Data-Driven Development

This section looks at how access to the internet and its content is priced and which parties earn the revenue. It examines the different ways for earning revenues from network access and content, including subscriptions (postpaid and prepaid), advertising, and transaction fees. It looks at the consequences of data-driven business models for the traditional pricing approach used by telecommunication operators, with potential impacts on network investment, market concentration, and net neutrality.

In reviewing alternative pricing structures, the difference between access to the internet and to its content varies in their financial significance. Users typically pay for access to the internet based on the volume of data they use, whereas content is not generally priced by volume. While a user may view free video streaming services that generate large amounts of traffic, an online shopping transaction generates little data traffic but, in aggregate, creates significant value for the seller. This contradiction poses a significant challenge for measuring the economics of data:

Consumer pricing at telecom network operators has evolved with technological change, from price per minute, to price per speed, to price per data consumed. Before the internet, voice was king and operators generally charged for it on a per-minute, metered basis. Before broadband, consumers mainly accessed the internet using dial-up connections with pricing as if it was a voice call (that is, minutes of use). The emergence of fixed broadband led to pricing by speed. With mobile broadband, it is harder to guarantee speeds because of different coverage ranges and a variety of handsets. Three main models exist for pricing data for mobile networks. One is flat rate, in which data usage is not metered but there is a fine-print data cap that, if surpassed, results in the speed being lowered. The second is postpaid, in which a certain amount of data is included with a monthly subscription.

The third, and most predominant, particularly in developing countries, is prepaid, in which the price is tied to a fixed amount of data. Prepaid must generally be used within a specific time, but does allow for flexibility in that a small amount of data can be purchased for a day or a week.

To justify investment in higher-bandwidth infrastructure, a number of telecom operators around the world have diversified into providing video services, enabling them to provide so-called triple-play offers (such as voice, data, and video). This has led to bundled offers in which all three services are offered together at a price higher than that for buying any single service alone, but at a discount compared with purchasing each service separately. Ironically, although traditional broadcast operators moved to protect their markets through legal challenges, neither they nor the telecom operators foresaw the greater threat looming from streaming video services.

The challenge today is that data traffic is exploding while the traditional revenue earner for telecom operators, voice and text, is declining as users shift to over-the-top (OTT) services delivered over the internet. Data traffic from smartphones surpassed voice traffic in 2011 and since then has grown at a tremendous rate, accounting for 96 percent of mobile traffic (figure 2.7, panel a). However, operator revenues have not kept pace. In 2015, voice still accounted for almost half (46 percent) of telecom operator revenue (figure 2.7, panel b), down sharply from two-thirds of operator revenue in 2010. Total telecom revenue has only been growing at 2 percent a year.

Figure 2.7Global network traffic and retail telecom revenue, selected countries

How to pay: Advertising-funded versus subscriber-funded networks

The internet has largely been characterized by "free" content. Although some sites, mainly news and streaming audio and video, charge subscriptions, the most popular websites are free. Most of these sites earn revenues from advertising through a two-sided market strategy of providing users with incentives to join their platform. As the internet audience has grown, firms are increasingly flocking to place ads on the web and smartphone apps.

The growth of digital advertising is astonishing. Global internet advertising accounted for more than a third of advertising expenditure in 2016, slightly behind television (figure 2.8, panel a).

Digital advertising spending is highly concentrated, with the vast majority going to two companies: Google and Facebook, who between them received US$106 billion in internet ad revenue in 2016, 64 percent of the global total (figure 2.8, panel b). Their share is growing, up 20 percentage points since 2010. They both have huge user bases, with each reporting more than a billion users of their services, which is attractive to advertisers. But reasons for placing ads on them differ, partly explaining why advertisers often put ads on both. According to one digital ad analyst, "Facebook believes the most important thing is identity in ensuring ad effectiveness . . . they know who you are and so much about you" whereas "Google believes identity is secondary to intent. What's important is what you want right now because advertising products and services fulfils a want or need".

Figure 2.8 Global advertising revenue

Digital advertising is causing a huge transfer of wealth from traditional advertising outlets (such as newspapers and radio) to internet companies. Telecom network operators have also not largely benefited from the value advertisers place on internet content. However, in a highly contested decision, the U.S. government decided in 2017 that ISPs can sell browsing histories to advertisers without the user's consent.

The concentration of advertising spending reinforces the large properties at the expense of the millions of smaller ones. It also threatens infrastructure investment, since the large websites sit on top of telecommunication networks, yet the networks do not necessarily receive advertising revenues. Although there are payments from content businesses to carriers for bandwidth, it is not clear that they make up for the huge amount of traffic generated. At the same time, if not transparent, the payments between content providers and telecommunication carriers pose net neutrality concerns, since a payment may imply an enhanced traffic service, to the detriment of other content providers.

Another concern is that as a few global sites thrive they are branching into areas they have no expertise in and for which their automated content controls pose challenges. One area is news, with traditional news outlets hit hard by a rapid decline in advertising. The rise of news on IT content sites is of particular concern, with objectivity and facts increasingly called into doubt given the explosion of sources. At the same time, legitimate news and information is sometimes blocked, illustrating the weaknesses of robotic software agents trying to determine what is appropriate. For example, Facebook blocked a 1972 Pulitzer Prize winning photo of a Vietnamese girl over concerns about nudity. The company was accused of abusing its power and the photo was later reinstated. Google has been under fire for placing ads next to extremist content on its YouTube video-sharing site. These examples have led to a growing argument that IT firms posting news stories should be subject to regulations similar to those that media firms face.

Advertising does support free services for users, which is particularly attractive for developing countries. These services include email, office applications, storage, and maps delivered over the cloud. The availability of free and legal cloud-based applications and services is putting a dent in software piracy, which has been dropping, and the focus has shifted from loss of revenues to the cybersecurity dangers of unlicensed software.

Besides advertising, some internet companies charge subscription fees, particularly those involved in content. One example is Netflix, which offers streaming television and film delivered over the internet. E-commerce sites such as Amazon, eBay, Rakuten, and Alibaba earn revenue in two ways. One is as a normal retailer, charging a markup over price. The second is when the platform is provided to third-party sellers, in which case the platform owner earns transaction fees.

Table 2.3 contrasts average revenue per user per month for the various internet payment models and for large internet-based companies. Prices have also been converted to purchasing power parity to adjust for differences in the cost of living. Subscription-based video viewing has the highest average revenue per user, and advertising on social media the lowest (but more subscribers). The telecom carrier in the list has the second highest average revenue per user (purchasing power parity).

Table 2.3 Average revenue per user from internet data, 2016

		Average revenue per user (month)
Company (Country)	Revenue model	US dollars	At purchasing power parity	Users (millions)	Note
Facebook (United States)	Advertising	2.63	2.63	1,860	World's largest social network platform
Alibaba (China)	Retail margin/ transaction fees	2.37	4.47	423	World's largest business-to-consumer retailer (gross merchandise value)
China Mobile (China)	Subscription and usage	4.31	8.14	849	World's largest mobile operator
Netflix (United States)	Subscription	8.61	8.61	94	World's largest paid video streaming service

Some content companies are becoming involved in developing networks, feeling that infrastructure development is not keeping up with the vast growth in data traffic. A desire to capture more users from developing countries by making it easier for them to get online is also driving this effort.

Facebook and Google are investing in a variety of communication ventures.

At Google, this includes providing fiber internet access in several U.S. cities, offering fiber backbones and Wi-Fi in Ghana and Uganda, using hot air balloons to extend internet connectivity, and supporting the use of white spaces for making more spectrum available. It also controls Android, the leading mobile phone operating system.

Facebook has been looking at satellites, drones, and solar-powered airplanes for extending internet access, and is getting involved in networking gear. The growing involvement in data transmission by large content firms raises questions about the separation of carriage and content, with the possibility of a few companies dominating both internet content and access.

The rise of "over-the-top" service providers

OTT refers to data services provided over telecommunication networks. The impact of OTT on telecommunication operators comes from either competing with traditional revenue sources, such as voice calls and messaging, or depositing a lot of traffic on their networks. A lack of clear metrics makes understanding the OTT market a challenge. Some of the most popular OTT services have been purchased by larger companies that do not provide disaggregated financial or operating indicators. Nevertheless, considerable circumstantial evidence suggests the impact is significant.

Notable OTT providers include the following:

• WhatsApp, purchased by Facebook in 2014, offers messaging and calling services and claims to have about a billion users in more than 180 countries.

• Viber offers calling, video, and messaging services to more than 800 million people and is owned by Japanese e-commerce company Rakuten.

• Skype, owned by Microsoft, offers calling, video, and messaging services.

• Netflix provides paid video to 94 million subscribers around the world. It competes with telecommunication service providers that also offer video (such as IPTV). However, the bigger impact on the telecoms is the volume of traffic Netflix generates.

The impact of OTT has been particularly strong in what has long been a traditional profit center for telecommunication carriers: international voice telephone calls. By 2012, Skype was already handling a third of international telephone traffic. By 2016, OTT traffic using voice over internet protocol exceeded international traffic provided by telecommunication operators (figure 2.9).

Figure 2.9 International carrier and over-the-top traffic (billions of minutes)

Similarly, telecom carriers have seen a sharp fall in conventional messaging services (such as SMS) used on their networks. In 2015, an analysis of 17 countries found that the number of messages declined in 14 of them (figure 2.10).

Figure 2.10 Mobile messages, year-over-year change (percent), 2014-15

The danger is that if carriers cannot offset the loss of revenues, less money may be available for future infrastructure investments to handle the rapid increases in data traffic. The ITU has created a study group to examine this issue. The West African telecommunication operator Sonatel paints a dire picture, estimating that between 2016 and 2020, its losses from OTT in the international segment will be CFA francs 256 billion (US$432 million) in Senegal, CFAF 164 billion (US$277 million) in Mali, CFAF 79 billion (US$132 million) in Guinea, and CFAF 12 billion (US$20 million) in Guinea Bissau. It also estimates that taxes paid to the government and dividends for its shareholders will fall CFAF 243 billion (US$410 million).

Telecom operators have developed several responses to OTT. They have argued for regulating OTT providers that provide voice and text services in the same way that telecom operators are regulated. Some operators are developing their own OTT products. Others are including large bundles of their own offerings, such as free calls or text in packages. Many are diversifying into opportunities in areas such as cloud computing, IoT, and mobile money. Some operators are trying to do all of the above. Also relevant is taxation for OTT firms without a physical presence in the country in which they are providing service. Though they may offer the service for "free," digital advertising revenues often subsidize this. The lack of OTT taxation in most developing countries gives them a cost-structure advantage over domestic telecommunication operators.

Some OTT services compete with traditional telecommunication services such as voice and text. However, others provide a different challenge, particularly OTT video providers. These include those that provide television and films through subscription services such as Netflix as well as free services such as YouTube and Facebook, in which video posts are increasing rapidly. They not only compete with telecom operators that provide video services, but are also responsible for a substantial portion of traffic going over the networks. Netflix, YouTube, and Facebook are among the top traffic applications in most regions, comprising 42 percent of fixed-access and 34 percent of mobile-access, peak-period traffic (table 2.4). When all of the properties of Facebook (that is, Instagram, WhatsApp, and so on) and of Google (YouTube, Google Cloud, Google Market, and so on) are considered, they account for an even larger share. The two account for more than 60 percent of total traffic on Latin American mobile networks, for instance.

Table 2.4 Percentage of aggregate peak period traffic by region, 2015-16

	Africa	Asia and Pacific	Europe	Latin America	Middle East	North America	Simple average
Fixed access
YouTube	16	27	21	30	NA	15	22
Netflix			4	6		34	15
Facebook	9	3	7	6	NA	3	6
Other (top 10)	44	45	38	41	NA	22	38
Other	31	24	29	17	NA	27	26
Top 10	69	76	71	83	NA	73	74
Mobile access
YouTube	10	17	20	20	21	19	18
Netflix				2		4	3
Facebook	6	8	16	26	10	16	14
Other (top 10)	49	36	38	34	43	35	39
Other	35	39	26	19	26	26	29
Top 10	65	61	74	81	74	74	71

The promise and perils of zero-rated services

Many internet applications are based on the so-called freemium model, in which consumers get basic features at no cost and can access premium functionality for a subscription fee. However, users must still pay for the data consumed using these applications. Zero-rated services provide access to certain content without it applying to a user's data cap. Some firms with desirable content, such as Facebook, have worked with operators, mainly in developing countries, to provide access to their services without its affecting a user's data allowance. At the same time, some operators are striking agreements to provide access to some services free to their customers. For example, T-Mobile in the United States does not charge data usage for customers on certain subscription bundles when they stream music. Other operators provide discounted access to a bundle of social networking services, but this is technically not zero rated, since users still pay for data access but at a discounted rate.

It is argued that zero-rated services provide a taste of a slimmed-down version of a service and users will eventually pay for access to the full internet. Text-only versions may also be relevant for users who do not have access to high-speed mobile internet and rely on slow 2G connections. One example is Facebook Zero, launched in 2010, providing access to a text-only version of the service. It has now been renamed the Free Basics service, available in more than 60 economies, with about 40 million users. In addition to Facebook, access is provided to other websites, such as Wikipedia, Accuweather, and Bing, as well as to local social impact sites for health and employment.

A variation on zero-rated services is sponsored data, in which companies pay for data usage if a user agrees to receive ads. Sponsored data is also used for companies to pay for their employees' mobile data usage for work. AT&T, a U.S. mobile operator, offers sponsored data in which users are not charged for usage if they access a sponsor's site. Syntonic, one of the AT&T sponsors, notes: "Millions of prepaid consumers ration their data, impeding discovery and exploration of mobile apps and content" and is expanding its product reach outside the United States to southeast Asia, India, and Mexico.

While providing free content seems commendable, only making certain content available runs contrary to the net neutrality principle of the internet. In February 2016, the Telecom Regulatory Authority of India issued a regulation prohibiting the use of what it called discriminatory tariffs for data services. The authority formed its decision based on "the principles of Net Neutrality seeking to ensure that consumers get unhindered and nondiscriminatory access to the internet. These Regulations intend to make data tariffs for access to the internet to be content agnostic". Several other countries have also banned zero-rated services. On the other hand, the United States Federal Communications Commission struck down net neutrality provisions in December 2017. The 3–2 vote by commissioners was along political lines in an allegedly "contentious and messy" public comment process. ISPs in that country are now allowed to block, throttle, and prioritize content. About 20 states have filed lawsuits against the ruling, and the United States Congress is considering overturning the ruling if it can muster the votes.

At the same time, it is argued that zero-rated services give an advantage to large companies to the detriment of new startups. As one report notes: "Ironically, if zero-rated services were available when large internet companies were startups, it is unlikely they would have scaled to the size they are now".

Importantly, cost is not the only or even, in many countries, the main barrier to internet access. Users often cite reasons such as no need or lack of skills as the reason they do not use the internet. In Thailand, 97 percent of those who do not use the internet said the main reason was because they did not know how to use it or it was unnecessary or a waste of time. In Brazil, 70 percent of those not using the internet cited a lack of interest as a reason. So it is not clear to what extent zero-rated services get new users online. As one report puts it: "Even with a zero-rated service, the user must still have a device and an active account with the operator that offers the zero-rated service. This raises the question of whether zero-rated services can bring people online who had not previously used the internet". The report looked at the impact of mobile data apps across eight developing countries, finding that 88 percent of users had already accessed the internet before using a zero-rated plan. This suggests that digital literacy challenges are arguably more important than affordability to get more people online.