is fairly generous and gives a high cost of £42 (about $68) per month. If contention could be raised to 6:1 the cost would be cut to about £27 ($43.50) per month.

Costing for a 3,000 end-user BT VideoStream Plus service

4.3 Service Management

VOD providers have many management requirements. Many VOD systems provide interfaces for management software that run in standard Web browsers. The management of a VOD service is usually split between two servers (or clusters of servers). One keeps the customer records, and holds information on their preferences, and possibly their viewing record. The other looks after system management and configuration, and monitoring video operations in real time. Another aspect of system management is using a mix of VOD or near-VOD to manage customer demand for popular content. Finally, VOD providers need to manage digital content rights, to ensure access charges and copyright fees are paid to the content owners.

Subscriber management functions include:

• Administering customers, for example switching off customers that have not paid, or switching customers’ preferences between different channel packages.
• Monitoring detailed VOD usage history for each customer
• System management functions include:
• Monitoring and updating software stored on the set top box hard drive
• Updating the user interfaces, menus and electronic program guides, usually using a standard e-commerce application using Java or XML
• Providing network management information, for example on network utilisation. Bandwidth allocation is largely automated, and carried out in real time.
• Loading content to video servers, copying (or ‘propagating’) content from one server to others, or removing content from VOD servers
• Providing reports on errors, warnings, or on particular content items, servers or ports

Managing customer demand involves striking a compromise between immediate on-demand response and aggregating demand for popular content.

The opportunity for TV and video over DSL is that there is a point-to-point mapping between each subscriber's set-top box and the corresponding DSLAM port. The problem is that providing this one-to-one connectivity over the core network is very expensive, and would bring network management problems if too many people tried to watch the same content at the same time, each with a separate media stream all the way back to the content originator.

One solution is to use IP multicast, or some variant of the same idea, for popular content, and dedicated connections for more niche programming. IP multicast makes it possible to provide a single media stream across the network for multiple users, breaking it out to serve each individual user over the last few links of the connection. For example, the top ten movies could be provided on an NVOD service using IP multicast to avoid network congestion. Less popular library titles could be offered as a true VOD service or with delayed download to use off-peak resources.

Digital rights management also has to be recognised as a dominant issue for any kind of content distribution service. Hollywood movies are not the only type of content that customers order from video-on-demand systems. In fact, the evidence from UK implementations is that childrens' programming and classic television comedy is more important. However, most service providers feel that it is important to have Hollywood movies on the menu. Notoriously conservative, studios are worried about:

• piracy - hackers stealing films, compressing the files, and posting them on file-swapping sites
• release dates – studios do not want to jeopardise VHS and DVD rental income, and are currently making VOD providers wait until the titles are due for release via conventional cable TV channels. VOD operators feel that this makes them uncompetitive with video rental shops
• fee to the studio – studios are worried about paying too much money to third and fourth party ‘middlemen’. A cut of 60:40 or even 65:35 in the studio’s favour may be the sort of deal that studios will insist on
• working with rivals – subscribers don’t want to watch only movies by one studio, but this is frequently what’s offered at the moment.

Studios are also worried about the emergence of a VOD equivalent of Blockbuster, which has huge power in the rental VHS and DVD market, much resented by the studios. Kingston re-branded its VOD service as Blockbuster in 2003, and FastWeb also has a Blockbuster link, indicating that Blockbuster continues to explore this market.

4.4 Billing

Billing for premium video-on-demand services tends to be on a per use basis. However, some services (eg Telekom Austria, see Section 6) with a more limited range of content offer the option of an ‘all you can eat’ monthly subscription.

Customers can set a credit limit (for example, to stop the children ordering too many movies every month). VOD servers have a range of sophisticated billing management options from which VOD operators can choose. For example, the server can maintain two-way communication with the set-top box to perform functions such as STB authentication or maximum credit set or check.

5 Business model

5.1 Value chain

The VOD value chain is highly complicated. The business of generating and delivering content to the end-user involves a whole series of players, with the main positions listed below. The content delivery chain rides on an infrastructure which brings together the contributions of many other players, from software suppliers through hardware vendors to network operators.

Video-on-demand content delivery value chain

Some players choose to occupy multiple positions in the value chain. FastWeb in particular does its own aggregation, sells the service to end users, and also owns and operates the network used for delivery.

The sharing of revenue across the value chain is still contentious. One split sometimes quoted is 50:25:25 for content providers, aggregators and video service providers respectively. But the actual split will vary greatly from country to country and depending on the service provider’s business mix.

Content providers will often require 60% or even more for the most valuable material, such as blockbuster movies. On the other hand, sales of older or less popular material will help to bring the average share of royalties down. A service provider getting 10% of its revenue from set-top box rental, 50% from broadcast and cable-TV type channels and 40% from VOD might aim to get royalties down to 33% of sales. The aggregator might get 25% of VOD sales but only 10% on broadcast and pay-per view channels. This would give a split much more favourable to the service provider at 33:15:52. This kind of share-out will be more or less essential if the service provider has to cover the cost of the broadband connection. A pure VOD service provided as an optional add-on for existing broadband users could be viable with a less favourable split from the service provider’s point of view.

5.2 Business case

Up until 2004, at least, the business case for VOD services over DSL has been difficult to make. The basic problem is simple:

• The initial capital cost per end-user is high
• Operating costs are high
• It is hard to assemble a sufficiently attractive catalogue of video content
• The revenue per end-user is relatively low.

The first generation of VOD services failed to overcome this barrier. Most if not all of them have lost money heavily and as a result they have either closed down or are shrinking rather than growing their user base.

Now the situation is changing. Initial capital costs have fallen sharply. Content providers are becoming more willing to make their products available. The most successful pioneers are achieving good and increasing ARPUs. It looks as if video-on-demand may be starting on a stable growth path at last.

Capital costs

The pioneers of VOD, such as Kingston Communications, are believed to have had initial capex per user as high as $2000. By 2002, capital cost estimates for setting up VOD services from vendors such as Net2Net Technologies were still as high as $1100 per user. Now typical estimates range from $500 to $750 per user.

The make-up of Net2Net’s $1100 estimate is shown below.

VOD capital costs (USD)

Now the component costs of this estimate are coming down. Set-top boxes in particular are down to about $180 per user. Much of the other expenditure can be charged against the cost of providing basic broadband services rather than being required specifically for VOD. Softbank has estimated that the marginal additional cost of providing VOD is only $300 per user.

Actual costs also vary widely depending on circumstances. Every operator has a different network, and network upgrade costs will be lower for a high-density modern copper network in Seoul than for an older, less dense network in Europe or North America. Many variables could alter the business case. The major ones are summarised below.

Variables affecting per user capital costs of video on demand

Operating costs

VOD operating costs can be divided into capital repayment, content royalties and general operating expenses, which include marketing and customer acquisition, maintenance and customer support, and managing and developing the VOD offer. A profit margin needs to be added as well. Based on initial capex of $1100 per user costs can easily reach over $70 per month, as shown below.

VOD operating costs (USD)

A set up cost of $1100 per customer requires around $26 per month over 5 years to pay back, when interest charges are taken into account. A set-up cost of only $300 would reduce this to about $7 per month.

Content costs depend on many variables, as discussed above. Obviously the total also depends on how much content the end-user buys or subscribes to. Here we have assumed an average 33% content cost on total consumption of $90 per month.

Operating expenses are also highly variable. Experience suggests that $20 per month is a reasonable target.

Adding a reasonable profit margin of $14 per month gives a total of $90 per month as an average revenue target for the service.

Revenues

Basic television service (the channels traditionally supplied by broadcast TV) is generally free to end-users, and providing this service over DSL will not generate significant revenue. That leaves premium content to generate the revenue to pay back the investment. A combination of pay per view, home movie rental, near video on demand and video on demand will contribute.

Softbank in Japan charges 2,500 Yen for a basic 14 channel service (about US$21). Add in another $20 for video-on-demand, and $10 for other services gives a potential revenue of $51 per month

FastWeb’s video subscribers were generating an ARPU of 341 Euros in the year ending December 2003, or 28.4 Euros ($34) per month. Approximately 60% of this revenue comes from pay TV services, 40% from video on demand.

Estimated average potential VOD revenues/month (USD)

Conclusions

The revenues being achieved by established VOD services are well short of the $90 per month which the older VOD services would need to recoup their investment and make a worthwhile profit. Add on sales tax, and the cost to residential end-users to achieve the target revenues would be over $100 per month in most countries. This is generally agreed to be well above the price level needed to reach a mass market for VOD. It is not surprising that most first-generation VOD services remain locked in financial problems.

On the other hand, simple estimates suggest that the cost of providing VOD is now moving down into the area where it can be covered by achievable ARPU levels. If the marginal cost of providing VOD can be brought down to $300 per user as Softbank suggests, the targets for content revenue and operating costs can be scaled down as well. This suggests a profitable business case could be based on additional VOD ARPU of about $36, plus sales taxes. This is still quite high, but it is in the region being achieved by FastWeb in Q1 2004, and targeted by Softbank.

On the other hand, it is based on the marginal extra cost of providing VOD over an existing broadband connection, not on the total cost. This suggests that the service would still be uneconomic for end-users who wanted broadband only for access to VOD. Further cost reductions, and ARPU increases, will be needed before broadband can be cost justified purely for video services.

Even if VOD profitability is small considering the resources required, it may still be valuable to include the service as part of a broadband bundle. For example TransACT, which offers VOD with other video services over a fibre network serving Canberra, the capital of Australia, reports that take-up of its service is twice as high where VOD and video services are included in the bundle and that it is experiencing churn of only 2% per annum. (For many broadband providers, churn is closer to 2% per month.) Thus a package including VOD may be more profitable overall for the service provider even if the VOD element is less profitable than other services in itself.

VOD over cable

VOD over cable offers a more immediately attractive business case than VOD over DSL. Costs per subscriber are still high, and the case depends on having a digital cable infrastructure in place. For example, just over a quarter of the 72 million cable subscribers in the USA have digital cable, and about a quarter of these subscribe to VOD.

The cable business case looks more attractive, because the set top box and local loop investment has largely already been made to enable existing cable TV services. Comcast’s existing STBs were updated online with new software to make them VOD compatible. Cable customers are also already proven TV subscribers, and marketing and customer acquisition costs should be lower for cable companies, compared to DSL operators. But the companies will still need to make significant headend investments in video servers and the associated software and systems integration. Video servers alone currently cost in the region of $250 per stream, and that does not include software and core network costs.

Core network upgrades are necessary for the future when a greater proportion of subscribers want VOD services than is the case today. Upgrading core networks to IP over Gigabit Ethernet, with MPEG taking over at the edge is a common upgrade path.

5.3 Entry opportunities

The complexity and potential size of the VOD market creates a large number of potential entry opportunities.

Incumbent telcos may increasingly find themselves obliged to offer VOD, generally combined with IPTV or TV over DSL, both to retain existing customers and to extend the appeal of broadband to households which do not have a PC.

Competing telcos can also offer VOD whether by reselling a wholesale offer, by providing their own technology using unbundled local loops, or by building their own access network. The value is likely to be more in acquiring and retaining customers for a profitable bundle of services than in the profitability of the VOD service on its own.

Aggregators are clearly going to be key players in the VOD business. New entrants will need to be well funded, with in-depth experience of the market and the ability to achieve large volume distribution. But in a rapidly developing business there are still opportunities, for example for telcos to team up to create a joint aggregator.

Content providers have the biggest and most varied opportunities as a result of VOD. There is a clear opportunity for the big players, such as the film studios and broadcasting networks. Now telcos, aggregators and ISPs are starting to offer content distribution and digital rights management services which make it easy for even small content providers to offer material over the Internet. While VOD is only a substitute for existing distribution channels for big content providers it offers a completely new business model for small ones.

Software for VOD includes middleware as well as management and billing packages. This is a classic area for entry by smaller companies at present, but the barriers to entry will rise rapidly as the experience and industry knowledge which needs to be built into these products increases.

Hardware products are, if anything, even more varied and more widely spread through the value chain than software. Set-top boxes are already known as a volatile business. From codecs to video servers, products will range in price from cents to millions of dollars. With standardisation and the development of mass services the main products are likely to become commoditised so that only the biggest and most efficient manufacturers will survive. But there will still be many niches which can be profitably served by smaller specialists.

5.4 Future development

In the medium to long term, VOD will move from a selection of movies on demand (with a choice of perhaps several hundred hours per month) to a service that is more like the TV on demand offered by personal video recorders. This would involve video servers capable of holding thousands of hours of programming, large chunks of which could be changed every day. Implementing systems with this power will depend on the cost of servers reducing significantly. But it will also depend on an industry standard for intellectual property rights management and content protection emerging. This will then enable a profitable and secure business model to establish itself. This will in turn encourage content owners to allow more material to be shown on demand. Only then will true ‘video dialtone’ be possible, although it may happen first for limited audience TV on PC, before mass-market TV.

6 Service listing

The following listing includes many of the TV and VOD over cable services that are currently in customer trials or operating commercial services. We estimate that by May 2004 there were over 520,000 active users of these services in total, about 390,000 of them with just two operators, PCCW (Hong Kong) and FastWeb (Italy), and another 130,000 shared between about ten other services.

These services all offer a mix of free-to-air broadcast channels, subscription channels and pay-per-view or NVOD. Most also offer true VOD, but many VOD services were still new at May 2004. Even where VOD is well established many users of real-time channels do not take advantage of it. The two main operators listed which did not offer VOD were PCCW and Qwest, with a total of 290,000 users. This suggests that worldwide there were at least 230,000 consumer end-users with access to VOD as of May 2004.

Video on Demand services

6.2 Sources

Business case sources

Orca Interactive ppt presentation, featuring Chunghwa case study, Telco TV conf, Barcelona, May 2003

NDS ppt presentation (Nsmith2), TelcoTV, Barcelona, May 03]

Ovum eBiscom profile, Access@Ovum service

Rogers Cable Q4 02 Analyst presentation]

Net2Net Technologies white paper

• BMM Service Homepage
• Tools & Applications
• Video on Demand
• Voice over IP