Cisco Blogs

Measuring Broadband Penetration – By population, household or something else?

October 7, 2010 - 1 Comment

In working across different countries on their broadband strategies, inevitably the question arises of how does one best measure broadband penetration. Some countries have published figures of broadband based on a population (per capita) basis, while others have opted to adopt measures of broadband penetration by household. Why the difference of approach and what’s the significance of using one over the other?

Historically, national statistics’ penetration figures have been computed by dividing by total population. National indicators relating to per capita growth, literacy, education levels, mortality, etc. were done on the basis of population as they are used as an indication of the maturity of a particular country across its citizen base.

The measures based on household numbers started to come about when people questioned whether it made sense to measure broadband per capita, where even in a country that is considered to be successful in broadband, it is the norm for family members to share only one connection (albeit a high speed one) rather than for each to have their own separate connection. In days of measuring fixed telephone lines, it may still make sense for each family member to have their own fixed voice line (since the family members cannot use the same voice line at the same time). However, in today’s broadband Internet where the technology is by nature capable of being shared, this measure by population does not seem to fit as well as by household. There are also those who make the case that as a target, getting one connection to each household is good enough to give the population base access to Internet, and it is not necessary to strive for one connection per person.

One side effect of using household instead of population is that the penetration numbers appear better (since we are dividing by a smaller base) and indirectly making it easier for governments to talk about broadband reaching homes as an objective to strive for in bridging the digital divide. Nonetheless, these are not the primary reasons for measuring by household, but instead because it is a better measure for the purposes of broadband and Internet based on the way the technology is deployed and used. Other traditional indicators such as literacy should continue to use population, because population is the more appropriate measure for the purpose of the indicator.

Two complications however.

One is mobile voice penetration versus mobile broadband. In measuring mobile voice penetration, it makes sense to continue using population as the base since mobile and smart phones tend to be individual rather than shared devices. However, mobile broadband technology has evolved to the point where it is capable of replacing a home fixed-line connection. Given that, how should broadband penetration numbers based on household capture the segment who use mobile broadband in place of fixed broadband, and how should they be distinguished from mobile voice? Since voice and mobile broadband can be on the same subscription plan from an operator, should it be double-counted for both mobile penetration and broadband penetration?

One compromise is to distinguish mobile data-only type subscriptions from voice-and-data subscriptions, and count only the former data-only subscriptions in the broadband penetration figure. Even then, it is not ideal as many of such data-only plans are used in personal devices like 3G dongles for computers or installed in data devices such as tablet and netbook computers — in such instances, a household may have more than one subscription as the devices are by nature more individual than shared. There are also wireless home routers today that use wireless data plans, but those are not yet the mainstream choices, especially when fixed line broadband offers comparatively much better performance, quality and value-for-money.

The second complication is about business subscriptions. Household numbers clearly do not include businesses, but in reality, businesses are also subscribers of broadband. Arguably, their use is of even greater importance from the perspective of economic impact of broadband to a country. However, a penetration figure based household would clearly not include businesses. The question is whether the total number of subscriptions include business subscriptions or only consumer subscriptions. I suspect most subscription numbers reported today do not make the distinction. So we lose some accuracy there in terms of achieving the purpose of the penetration indicator.

All-in-all, perhaps household penetration numbers make the most sense for broadband today, with the option of including data-only wireless broadband connections within the count for the total number of broadband subscriptions, and excluding the voice-and-data subscriptions. Perhaps there is room to explore another measure that includes both households and businesses, and divide the total subscriptions by this new measure. Alternatively, different indicators of household and business broadband penetration could be used. Organizations such as the OECD have been studying ways of computing the wireless broadband indicator and it will be interesting to observe how the landscape changes.

In any event, having a commonly understood basis to talk about penetration is important to make sure that our conversations comparing broadband developments across countries are made on an apple-for-apple basis. Too often, we talk about broadband penetration without paying too much attention to whether it is measured by population, household or something else, leading to some inappropriate comparisons. This is not to mention the difference of opinions about the definition of broadband in relation to minimum speed. But that’s a topic for a whole different discussion for another time.


In an effort to keep conversations fresh, Cisco Blogs closes comments after 60 days. Please visit the Cisco Blogs hub page for the latest content.


  1. Excellent points, Seow Hiong. There are many definitional issues surrounding internet and broadband penetration measures, and the importance of high speed internet adoption points to the need for standard definitions to be applied across geographies so penetration numbers can be appropriately compared. A few other issues to be aware of:

    1) Per capita penetration values may under represent the percentage of a population with access to the internet. This may be the case when comparing per capita values between two countries where average household sizes differ significantly (e.g. between Germany, where it is low, and say India, where average sizes are around twice as large)
    2) The actual definition of ‘broadband’ itself varies. The latest UN/ITU statistics still define it as connections with download speeds of 256kpbs or higher, though may be in the process of redefining it at a higher speed. The speed definition also is very important when comparing quality of access say between wireless versus fixed wireline broadband (is the ‘broadband’ experience via a mobile phone comparable to cable modem or DSL access connected to a desktop PC?).
    3) As mobile broadband, and more broadly, wireless broadband becomes more prevalent, the measure of mobile access is continuing to be refined. For accuracy, it will be important to measure mobile subscriptions based on active subscriptions of data-transmission, as opposed to just the aggregate of mobile devices with the possibility of data transmission. (And the ITU is in the process of refining this measure as well)