The Communications Authority of Kenya puts Internet penetration at 112 per cent, translating to 51m users.
Benta Sila lives in Nairobi’s Kasarani neighbourhood with her infant daughter, husband and sister. Yet, in Communications Authority of Kenya data, this household alone accounts for more than 100 Internet users.
This is because Ms Sila has a fibre subscription and in the formula used by the Communications Authority (CA) to estimate the number of Internet users, it assumes every fibre subscription is used by 100 people.
This is a formula that might have been relevant a decade ago but which, in the era of the fibre to the home, will produce highly distorted figures. For every mobile data subscription CA counts one Internet user, which does not take into account multiple SIM card ownership or even tethering devices in the same household.
It is using this formula that the CA recently announced that Internet penetration in Kenya had reached 112 per cent in the quarter to September 2017, translating to an estimated 51.1 million Internet users in a country of about 45 million.
At face value, these figures would suggest that every Kenyan, including Ms Sila’s infant and about 20 per cent of Kenyans who are illiterate, use the Internet.
Dirk-Jan Koeman, a veteran of the telecommunications industry and chief business development officer at Poa Internet, says while CA method of calculating penetration is not “uncommon” globally, it is prone to produce “inflated” numbers.
“These figures are not really reflective of what is on the ground at all. This is a view based on figures from operators that have not been verified on the ground,” says Tom Makau, a telecoms analyst.
So, how many Internet users are in Kenya? The short answer to this question is that no one knows for sure.
The CA admits that its own formula is a weak stand-in that is “not ideal for the Kenyan scenario” or for other African countries.”
The Authority says there is no accepted scientific method for measuring Internet penetration and that ideally, a national survey or census would be the best route. It is a costly route.
Our research found a wide array of statistics by many credible institutions.
In its most recent report on Measuring the Information Society, the International Telecommunications Union (ITU) estimates that 26 per cent of the Kenyan population were online in 2016, a number that is echoed by the World Bank.
Research ICT Africa, a South Africa based body, also reports that 26 per cent of the population is using the Internet. The GSM Association, a global trade body for the telecommunication sector, reports that at least 52 per cent of Kenyans were using mobile Internet in quarter four of 2016.
The formulae used by the World Bank, ITU and GSMA were not immediately available but Research ICT Africa carried out a survey based on a representative population of 1,171 Kenyans.
Contrast these figures with the December 2016 estimate from the CA indicating that 89.7 per cent of Kenya’s population was using the Internet.
Because CA does not clean up its data to remove duplicates or verify the number of users that are actually connecting to the internet via a connection, this is essentially the number of Internet subscriptions in Kenya, including duplicate subscriptions.
“The challenge of double counting is almost inevitable …In Kenya people mostly access Internet through their mobile devices, the same person has Internet at home, and also can use the Internet at work, cyber cafes etc,” says the CA.
Nevertheless, CA disputes reports of a 26 per cent Internet penetration, saying that this is low.
The 2016 ICT Gaps Study found that at least 78 per cent of Kenyans were covered by 3G, suggesting that a similar level of access, if not use.
Taken into consideration with rising fibre connectivity and Kenyans who use the Internet in public access points like schools or cyber cafés, the CA says, these figures rubbish the idea that penetration in Kenya could be as low as recorded by the ITU.
“The percentage accessing Internet is definitely higher than 26 per cent,” the Authority says.
However, Mr Koeman, whose company connects low-income communities to cheap Internet, says that the 26 per cent penetration rate does not sound immediately wrong.
Further, he points out that low penetration numbers are not necessarily a bad thing as they could draw funding and investment. Discrepancies in the definition of an Internet user may also account for the different figures from different organisations.
While the Authority counts all subscriptions that have brought in money to telecoms over the 90 days preceding their reports, Research ICT Africa counted people who had used the Internet over the 60 days preceding their report.
Safaricom #ticker:SCOM reports 30-day active users in their financial results, meaning that these figures reported to shareholders can seem deflated in comparison to the CA’s.
The CA says that it is working to come up with a better way of measuring Internet penetration, borrowing a leaf from other African countries that are experimenting with new formulae.
Nigeria, for instance, focuses only on cellular connections, more appropriate in a continent where mobile still accounts for 99 per cent of all subscriptions. The country recorded a 47.7 per cent internet penetration in June 2017.
Uganda uses a three-year old household survey as the base for its penetration statistics, adding new subscriptions to the number of Internet users recorded in 2015. The Authority says that the new methodology it is developing will be presented to its board by February.
“Once approval is granted, we shall immediately implement the Board decision, probably in the Quarter Sector Statistics Report of the current fiscal year (October-December 2017),” the regulator said in a statement.
In the long-term, however, the most accurate answer to the question of how many Internet users will have to wait.
The CA says that Internet usage was measured during the just concluded Kenya Integrated Population Housing Census. The data is also set to be collected again when the Kenya National Bureau of Statistics carries out the national census next year.