Dont worry C Mobile broadband is profitable by dfgh4bnmu


									 business case mobile broadband

 Don’t worry – Mobile
 broadband is profitable
                              figuring out the right price
                                      real operator cases
           in fact

                                  ▶ RECENT REPORTS (from Yankee Group and          month, of which  percent at busy hour,
                                  Heavy Reading) warn operators to be careful      the network cost comes to around eur
                                  of the traps they may face when calculating      .,– per subscriber in a suburban or
                                  business cases. At the same time, an increas-    rural area if the network is reasonably well
                                  ing number of operators are betting heavily      utilized. In a network where the operator
                                  on mobile broadband and positioning it as        already has sites available, from a G net-
                                  an alternative to dsl or cable. These opera-     work, and in an urban area with higher uti-
                                  tors are seeing tremendous subscriber            lization, the equivalent network cost per
                                  growth, reaching typically  percent – even      subscriber is below eur .
                                  as high as  percent – of the population after
                                  two or fewer years of offering the service.      H OW WE CALCULATED
                                     Ericsson, after more than two years of        To clarify our method, let’s start with a
                                  research in close cooperation with             cost analysis for just a radio base station.
                                  operators, has come to a clear conclusion:         Base stations account for a substantial
                                  Mobile broadband business cases show             part of an operator’s mobile broadband
                                  high profit margins, even if the operator is     costs, because of the large number
                                  only a pure bit-pipe provider catering to        deployed. A typical radio base station
                                  the needs of subscribers using computers.        (NodeB) in an hspa mobile network, offer-
                                     Here is a snapshot of a cost-benefit anal-    ing .Mbps, with a three-sector configu-
                                  ysis based on real and forecasted costs and      ration and a cell capacity of approximately
                                  revenues. (see graph )                          Mbps ( ×  Mbps totaling  Mbps), costs
                                     This analysis is made as an investment        up to eur ,. This investment is
                                  case, meaning that any costs that occurred       depreciated over – years.
                                  prior to the decision to launch mobile             Most people look at mobile broadband
                                  broadband are considered to be sunk costs.       production cost in terms of cost per giga-
                                  The result, even after only a few years and      byte (gb), and sometimes as cost per sub-
                                  with good but not aggressive growth,             scriber, as we did in the dsl comparison.
                                  shows margins in line with or above what         Let’s start with the gb cost to see what an
                                  operators typically generate today. The          operator can get out of one NodeB. In this
                                  conclusion at this stage is that mobile          calculation we leave out site acquisition/
                                  broadband provides a great addition to any       build, which is typically depreciated over
                                  operator’s business, and can match and           – years and only has a small impact
                                  compete effectively with dsl.                    on the result, although it represents a large
                                     To substantiate this, we must look at         investment. Some basic assumptions are:
                                  what costs a dsl operator may have. The          ▶ At Mbps it is possible to download
                                  typical cost per subscriber per month for          approximately gb in a month (
                                  the unbundling fee can vary by eur –            days). (Mb = , bytes × ,
                                  (plus equipment). It can go up to eur –        seconds per day.)
                                  in a wholesale (bitstream) situation for fiber   ▶ Each subscriber generates an average of
                                  and vdsl. For an operator owning the cop-          gb per month.
                                  per lines (assuming all equipment is written     ▶ Depreciation is five years.
                                  off ), the opex per subscriber per month         ▶ Cell capacity is  Mbps (grake,
                                  ranges from eur – in urban areas and             Qam,  Codes, and a m suburban
                                  rises to eur – in rural areas, depending        cell. Result from Ericsson Radio Network
                                  on the quality of the copper network.              Planning Tool and measurements in
                                     The comparable costs for mobile broad-          real-life networks.)
                                  band are the network opex and capex.
                                  Adding the cost for the sites, assuming            With a theoretical maximum of over
                                  average traffic of gb per subscriber per        ,gb per year, the production cost

                                                                                                                      mobile broadband business                       case

using this site would be eur . per gb or      efficiency important; otherwise, margins
eur . per subscriber per month. No site       would slowly deteriorate as users demand
will ever be utilized to its theoretical maxi-   more capacity and tariffs are lowered.
mum, but it could be used up to and above        Meeting this requirement is what the tech-
 percent. Thus the cost per gb lands at        nical evolution is all about. Consider a
eur . and the cost per subscriber at          given site configuration, starting at
eur . per month. Adding the site acqui-       . Mbps and going up to  Mbps using
sition/build into the equation with, for         hspa Evolution, which adds a cost of
example, a eur , investment depre-         around – percent. At the same time,
ciated over  years results in around eur       capacity increase is around  percent.
. per gb. A more loaded site with a ×       Going to even higher speeds using dual
configuration brings this cost down to           carriers and other features improves the
around eur . per gb. A shared G/G           site’s efficiency.
site with a × configuration costs around          Let’s look at a site with a cell radius of
eur . per gb.                                 m using Mhz spectrum and mimo
   If we do the same type of calculation for
the equivalent of a NodeB in a dsl net-            [graph 1]
work, namely the dslam, the result is as           Profitability of mobile broadband
follows: Assuming a price of around eur
, and an existing site in a suburban
                                                                                                                             Core 16%                    Core 14%
scenario with an average speed of around                                                                                                                Backhaul
                                                   Income statement
Mbps, the price per gb is around eur ..                                                                                   Backhaul                      31%
                                                   100%                                                                        54%        5%
The opex related to maintaining the cop-
per lines will of course add to this cost and                                                                                                            Radio 30%
                                                                                                 Non-                        Radio 30%
adds another eur .–. if we assume              80%
                                                                                               network                  Network OPEX                 Network CAPEX
the cost to be shared with voice (pots) and
an average consumption of  GB per                  60%                                                                     55%
month.                                                                                     Network OPEX                                                         47%
   Traffic distribution in the networks Eric-        40%                                         -45%                                    Network CAPEX
sson is monitoring (some  deployed                                                                                                           -8%
around the world) shows that only a few              20%
sites carry most of the traffic. A normal
scenario shows around  percent of the               0%
sites carrying more than  percent of total                         Revenue                     Opex                   EBITDA                 DA              EBIT
network traffic.                                   Source: Ericsson Analysis; BNET Sales Development Mobile Broadband Team
   This means that most of the sites
deployed in a network, whether for voice          Assuming 4 percent of the population has been reached four years after launch, with an ARPU of EUR 20 using
                                                  unlimited flat rate and average traffic per subscriber of 2GB per month. All costs related to mobile broadband
or broadband, can be considered “cover-           are included in the case. The revenue bar is aggregated revenue based on subscribers times ARPU (EUR 20).
age” sites. The number of sites carrying          Non-network costs include terminal subsidies, marketing, customer care, and IS/IT. Network opex includes
heavy traffic is even lower, perhaps –          power consumption and support for related equipment. Capex includes any expansions or additions required
                                                  to support HSPA depreciated over 8 years for hardware and 3 years for software.
percent for broadband. These sites will be
the first to require upgrading to “second
carrier” or higher modulation schemes to           [graph 2]
provide better capacity. But they are usu-         Projected 30 percent YoY growth
ally the sites that provide the shortest pay-
back time.                                         Monthly usage per subscriber (GB)
   Considering that cost per gb is related to
                                                   4,50                                                           EBITDA margin - (Normal)                            60%
how the nodes are used, the operator’s
challenge is to make use of available free         4,00                                                                                                               58%
capacity rather than risk congestion.
                                                   3,50                                                                                                               56%
THE IMPORTANCE OF SCALE                            3,00                        EBIT margin - (Normal)
When subscriber numbers increase, both             2,50
traffic and revenue rise. (see graph ) The                                                                                                                           52%
operator will eventually have to invest in         2,00
more capacity, in the form of additional                                                                                                                              50%
carriers, each using MHz of the wcdma                                                                                                                                48%
spectrum. Each added carrier represents an         1,00                                        Usage per subscriber - (Normal)
investment. Because these investments are          0,50                                                                                                               46%
driven by traffic from more subscribers,
there is, of course, a correlation to revenue.     0,00                                                                                                               44%
                                                                Year 1                   Year 2                    Year 3                Year 4            Year 5
  That makes the ability to improve cost-

 business case mobile broadband

                                                          In the long run, unlimited flat rate with a fair-
                                                          use clause is potentially cheaper and more profita-
                                                          ble for the operator than bucket plans.

                                                          (GB/month,  percent busy hour). It has a      voice and sms traffic, though not part of
                                                          capability to handle over  subscribers      mobile broadband, should be considered
                                                          at  percent load. That cell radius equals     because they affect overall network dimen-
                                                          approximately ½ km² and can be compared         sions. Mobile broadband must share net-
                                                          to New York with around , people           work capacity with other services, espe-
                                                          per km². Assuming an operator gets            cially the radio bearer, and voice in partic-
                                                          percent of the total population, that equals    ular. All operators in our research have a
                                                          , subscribers per site.                     G network covering – percent of
                                                             These examples show only the cost per        their respective populations.
                                                          gb for a NodeB. But if we look at the entire       We have noticed that operators often
                                                          hspa network (reasonably well utilized,         struggle with their own calculations
                                                          with the cost of site acquisition and build     because of difficulties in identifying or
                                                          included), we see that all costs, including     allocating costs that are strictly related to
                                                          the radio network controller (rnc) and the      mobile broadband.
                                                          core nodes sgsn and ggsn, typically rep-           In our work to produce real business case
                                                          resent a small part of the total cost per gb.   examples, we have proceeded as follows:
                                                          The mobile backhaul and optical transmis-          Analysis is based on existing traffic pat-
                                                          sion in the core cost less than eur . per     terns and forecasts, creating a scenario for
                                                          gb compared to a × NodeB at eur .          the next five years. The traffic and sub-
                                                          per gb. (see graph )                           scriber growth scenario determines the
                                                                                                          capacity required in all nodes including
                                                          R EAL- LIFE BUSINESS CASES                      radio and backhaul, and thus drives cost
                                                          Simple calculations don’t tell the full story   over the five-year scenario. The case
                                                          of a real business, but they provide a good     includes all non-network costs, such as
                                                          indicator. Let’s turn our focus to real-life    handset subsidies, marketing, and cus-
                                                          cases, based on research in cooperation         tomer care. Although marketing costs, for
                                                          with established operators from all parts of    example, may be much higher initially, we
                                                          the world.                                      believe that within three to five years they
                                                             Ericsson has developed a tool to make a      will stabilize at a level similar to today.
                                                          complete end-to-end analysis, including all     Therefore, the non-network costs will be
                                                          aspects relevant to the business case. Even     around  percent of revenue for a West-
                                                                                                          ern world operator and somewhat lower in
                                                                                                          low-arpu regions.
     [graph 3]                                                                                               Analyzing a static network won’t satisfy
     HSPA network costs
                                                                                                          an operator looking for real-life answers.
      Network costs are less than EUR 1,0 per GB                                                          So we look at key areas where there are
      New sites including site acquisition and build                                                      question marks.
                                                                                                             What would happen, for instance, if traffic
       7.2 Mb/s - 3x2                                          SGSN       < 0,6 €/Sub/Year                per subscriber increased dramatically? Not
       50% Utilization                                                                                    the statistical average, which can increase
                                   80% Utilization                      80% Utilization
                                                                                                          because a few users generate huge amounts
                     80% Utilization               80% Utilization                                        of data, such as when using peer-to-peer;
                                     RNC/BSC                            GGSN
                                                                                                          but rather a traffic increase that an operator
                                                                                                          must consider when dimensioning the net-
              RADIO         BACKHAUL CONTROLLER TRANSPORT                  PACKET             TOTAL       work. We also investigate what would hap-
     CAPEX 0,37 €/GB        0,01 €/GB       0,08 €/GB      <0,01 €/GB     0,08 €/GB          0,55 €/GB    pen if an operator moved away from E/T
                                                                                                          (backhaul on leased lines) and used micro-
                                                                                                          wave links instead. Finally we consider what
     OPEX 0,19 €/GB         0,01 €/GB       0,08 €/GB      <0,01 €/GB     0,08 €/GB          0,37 €/GB    effects that variations in subscriber uptake
     Shared                                                                                               could have on the business case.
     with 2G                                                                                                 The graph (see graph ) shows how an
                                                                                      SUM    0,92 €/GB
                                                                                                          increase of traffic per subscriber affects
      Note: Well utilized network, fully allocated costs – new site                                       profitability, assuming all other things are
      100,000 € for passive part, 20 years depreciation                                                   equal. We conclude that, within limits, we
                                                                                                          can maintain strong profitability even if

                                                                                                        mobile broadband business                   case

traffic increases drastically. In the scenarios
                                                   [graph 4]
modeled with operators, shown in the               Economies of scale
graph, we have calculated with network
capabilities up to  Mbps, being intro-
duced in some networks during . But                 Data usage does not significantly impact the business
we need to remember that technology is
evolving very fast. When allowing the net-              case due to economies of scale
work to evolve all the way into a five-year
scenario and assuming speeds up to                   Network cost per GB per month                                                           Profitability
 Mbps, we have the tools we need to keep              €3                                                                                        60%
up with demand and remain profitable.                                                                                            EBITDA
   The most important element of profita-
bility is subscribers paying for the service.                         2.2
The graph shows how profitability drops                 €2                                                                                        40%
unless enough subscribers are added (see                                OPEX                                                          EBIT
graph ). Many operators still have a long                                                   1.3
way to go before they have enough sub-
scribers to be profitable. At the same time,            €1                                                    0.8                                 20%
we do see a few operators who, after two                               CAPEX                                                    0.6
years, have reached a population penetra-
tion above what we use in this sensitivity
analysis. What still has an effect on the end                                                                                                      0%
                                                                         1GB                2GB               4GB               8GB
result for these operators is the amount of                                               Base case
nodes that they have built out so far. The
population coverage does vary between                                            Dimensioning traffic per subscriber per month
operators and the prerequisites vary by             Source: Ericsson Analysis

country. The Nordic countries for example
have quite low population density and
therefore require more sites per inhabitant        [graph 5]
                                                   Customer uptake: The driver for profitability due to economies of scale
than countries with high population density.
                                                     Network cost per GB per month                                                           Profitability
If distribution cost per gb is counted in              €25                                                                                        60%
euro cents, and traffic is not an issue at                                                                                            EBITDA
most of the sites, why do we keep hearing              €20                18.7
that traffic will kill the networks?                                                                                            EBIT
   The most common argument is that                                                                                                               40%
heavy downloading through file sharing via
peer-to-peer applications generates huge
amounts of traffic. As a consequence, few              €10
operators dare adopt the de facto price
model on wireline broadband, namely,                    €5              CAPEX                 4.3
                                                                                                               2.5              1.9
unlimited flat rate. Even though this flat
rate promotes subscriber uptake and is the              €0                                                                                         0%
easiest pricing for consumers to under-                                                  1,25 M Subs       Base case        3,75 M Subs
                                                                    0,25 M subs                            2,5 M Subs
stand, operators still worry about uncon-                            1.5xPrice
trolled costs from heavy increases in traffic.
   All or most mobile networks today have           Source: Ericsson Analysis
been running voice, sms, mms, and some
mobile data traffic. None of these have gen-
erated much traffic per subscriber. Revenue       and video streaming. Peer-to-peer alone
growth has been well aligned with traffic         accounts for over  percent of all house-
growth (and thus traffic cost) per sub-           hold-generated traffic. And with traffic per
scriber. Then along came this new service         subscriber increasing at a yearly rate of 
that, compared to sms, for example, easily        percent, driven mainly by file sharing, we
generates , , or even  times more        should look at peer-to-peer for opportunity.
traffic per subscriber. This has triggered a        Operators want as many profitable sub-
knee-jerk reaction among operators, who           scribers as possible. This means that
think that such a service can’t be profitable.    investments made in the network are
   The three applications generating the          driven by the bulk of subscribers and not
highest volume on the internet today are          by a few heavy users. Ericsson addresses
peer-to-peer file sharing, web browsing,          this by introducing traffic-handling prior-

 business case mobile broadband

                                                         ity throughout the network, which allows         even  gb already on the market must be
                                                         the network itself to manage its resources.      fulfilled by the operators. With bucket
                                                         The operator must introduce a fair-use           plans, the traffic volumes for dimensioning
                                                         clause in the subscriber’s contract so it can    the network would continually increase,
                                                         manage heavy usage intelligently. Most           and the only limitation the operator has is
                                                         commonly, mobile broadband operators             the bucket size. However, the fair-use level
                                                         use unconditional throttling today, which        for an unlimited flat rate offer may not
                                                         means that once the fair-use level is            need to change at all, or at least very little,
                                                         reached, the throughput drops to a prede-        over time since it does provide an “all you
                                                         termined level. Typically, though, these         can eat” model. The segmentation is
                                                         speeds don’t allow for meaningful use of         instead achieved through speed and price.
                                                         the broadband connection.                           Unlimited flat rate is a complex issue,
                                                            Traffic-handling priority (see graph )       and it is difficult to predict what will actu-
                                                         gives the heavy user a lower priority in the     ally happen in a network when this model
                                                         network once the fair-use level is reached.      is applied. Although the model does allow
                                                         The heavy user experiences a degradation         each user to generate as much as they
                                                         of the service only when competing for           want, other factors influence the outcome.
                                                         resources in a congested situation. But in       The operator’s chosen position in the mar-
                                                         peer-to-peer, the experienced reduction of       ket determines which subscribers it
                                                         the throughput will, over time, be limited.      attracts. This in turn defines the behavior
                                                         Only in heavily loaded cells does a peer-to-     of its subscriber base. Great variations
                                                         peer user experience serious problems.           exist between operators in the same mar-
                                                         Those sites would soon be targeted for           ket with similar packaging and pricing.
                                                         capacity upgrades since it is normal usage          We have based this pricing discussion
                                                         that is creating the congestion.                 entirely on pc-based subscribers to prove
                                                            Traffic-handling priority allows an oper-     that there is good profitability even in
                                                         ator to focus on dimensioning the network        offering a simple bit pipe. Introducing
                                                         for normal usage while still allowing            intelligent management functions in the
                                                         unlimited or “all you can eat” traffic. The      network allows the operator to handle all
                                                         consumer gets better overall quality and         sorts of situations, such as separating appli-
                                                         the comfort of using an unlimited service        cation streams from each other, or varying
                                                         that does not generate surprises on the bill.    traffic – and perhaps pricing – depending
                                                            In the long run, unlimited flat rate with a   on time of day; or giving different priority
                                                         fair-use clause is potentially cheaper and       to smartphone users over pc users; or giv-
                                                         more profitable for the operator than            ing paying mobile-tv viewers a higher pri-
                                                         bucket plans. Subscriber uptake aside, we        ority than “best effort” internet. The possi-
                                                         see that bucket size is increasing drasti-       bilities are nearly endless, and it comforts
                                                         cally, driven by competition and as a way        operators to know they exist.
                                                         means to segment the market. It’s probably          When we add them all together and put
                                                         fair to assume that buckets of , , or        them on top of the simple bit pipe, we
                                                                                                          improve on an already powerful concept
     [graph 6]                                                                                            that will continue to generate good profit
     Traffic handling priority                                                                            for operators in the future.
                                                                                                             And finally, it is nice to get the kind of
     Heavily loaded cell, filled by p2p traffic –                                                           confirmation we recently received when
     Regular users still experience the full bandwidth                          Bittorrent                one of Ericsson’s customers reported to us
                                                                                Internet Radio            that its current cost per gb for mobile
     Bandwidth                                                                  YouTube                   broadband is now down below eur , after
                  “7.2” Mbps limit                                                   only two years of operation.

                                                                                Average throughput                                      ▶ GREGER BLENNERUD is
                                                                                experienced by                                          Director of Business Develop-
                                                                                heavy user                                              ment at Ericsson Business
                                                                                                                                        Unit Networks, responsible
                                                                                                                                        for mobile broadband for op-
                                                              Experienced throughput,                     erators and consumers. He has over 20 years in tele-
                                                              all others (web, email,                     com with experience in software development, busi-
                                                              streaming)                                  ness intelligence, sales, and marketing. He holds a
                                                                                                          master’s degree in Business Administration and Econ-
                                                                                                          omy from the University of Uppsala, Sweden.


To top