Understanding data economics:

Understanding data economics:

The Delta Perspective

The top-line impact of OTTs

Authors Juan Jose Rio - Partner - jri@deltapartnersgroup.com
Saahil Malik - Senior Associate - saa@deltapartnersgroup.com


Traditional telco revenues from voice and SMS are being eroded by over-thetop (OTT) applications such as Viber, WhatsApp and KakaoTalk. Delta Partners estimates that this substitution could cost mobile operators as much as US$40 billion – nearly 4% of total industry earnings - in lost revenues by 2014. Although this erosion is partially offset by data ARPU uplift and churn reduction, the long term top-line impact of OTTs cannot be negated by just these. Operators can, however, minimise this impact by assessing:
  • Where they are on the substitution curve (a measure of smartphone and mVoIP penetration)
  • When the OTT impact will become material for them, and
  • How they can address these to drive higher data margins and sustainable
  • ROIC 
  • Our experience suggests these aims are best realised through:
Evolution of commercial management (e.g. re-evaluation of pricing and subsidies)
2. Focus on differentiation and adjacent businesses (e.g. advertising and ICT services)
3. Mediation and enablement of third-party services (e.g. OTT-operator billing partnerships through APIs and differentiated OTT rates)
4. Shift towards cost management (e.g. managing network capex / opex)


The future growth of mobile data users and their willingness to pay for data services continues to remain promising. Global fixed and mobile data subscribers are expected to grow from 2.5 billion in 2012 to 4.5 billion in 2016, with mobile subscribers constituting 75% of these. Although there are some doubts on the contribution of mobile to total data growth – especially in countries with advanced fixed infrastructure – key drivers are likely to continue to fuel the growth in global, mobile-data demand:
1. Data-enabled devices:  Global smartphone sales are expected to almost double from 657 million in 2012 to 1,248 million in 20151 while tablet sales are expected to grow from 122 million in 2012 to over 275 million in 20152.
2. Traffic growth per device: Monthly mobile data traffic per tablet is expected to surpass 4 GB per month in 2016, an eight-fold increase in four years, while smartphone traffic is expected to deliver a 17-fold increase to 2.6 GB per month^3.
3. Data pricing: Operators are trying to retreat from flat-fee pricing structures that strongly incentivise traffic growth. However, not all operators have been successful in doing so, either due to competition (e.g. T-Mobile US) or regulatory requirements (e.g. Hong Kong-based SmarTone)^4.
4. Applications: The total number of apps downloaded has increased more than 10 times in three years to 85 billion in 2012. In particular, mobile video apps (e.g. YouTube) contribute significantly to traffic growth and are forecasted to account for over 70% of mobile data traffic by 2016^3.
5. Network speed: HSPA+ and LTE network upgrades are expected to increase the mobile connection speed six-fold from 504 Kbps in 2012 to 2.9 Mbps on average in 2016^3.
This is the second publication in our Mobile Data Economics series (after “Navigating the Challenges of Data Monetisation”) and dives into the OTT conundrum facing operators by quantifying the trade-off between ARPU uplift from data usage versus mobile voice and SMS substitution.

ARPU uplift

Oft-cited metrics on the positive impact of smartphones may in reality present an inaccurate view. Although some industry and financial analysts estimate ARPU uplift to be as high as 100%5, most of these forecasts appear to be based on a comparison of the average feature phone (non-smartphone) and smartphone ARPUs.
On the contrary, although smartphones will most likely stimulate data usage, ARPUs may actually decline over time given the volume discount afforded to this greater consumption. For example, KT’s smartphone ARPU (and accordingly smartphone ARPU uplift) decreased from US$48 in 2010 to US$35 in 2012 despite a corresponding increase in data usage from ~200 MB to ~1 GB per month^6.
Rather than comparing the average feature phone with smartphone ARPUs, a more accurate approach considers the ARPU uplift resulting from the migration of a feature phone customer to a 2G, 3G or LTE smartphone, or from a low-end smartphone to a higher-end smartphone^7.
Forecasts of smartphone ARPU uplift need to take into account consumer-specific factors, which may include device price and technology. In China for example, ARPU uplift was 25% on lower-end handsets versus 36% on higher-end handsets and ~13-26% for 2G smartphones versus ~22-32% for 3G^8.

However, operator-specific drivers such as voice and data package pricing and related usage limits are also important. Singapore, for example, saw limited ARPU uplift among higher-end, heavier-usage
customers, due in part to the higher ARPU from traditional voice and SMS services, limited pricing options for voice and voice-and-data bundles and a potentially excessive data allowance in the packages^9.
As seen in Exhibit 1, our estimates for the ARPU uplift from feature phones to smartphones are significantly lower than estimates based on average ARPU comparisons. Moreover, we estimate that the revenue uplift from feature phones to smartphones is likely to further decline over time. Accordingly, our experience suggests operators are better served enabling the migration earlier although not converting the entire feature phone base by, for example, managing handset subsidies.


As seen in Exhibit 2, traditional telco services of voice and SMS can be substituted by various data applications, including VoIP (particularly international calls), IP messaging (or SMSoIP) and social media. However, substitution of SMS by data has the most visible effect – a higher impact on margins, and is more relevant in several high-growth (and SMS-heavy) markets. By 2014, the worldwide effect of SMS substitution on operators’ revenues could be as high as US$21 billion.
For IP messaging, the two main two drivers of substitution are not only a significantly lower rate per message (free messaging yet some data access fees), but also a better user experience. While often overlooked, IP-based messaging services improve user experience across a number of relevant dimensions:
  • The status of the message can easily be checked (sent/delivered/read)
  • Both communicating parties can  see each other’s availability (online/offline/writing a message/other user-defined status)
  • The maximum length of an IP based message goes far beyond the maximum SMS length
  • Self-explanatory menus ease the  process of sharing media files or creating group chats
  • Services such as emoticons, images and voice files provide for a richer communications experience, adding to their popularity

Operator responses to IP messaging have varied from initially blocking access to charging for such applications10, and more recently to developing their own competing services. Rich Communication Services (RCS) is one such play by the operator community to compete against OTT services through the creation of a consumer-facing brand, joyn^11.
However, the challenge that operators will face is not necessarily support from other operators, but instead consumer commitment – particularly given the proliferation and pervasiveness of existing messaging services. A value proposition that combines a partner ecosystem (e.g. device manufacturers), global interoperability, ease of use, security and QoS assurance will be critical for driving penetration.
Besides the drivers of substitution, we observe two key accelerators of the level of substitution: the square and community effect and platform-neutral applications.
1. Square and community effect
The bigger the network, the bigger the benefit for the individual user. If 1% of the subscriber base makes use of a certain IP-based messaging application (e.g. BBM), the expected maximum share of SMS substituted by IP messaging equals 1% x 1% = 0.01%. If the penetration of the application reaches 50%, that share increases to 50% x 50% = 25%. Hence the marginal substitution rate accelerates with increasing IP messaging adoption.


Furthermore, the community effect drives substitution through mass adoption of IP messaging and VoIP applications within communities (or calling circles). Even if the total user penetration of an IP messaging or VoIP application in a country may be low, it can be high within certain groups of people, for example, students and clusters of professionals. As individuals within these communities primarily communicate with fellow members of the community, substitution intensity can go far beyond the levels forecasted by the overall square effect.

2. Platform-neutral applications
The substitution rate is increased further with applications that can be used across types of operating0020systems and/or devices. Platform-neutral applications such as WhatsApp and Facebook can be used across operating systems and devices, which adds to their popularity and simultaneously fosters substitution.

Conversely, the BlackBerry Messenger application enjoys a strong following among BlackBerry owners even though, historically, one could only reach other BlackBerry users through the application. However, to avoid BlackBerry users switching to platform-neutral applications or, worse still, other smartphones, RIM has very recently launched a version of BlackBerry Messenger for the Andriod and iOS operating systems.

To measure the impact of SMS substitution on revenue, operators need to assess the combined effects of IP messaging user penetration, SMS revenue cannibalisation versus data ARPU uplift and elasticity.

1. User penetration
Juniper forecasted that the number of IP-based messaging users will surpass 1.3 billion worldwide by 2016 equating to approximately 18% of the total messaging users12. The trend of IM Apps that start to charge users may lessen penetration growth; however, the fees charged (i.e. US$1 per year for WhatsApp) are relatively low.

2. SMS ARPU cannibalisation and data ARPU uplift
Developed markets show the greatest annual decrease in SMS traffic per user. SMS revenues have decreased by 33% in Singapore, 10% in The Netherlands and 9% in Australia. Western Europe’s SMS ARPU is estimated to have fallen by approximately US$4 in 2012 while the US is expected to report declining SMS traffic this year13. Emerging markets are also likely to follow.

Given the relatively-high prices of SMS compared to IP messages in Western Europe, cannibalisation dominates the overall equation and is not offset by increased data ARPU from IP messaging. This results in a net ARPU decline of approximately US$4. In other regions, the data ARPU uplift again does not offset the cannibalisation of SMS by IP messaging, resulting in a net ARPU decline of between US$0.04 and US$1.20.

3. Elasticity
Although the financial impact of the elasticity effect (i.e. subscribers messaging more due to the significantly lower price per message) is currently marginal – particularly in the context of unlimited data packages – this effect will become more significant with the prevalence of picture, video and other short-form content in IP messaging applications.

Taking these three factors into consideration, Delta Partners estimates that by 2014 the impact of SMS substitution on revenues will range between ~7% (in MEA) and ~31% (in North America). This equates to a 19% decline in global SMS revenues resulting in a worldwide loss of US$21 billion by 2014 (Exhibit 4).


Voice substitution

On top of SMS substitution, customers are replacing mobile voice services with VoIP solutions. Voice ARPUs in the US and Europe decreased by 25% and 18% respectively between 2008 and 2012 while data ARPUs remained stable14. The principles of SMS substitution apply to a large extent to voice substitution, with three key exceptions:
1. User experience
Where IP messaging applications in general provide a better user experience than SMS services, the quality of mobile VoIP, or mVoIP, usually falls behind mobile voice services. With the rise of Wi-Fi and LTE, these quality issues are likely to be - at least partially - addressed.
2. Tariffs
VoIP is mostly used as a substitute for international calls. Based on the average revenue/MB and the data consumption related to VoIP (~250 KB per minute per connected party), average data charges per mVoIP minute are below one-fourth of a dollar cent (when not using Wi-Fi).
As a result, the price arbitrage between mVoIP and premium-priced international calls (e.g. 30¢ per minutes in Western Europe) is far greater than the gap between IP messages and SMS (6¢ per message in Western Europe).
Additionally, this affects international outgoing, incoming and visitor revenues – all of which contribute to EBITDA for operators.
3. Penetration
While the penetration of IP messaging has already reached meaningful levels, mVoIP user penetration is at an early stage. Future projections on growth vary largely, but even the more aggressive estimations (such as 410 million mVoIP users by 2015 from Infonetics Research) are low compared to the number of IP messaging users, which reached the same level in 2010.
Nevertheless Facebook, with its user base of more than 1.1 billion (of which the majority have mobile phones), may boost mVoIP penetration when it rolls out its VoIP services beyond Canada and the US. Through Facebook Home15, Facebook could enhance overall mobile engagement, potentially augmenting user penetration in the process as well.
In addition, Apple’s recent launch of its FaceTime Audio VoIP service as part of the iOS 7 operating system may further boost mVoIP penetration. Through FaceTime Audio, users can make free long distance and international calls to other Apple users through seamless integration with the FaceTime and iMessage apps and across both Wi-Fi and mobile networks.

Taking these factors into consideration, Delta Partners estimates that by 2014 the impact of voice substitution could be up to 6% of voice revenues depending on the region. This equates to a 3% decline in global voice revenues resulting in a worldwide loss of US$19 billion (Exhibit 5).
While globally, SMS substitution will have a greater impact than voice substitution in 2014, our projections indicate that in certain regions (e.g. North America), voice substitution will outpace SMS substitution, contributing more than half of the total substitution impact.
Overall, the substitution from mVoIP and mobile-based IP messaging is expected to result in a US$40 billion revenue loss worldwide by 2014.
Our forecasts are however contingent upon the future balance of power between OTT providers and telecom operators. It is noteworthy then that certain OTT providers (such as Apple and Microsoft)16 are relatively more dependent (compared to Facebook and WhatsApp) on operators to subsidise their devices. Accordingly, these providers are likely to be less aggressive in pushing their VoIP and IP messaging services. Others, such as Facebook, have recently employed a combination of approaches including operator partnerships17 and new service launches (e.g. VoIP)18 to increase OTT penetration and usage – both drivers of their advertising-based revenue model.
Delta Partners estimates that in the worst-case scenario where OTT providers are aggressive and operators’ responses are muted, losses can be significantly bigger – totalling as much as US$100 billion (US$57 billion from SMS and US$43 billion from VoIP substitution).


Despite their potential to enable substitution, smartphones nonetheless remain a retention tool.
  • Postpaid smartphone customers are locked into long-term contracts with high early-termination fess (to protect the operator’s device subsidy).
  • Customers migrating from non-smartphones to smartphones typically report increased satisfaction with their provider. As devices are a key determinant of provider choice, improved device experience for the customer is likely to minimise churn risk for the operator.
  • Additionally, if exclusivity agreements are reached, being the sole provider of certain devices may also improve an operator’s positioning. (Discussed in more detailed in our article, “The Smartphone Subsidy Conundrum”.)
Operators, including Sprint, report having decreased churn levels through the introduction of high-end smartphones such as the iPhone. In the period which the iPhone was introduced in the UK, the only operator to reduce its churn rate was O2 which offered the iPhone to its subscribers on an exclusive basis.

Addressing the Substitution Challenge

Operators first need to conduct a diagnostic to determine the degree of substitution. As noted, the key drivers of the substitution effect are smartphone penetration and the penetration of mVoIP and IP messaging applications within the smartphone base. Both of these drivers impact the number of subscribers that might substitute traditional voice and SMS services with VoIP minutes and IP-based messages and the extent to which they will do so.
Once operators understand where they are on the substitution curve, they will need to employ a combination of interdependent levers through four strategic programmes across both the revenue and the cost side of the business (as shown in Exhibit 6).
Specifically, in the early adoption phase (with relatively low smartphone penetration levels), operators may focus on data pricing. However, as penetration increases, the focus should shift to subscriber acquisition and retention costs. A retention subsidy based on customer lifetime value (incorporating revised ARPU uplift assumptions) and gradual network upgrades could manage network congestion and secure high-value customer lock-in.
Finally, smartphone penetration levels such as those in advanced markets (e.g. the United States, Europe or Singapore) will require rebalancing to position data at the centre of the proposition with voice positioned as a data-enabled application.
As operator responses to OTT evolve from blocking to tolerating to partnering, can partnerships with OTT players result in better monetisation of data traffic? The answer depends, at least in part, on where operators are on the substitution curve.



The net impact from voice and SMS substitution on operators’ top lines could be as high as US$40 billion by 2014 with IP messaging driving the majority of the revenue decline. Because the two most affected areas, international voice and SMS, are also the two most profitable products, this substitution will weigh on EBITDA and thereby pressure operators’ free cash flows.
Even when factoring in the positive impacts of smartphone penetration including data ARPU uplift and reduction in churn, operators need to re-evaluate their approach to OTT services, mobile data pricing and the level of handset subsidies reflected in their subscriber acquisition and retention costs.
So can data be profitable? In short, the answer is, “Yes - but substitution needs to be managed carefully.”
Delta Partners’ proprietary model provides a framework that can be tailored for individual operators to determine the amount of the substitution impact and propose commercial guidelines based on where the individual operator’s base is in the adoption curve of smartphones and its local mix of voice and SMS.