The telecom field has always been an interesting one but data has been sparse in India. Most private telecom operators have been hesitant from releasing data for various reasons. Airtel for example initially used to report the number of 4G subscribers only to drop it midway. Although TRAI releases several data regarding various operators for the circles they operate in, there’s still quite some data that’s left out. One of the most crucial data points used to be the number of towers and their specific composition circle wise. Up until recently, this data was hard to come by unless you were working in the telecom industry or had inside sources. However, with Project Leap, Airtel decided to make this data publicly available in one single place at www.airtel.in/leap. One key metric that always had been missing was tower data, and with Airtel making that available, it fills a major part of the puzzle to analyze Airtel.
What is Project Leap?
Airtel has been planning a massive Rs 60,000 crore investment in India over the next few years in an attempt to further strengthen its position as the top Indian telecom player and prepare itself for an impending onslaught from Reliance Jio. To achieve this, Airtel plans to expand its 3G and 4G coverage all across India, as well as improve its fixed broadband services significantly. In order to enable consumers to be able to track Airtel’s progress, the company launched Project Leap where customers can see the current 3G/4G sites of Airtel on a map, its voice sites and existing as well as upcoming sites. Now that we have all this data in our hands, it’s time to crunch some numbers.
1. All values in this article are weighted averages and not indicative of real life numbers which for sure vary greatly on a number of factors. Think of it as 10 litres of water were given to 5 plants, on average every plant got 2 litres of water. But in reality, water consumption of every plant would vary greatly depending on the characteristics of the plant, type of soil and various other factors.
2. I have done more calculations in this article than I have done in my entire school life, so there are chances that certain calculations might go wrong. If so, please comment below and I’ll have it corrected.
A. Tower density
Now this has been one data point that I’ve always wanted to know and I’m pretty sure a lot of people must be craving to know about. How much area does Airtel cover in your circle for every tower? And how does that help? Well, it helps in two ways. First off the tower density and coverage are proportional for the most part. Higher the tower density, better the coverage and secondly better tower density also helps in capacity. It’s important to note here that both coverage and capacity are dependent on three more factors namely quantum of spectrum, type of spectrum and population density.
Quantum of Spectrum – Higher the quantum of spectrum better is the coverage and capacity. In short, higher spectrum quantum is always better from a coverage and capacity perspective.
Type of Spectrum – Lower band spectrum i.e the ones below 1 GHz are great from a coverage perspective but not so from a capacity perspective. High band spectrum such as the 2.3 GHz band spectrum is great from a capacity perspective but not so much from a coverage perspective.
Population Density – Generally, larger the number of people in a given square km, higher will be the congestion and coverage would be lower since high-density areas are also bound to have more buildings in more compact space leading to signal attenuation.
Despite all the above-mentioned factors, having a general idea of km2 per tower gives a rough idea of the kind of coverage and capacity one could expect in their circle.
Note – I’m including the towers that are scheduled to be coming by March 2016 in this table.
Note 2 – Area covered is the geographical area of the state(s) the circle lies in. Most of the times the Licensed Service Area (circle) and the geographic area of a state overlap each other and differences are minor at best.
Note 3 – Data regarding Jammu & Kashmir hasn’t been given mostly because of security reasons.
|Circle||Voice Towers||3G/4G Towers||Area covered||Voice tower density (km2/tower)||3G/4G Tower density (km2/tower)|
|Andhra Pradesh & Telangana||12,436||10,903||275,045||22.11||25.22|
|Bihar & Jharkhand||9,874||4,888||173,877||17.65||35.57|
|Gujarat & Daman & Diu||7712||4367||196126||25.43||44.81|
|Madhya Pradesh & Chhattisgarh||9839||4348||443439||45.06||101.98|
|Maharashtra & Goa||15404||14762||311415||20.21||21.09|
|Tamil Nadu including Chennai||13168||13585||130058||9.87||9.57|
|Uttar Pradesh(East) & Uttrakhand||17230||7906||296769||17.13||37.53|
1) Lower the Voice tower density and 3G/4G tower density, it’s better because this shows the km2 covered by every tower. If every tower has to cover higher areas in terms of km2, then there are lesser towers in that circle compared to the area of the circle. So lower is better.
2a) There are now quite a few areas where the density of 3G/4G towers is lower/better than that of voice towers. These circles are Delhi, Karnataka, Mumbai and Tamil Nadu including Chennai where the number of 3G/4G towers is now actually exceeding that of voice towers. This also means that there are standalone 3G/4G in these circles.
2b) Maharashtra is very close in terms of having an equal number of voice towers and 3G/4G towers as shown by the density.
2c) In West Bengal, Himachal Pradesh and Andhra Pradesh including Telangana, the density of 3G/4G towers is just slightly higher than that of voice towers.
2d) Certain circles like Madhya Pradesh & Chhattisgarh and Kerala have a huge difference between the 3G/4G tower density and voice tower density where the 3G/4G tower density is dramatically lower than that of voice towers.
B. ARPU per Tower
With the help of ARPU data obtained from COAI (latest available data was till Q4 2014) and the tower related data made available by Airtel, it’s now possible to calculate just ARPU/tower data just as we calculate km2/tower data above.
|Circle||Voice Towers||3G/4G Towers||ARPU||Voice tower (ARPU/tower)||3G/4G Tower density (ARPU/tower)|
|Andhra Pradesh & Telangana||12,436||10,903||177.59||0.014||0.016|
|Bihar & Jharkhand||9,874||4,888||129.81||0.013||0.026|
|Gujarat & Daman & Diu||7712||4367||83.85||0.010||0.019|
|Madhya Pradesh & Chhattisgarh||9839||4348||101.22||0.010||0.023|
|Maharashtra & Goa||15404||14762||126.20||0.008||0.008|
|Tamil Nadu including Chennai||13168||13585||203.11||0.015||0.014|
|Uttar Pradesh & Uttrakhand||17230||7906||123.68||0.007||0.015|
1. ARPU stands for Average revenue per user which is nothing else but total revenue for the circle divided by the number of subscribers in that circle. So ARPU is the basically the revenue for every user in that particular circle/LSA. When ARPU is further divided by the number of towers, it helps us know how much revenue every tower is generating per subscriber.
2. This is just the revenue calculation and does not indicate profit as tower rents keep varying from area to area and there’s no data available on that. So a higher ARPU/Tower isn’t indicative of a higher profit/tower as tower costs remain unknown.
3. Even though it’s West Bengal that has the least ARPU for Airtel, but when ARPU/Tower is taken into consideration, it’s Maharashtra that’s the least with a score of 0.008 for both voice and 3G/4G towers. Similarly, from an ARPU/Tower perspective, Himachal Pradesh ends up giving the best score for voice towers and both Kerala and Himachal Pradesh end up giving the best score for 3G/4G ARPU/Tower. It’s important to note here that we don’t know what percentage of revenue originates from 3G/4G data in Kerala and Himachal Pradesh, so even though with the data available they seem to be the winners that may not actually be the case.
Note – I haven’t taken into consideration Uttar Pradesh and Uttrakhand because on their leap website, Airtel has mixed two circles namely U.P East and U.P West into one circle although the two circles have very different revenues and subscriber base for Airtel.
C. ARPU per km2
With the calculation of the above two metrics, it is now possible to calculate the price every user is paying for 1 km2 of the area covered by Airtel towers. This, however, doesn’t indicate the amount a user is paying for coverage from Airtel because that’s dependent on the type of spectrum, its quantity and how much Airtel has paid for it. It is necessary to divide ARPU/Tower with km2/Tower in order to achieve this metric.
|Circle||Price per km2 for voice towers|
(ARPU/km2 x 10-4)
|Voice tower density (km2/tower)||3G/4G Tower density (km2/tower)||Price per km2 for 3G/4G towers|
(ARPU/km2 x 10-4)
|Voice tower (ARPU/tower)||3G/4G Tower density (ARPU/tower)|
|Andhra Pradesh & Telangana||6.33||22.11||25.22||6.34||0.014||0.016|
|Bihar & Jharkhand||7.3 ||17.65||35.57||7.3||0.013||0.026|
|Gujarat & Daman & Diu||3.93 ||25.43||44.81||4.2||0.010||0.019|
|Himachal Pradesh||27.10 ||36.15||39.17||27.06||0.098||0.106|
|Madhya Pradesh & Chhattisgarh||2.22 ||45.06||101.98||2.25||0.010||0.023|
|Maharashtra & Goa||3.95||20.21||21.09||3.7||0.008||0.008|
|North East||9.12 ||82.22||105.74||9.17||0.075||0.097|
|Rajasthan||4.2 ||40.47||61.19||4.41 ||0.017||0.027|
|Tamil Nadu including Chennai ||15.19 ||9.87||9.57||14.62||0.015||0.014|
|Uttar Pradesh & Uttrakhand||4.08 ||17.13||37.53||3.99||0.007||0.015|
|West Bengal||9.96 ||10.04||12.76||9.40||0.010||0.012|
1. It is pretty clear that the price per km2 for the coverage of both Voice and 3G/4G towers appear rather similar and rightfully so since a common ARPU/Revenue has been taken into consideration for calculating both of them.
2. It’s amazing how different the revenue per km2 for every circle varies. Certain circles such as the metro circles of Mumbai, Delhi and Kolkata have rather astonishing amounts of revenue generated by every subscriber for every km2. The most ARPU/km2 circle seems to be Kolkata while the least seems to be Madhya Pradesh & Chhattisgarh.
3. Remember again that this is only the ARPU/km2 and in no ways is indicative of the profit since tower rents vary significantly from area to area and definitely circle to circle.
4. This again is a weighted average like everything else in this article and the ARPU/km2 vary very significantly from circle to circle to be able to draw any trends although ARPU/km2 keeps going down as we move down from metro to Category A to Category B to Category C circles.
D. Subscribers per tower
It is also now possible to determine the number of subscribers per tower although even over here it must be noted that Airtel hasn’t broken down the number of 3G/4G subscribers in every circle, so subscribers per tower for 3G/4G might be as indicative as real life levels might be. One could argue that Airtel has disclosed what percentage of users use data which is 2G as well as 3G/4G however this again is highly skewed. For example, Delhi is prone to have more data subscribers than a circle like Madhya Pradesh & Chhattisgarh.
|Circle||Voice Towers||3G/4G Towers||Subscribers||Subscribers per voice tower||Subscribers per 3G/4G tower|
|Andhra Pradesh & Telangana||12,436||10,903||23227059||1867.72||2130.33|
|Bihar & Jharkhand||9,874||4,888||26927312||2727.09||5508.86|
|Gujarat & Daman & Diu||7712||4367||8272527||1072.82||1894.32|
|Madhya Pradesh & Chhattisgarh||9839||4348||12111570||1230.97||2785.54|
|Maharashtra & Goa ||15404||14762||12687714||823.66||869.48|
|Tamil Nadu including Chennai||13168||13585||16829940||1278.09||1238.68|
|Uttar Pradesh(East) & Uttrakhand||17230||7906||27955343 ||1622.480||3535.97|
1. Mumbai seems to have the least number of subscribers per tower from a 3G/4G perspective. Mumbai therefore would have good amount of capacity from a per user basis as Airtel has 20 MHz of 4G spectrum in Mumbai and 5 MHz of 2100 MHz as well as 5 MHz of 900 MHz spectrum for 3G. Maharashtra also seems to be having far lesser congestion.
2. Orissa seems to be the most congested for Airtel from a 3G/4G perspective considering that around 9511.89 people latch to every 3G/4G tower if one were to assume that every subscriber belongs to 3G/4G which isn’t the case but even I were to assume that a far lower percentage the congestion is still pretty great. Airtel should definitely consider adding more 2100 MHz spectrum and probably win a block of 20 Mhz 2300 Mhz spectrum for Orissa. Bihar and Jharkhand seems to be the second most congested place.
3. In terms of voice, Rajasthan seems to be the most congested while Maharashtra the least.
Note: Congestion in real life varies from area to area and less congestion != more speeds, as speeds depend on a lot of factors such as quantum of spectrum, type of spectrum, backhaul and stuffs like MIMO etc.
Airtel Leap and the associated data disclosed by Airtel has been a great step forward in Indian telecom industry. I have used the data, however, the lack of 3G/4G revenue share circle wise has made ARPU/km2 and ARPU/Tower a little murky. I would have loved to determine the capacity per user for every circle but Airtel has started using 900 MHz band for both 2G and 3G without breaking out how much is just for 2G and how much for 3G circle wise. Similarly, the 1800 MHz band is being used for both 2G and 4G without breaking out how much is used for 2G and how much for 4G. If other telecom operators also disclose tower related data circle wise, we could compare them against each other. Another personal request to telecom executives is to be a little more social so that telecom enthusiasts can contact them and get out better info and data to the world. A great start would be to join Twitter with open DMs or, at least, reply or accept the LinkedIn invitation. Compared to other countries like the US, there is very little processed data available in Indian telecom industry and even if someone wants to crunch all these data and get it out to the world, it wouldn’t be possible without the help from telecom professionals themselves.