estimating market share through mobile traffic analysis linkedin
TRANSCRIPT
ESTIMATING MARKET SHARE
Through Traffic Analysis
Dr. Asoka Korale, C.Eng. MIET
What is Market Share Based on
what the Telecom Regulatory Commission says
Subscribers Revenues Traffic
Require Objective Method Telecom Regulatory Commission – as reported by operators
Subscribers Defined in terms of revenue generating base – Based on window in time over which subscribers are active
large variation depending on selected time interval
Revenues From published accounts where available
Traffic – Contribution of each operator to total traffic produced Intra (on net)/Inter (off net) operator call volumes and
aggregate minutes of use
Traffic Based Estimate Use known intra and inter operator traffic volume
measurements
Express each operator’s traffic contribution in terms of known intra net traffic volume contribution Amount of traffic produced proportional to subscriber base No direct dependence on composition of the base (pre / post paid), time
windows or subscribers (can be related to subscribers if necessary) As market share is expressed as a proportion of known (Dialog) traffic no
impact of variation in RGB during different time windows Predict corresponding subscriber shares of other operators utilizing own
(Dialog) RGB and each operator’s estimated market share
Another view of market share - the proportion that each operator contributes to total traffic generated
Assumptions All subscribers are uniform across all of the networks, and they are all
equally likely to make a call to any other subscriber in any network. Under this assumption of uniformity all subscribers are also equally likely to
receive a call from any other subscriber in any network.
The number of calls (traffic) originated by a particular network is directly proportional to the number of subscribers belonging to that network.
The number of calls “attracted” to (terminated) in a particular network is also directly proportional to the number of subscribers belonging to that network.
All subscribers are not uniform as they generate traffic according to their specific packages and prepaid/ post paid segment
Adjust for effect of differing tariffs and impact on generated traffic
Total subscribers10,000
Airtel-500
Mobitel-2000
Tigo-750
Hutch-750
Dialog - 6000
Total calls 1000
600
30
120
45
45
1000/10000 *6000 600/10000*6000
600/10000*4000
360
240
240/4000*500
240/4000*2000
240/4000*750
240/4000*750
50
200
75
75
1000/10000 *2000
1000/10000 *500
1000/10000 *750
1000/10000 *750
Example Network Consider a network space of 10,000 subscribers
that generate 1000 calls in a busy hour the view from Dialog’s position
Figure 1
Dialog6000 subs
Mobitel2000 subs
Tigo750 subs
Hutch750 subs
360 calls
240 calls
30 calls
120 calls
Airtel500 subs
45 calls
45 calls
The Network Space from Dialog’s Position
Dialog will originate (1000/10000)*6000 = 600 calls uniformly to all subscribers (in all networks).
(600/10000)*6000 = 360 calls will be to Dialogs own (D2D) subscribers.
Of the 600 calls, (600/10000)*4000 = 240 calls will be to other networks The 240 outgoing calls to other networks are distributed as (240/4000)*500 = 30 calls to Airtel (240/4000)*2000 = 120 calls to Mobitel, and so on…
The Network Space from Airtel’s Position
The view from Airtel’s position
Airtel
500 subs
Mobitel
2000 subs
Tigo
750 subs
Hutch
750 subs
2.5 calls
47.5 calls
30 calls
10 calls
Dialog
6000 subs
3.75 calls calls
3.75 calls
Figure 2
View from Airtel’s Position Airtel will originate (1000/10000)*500 = 50 calls –
uniformly to all subscribers (across all operators). (50/10000)*500 = 2.5 calls will be to Airtel’s own subscribers. (50/10000)*9500 = 47.5 calls will be to other networks
The 47.5 outgoing calls to other networks are distributed as (47.5/9500)*6000 = 30 calls to Dialog (47.5/9500)*2000 = 10 calls to Mobitel, and so on…
Predicted symmetry in the traffic between operators Large operator makes a smaller proportion of OG calls to smaller
operator Small operator makes a larger proportion of OG calls to large operator Expect similar number of calls (MOU) from Dialog to Airtel and
from Airtel to Dialog
Results on Inter Network Traffic Symmetry
Total traffic between Dialog and other networks in a busy hour - over 7 day period (in 105)
Within +/- 10%
Total Incoming
MOUs
Total outgoing
MOUs
Total incoming
calls
Total outgoing
Calls
Airtel 3.5475 3.3206 2.173 2.06
Etisalat 6.8092 8.3425 4.407 4.881
Hutch 0.8331 0.8114 0.461 0.472
Mobitel 10.371 11.0604 5.449 5.756
Adjusting for Tariffs Differing on net and off net tariffs affect average
duration of on net and off net calls Assumption: Call length is inversely proportional to
tariff Ex: If on net tariff is 1 Rs per min and off net tariff is 2 Rs
per min. A subscriber who spends 4 minutes on an on net call will spend ½ the time on a off net call or two minutes.
Adjust the MOUs by the ratio between the off net tariff to the on net tariff as the off net MOUs will be under stated by this factor
(any proportionality constant will cancel)
The traffic attracted to an operator is proportional to the subscriber base (market share)
Adjusting MOUs Off net MOUs understated due to difference in tariffs
Scale off net MOUs by the ratio between off net to on net tariff
Market share independent of “absolute” number of subscribers ie RGB can differ based on time window but share wouldn’t change
D2D: 360 calls => 360*4 = 1440MOUs
Share Mobitel = 1/3/(1+1/3) = 25%
Dialog6000 subs Mobitel
2000 subs120 calls => 120*2 MOUs
Adjusted MOUs = 240 * (D2ND tariff/ D2D tariff) = 480 MOUs
Proportion Mobitel = 480/1440 = 1/3
Share Dialog = 1/(1+1/3) = 75%
Estimate Based on Aggregate Traffic
Using aggregate MOUs to individual operators and average on net and off net tariffs, with the pre and post paid split
Dialog
6000 subs
Mobitel
2000 subs
ToT_D2D_MOU: 360 calls => 360*4 = 1440MOUs
120 calls => 120*2 MOUs
iinetworkproportion
MOBproportionMOBshare_1
__
MOUDDTotMOUadjePTotMOUadjPoPTotMOBproportion
_2_)__Pr____(_
tariffDDePtariffNDDePMOUePMOUadjePTot
_2_Pr_2_Pr_Pr__Pr_
tariffDDPoPtariffNDDPoPMOUPoPMOUadjPoPTot
_2__2_____
Required Measurements
D2D MOUs (pre pay and post pay) Outgoing MOUs to each Operator Average pre pay D2D tariff and D2ND
tariff Average post pay D2D tariff and D2ND
tariff
Implementation Required data already available in
monthly spreadsheets and easy to automate inter operator MOUs
Interconnection details voice spreadsheet Intra net MOUs, average off net / on net,
pre pay and post pay tariffs Daily performance spreadsheet
Easy calculation
Estimate Based on Aggregate Traffic
Using aggregate MOUs to individual operators and average on net and off net tariffs
March April MayShare Airtel 0.0780 0.0765 0.0747Share Hutch 0.0164 0.0173 0.0181Share Mobitel 0.2526 0.2534 0.2542Share Etisalat 0.1863 0.1865 0.1864Share Dialog 0.4664 0.4660 0.4663
Validation: Call by Call Analysis Observing predicted symmetry The tariffs applicable to the individual subscriber package used Results of a call by call analysis in a busy hour over 7 day period,
using tariffs from packages applicable to individual subscribers The short term estimates converge to the aggregate based
estimate
MayShare Airtel 0.0763Share Hutch 0.0187Share Mobitel 0.2599Share Etisalat 0.1787Share Dialog 0.4663
Conclusion: Main Points in the Estimation
Estimates computed by expressing traffic volume to other operators (off net traffic) in terms of the intra net traffic (D2D traffic) The outgoing traffic to each operator gives an indication of the weight of
that operator in terms of Dialogs own subscriber base Traffic attracted by other networks is proportional to their internal
subscriber base (oranges) – but expressed in terms of Dialog base (apples) Indirectly relating each operator’s base to Dialog base
As outgoing traffic is utilized - estimate not dependent on specific details of each operator’s internal pre /post split or on net / off net tariffs
Even loss of symmetry in the traffic will not affect validity of the estimates as calculation is based on relating the outgoing traffic to on net traffic
Once other operator traffic (oranges) expressed in terms of Dialogs intra net traffic (apples) - The ratio (of apples to apples) of the off net / on net traffic volumes will give proportion and indicate market share
Thank You
Relating the OG MOUs
X X X X X X
Y Y
12 MOU
4 MOU(adjusted)
Dialog
Operator Y
Irrespective of the type of subscriber that operator Y has, the market share estimatedepends on the outgoing D2D and D2ND MOUs. The weight of operator Y will be expressed in terms of Dialog subscribers (MOUs).
If 6X attract 12 MOUs OG, then 4 MOUs OG (attracted by Y) Imply Op Y has (6/12)*4->, 2X type subscribersShare = 1/3 /[1 + 1/3]
Estimate independent of OpY tariffs and subscriber composition