How many Internet users in SA?

People often ask me about levels of Internet access in South Africa. I must confess that I watch these stats like other people watch football scores. It’s fascinating to lift the lid off the studies and see how the researchers have arrived at their deceptively simple numbers.

Anyhow, (and because some of you have asked!) here are some starting points for understanding internet access and usage in South Africa:

Mobile subscribers and internet adoption in SA: 1996-2009
Mobile subscribers and internet adoption in SA: 1996-2009 : ITU

A good source for annual cross-country comparisons of desktop internet access (and mobile subscription data ) is the ITU( International Telecommunication Union).  They currently put SA at  12.3%  fixed line internet access (2010) . Although access is growing, it’s shown nothing like the massive expansion of mobile subscriptions which the ITU’s data also reveals. (CAVEAT – the impressive S-shaped curve in the graph reveals growth in the number of SIM cards rather than users).

For more specifics about access in South Africa there’s AMPS (All Media Products Survey) conducted by the South African Audience Research Foundation. This survey provides overwhelming levels of detail about media use (15+), particularly about that of higher income groups (or higher ‘living standards’, to use their parlance). It comes out twice annually. Largely informed by the preoccupations of market researchers, it is thus not the best source for understanding use in lower income brackets. Although it’s expensive to subscribe to the full AMPS database you can get quite a bit of info from the free queries they allow to registered users via their eighty20 online interface.

At the moment, according to the  AMPS 2011 Individual survey (Jul’10 – Jun ’11),  these are their numbers:

  • 18%  (accessed internet in last 4 weeks)
  • 20% (accessed internet in last year)

Other surveys (notably ResearchICTAfrica) aim to provide information about internet access and usage, and to identify those who are currently excluded. These surveys are more informative about the challenges of access to internet and phones for South Africans in lower income brackets and in rural areas. The ResearchICT Africa survey allows comparisons with other African countries, but is conducted less frequently, and so unfortunately the numbers are quickly out of date.

AMPS now allows us to piece together a great deal of info about mobile internet. It tells us, for example, that although 22.88% of South Africans use mobile instant messengers such as MXit daily (which shows how many people are using data/internet on their phones), only 7,4%  said that their primary means of internet access was a cellphone and only 6.03% said they accessed the web or internet daily from their phones. AMPS

This discrepancy shows the importance of mobile messaging rather than web-style information access, but also raises questions about how mobile internet access is defined. These questions were originally raised in Tino Kreutzer’s research, which I’ve mentioned in a previous post. Many people have phones which are data-enabled, but they are not all internet users. It’s also likely that many people who do use the internet aren’t telling interviewers that they use the internet, when in fact they do. If we include all use of the TCP/IP protocol as ‘internet’, then use of MXit and other mobile apps should be counted along with desktop-style ‘web’ use (http protocol) as mobile internet use.
The diversity of mobile internet use makes things particularly tricky. In  Arthur Goldstuck’s terms, we should be thinking about the different ‘tiers’ of the mobile internet: (i) WAP, (ii) mobile apps such as MXit and (iii) web browsing.  I think survey questions are better when they ask specifically about particular online brands – Google, MXit, Facebook etc  – people don’t know the term ‘instant messaging’  but they do know whether they use MXit.

More nuanced figures are available from Goldstuck’s 2011 World Wide Worx study, which details Facebook and MXit use, and shows the gap between rural and urban areas. This captures some of the dramatic growth happening at the moment. World Wide Worx put the figures for mobile internet at urban (39%) and rural (27%). According to a large potential client, MXit currently reports 11 million active users (which is a good match for the AMPS figures cited above). A large live activity map photographed on 1 April 2012 shows how MXit use is spread around the country, but concentrated in urban areas (as is the population!):

Activity map of MXit use - posted by Arthur Goldstuck @art2gee

The latest World Wide Worx study shows Facebook growing rapidly and starting to trump MXit in urban areas. Twitter is also on the rise, no doubt driven by its popularity with journalists, celebs and the mass media. MXit is likely to hold onto its early advantage,  particularly among low income users because of the high cost/bit in South Africa and Facebook Zero is not available on the SA networks. In some new research, Jonathan Donner and I have also found that many people pay more than they should for data because very few buy data bundles. Also, many people don’t like to buy more than R5 or R10 airtime at a time, but I’m sure Facebook in particular will get a boost from some of the cheaper current data offerings by MTN and Cell C.   Again it’s important to consider these numbers in relation to the fact that the World Wide Worx survey doesn’t include informal settlements or ‘deep’ (less accessible) rural areas. There’s also a big gap in relation to under 16s, who are big early adopters but are not included in any of the big surveys. As the landscape changes so fast it’s also worth trying to ask about forms of communication which are growing in popularity. My own research suggests that mobile internet is rarely used in the same way as desktop internet, and that we should rather conceptualise mobile internet as configuring a range of socio-technical contexts.

To illustrate some of the difficulties in answering the question ‘how many Internet users in SA’, here are some survey questions from past surveys:

Survey A: Research Internet Africa

1.      Do you know what the internet is?

a.       Yes

b.      No

2.      Do you ever use the internet?

a.       Yes

b.      No

Survey B: All Media Products Survey

D2. Have you PERSONALLY accessed the Internet/World Wide Web in the PAST 12 MONTHS?

Record one answer:



If  the participant answers ‘Yes’ in question D2, continue with question D3, otherwise skip to question D10

The key difference between Survey A (RIA) and Survey B (AMPS) is the use of the skip question in Survey B – this means that anyone who doesn’t understand the words Internet or World Wide Web will not be given the opportunity to answer any of the subsequent questions (about email use for example) and so the structure will likely generate false negatives. This problem is compounded by the fact that, unlike Survey A, Survey B makes no attempt to identify the respondent’s existing awareness of the Internet.

In both surveys the closed structure  inevitably disguises a lot of variation. For example, respondents who want to answer ‘I don’t know what the internet is’ or ‘I’m Not sure if MXit/Facebook/gmail counts’ ‘my friend helped me, does that count?’ or ‘I don’t remember’ might all answer ‘No’ in Survey B. Also notice that the question requires recall (‘ever’ in survey A and 12 months in survey B).

There are thus serious issues of validity for both surveys – you can use the internet without knowing that you are doing so. The question strategy in both surveys requires the participant to understand the word ‘internet’, which is pretty jargonistic. On the one hand there’s a lack of popular understanding of the term ‘internet’ (only 50.8% of SA respondents in the Research Internet Africa survey said that they knew what the word meant). On the other hand those who do know the term may associate it with computers rather than phones or other devices. Not only participants but fieldworkers who administer the survey may also have this association .