This will be remembered as a watershed year in South Africa when, after more than 200 years, printed newspapers were surpassed by digital computers, tablets and cellphones as the primary source of news for most of the country’s inhabitants. Over the next three years, classic print ad spend will lose R2 billion and digital spend will more than triple as ad spend always follows eyeballs. It’s no wonder then, that two of South Africa’s largest newspaper publishing groups, Times Media Group and Independent Newspapers, changed hands earlier this year.
South Africa is believed to have the best media research in the world. The All Media and Products Survey (AMPS) has a massive sample in excess of 25 000 respondents annually that is representative of the population. But while the survey and sample may be sound, it is unfortunately conducted among humans. And we are notoriously bad at remembering how many beers we had in the past week, when we last read the Sunday Times, or that cellphone browsing is also accessing the internet. Researchers refer to this as ‘respondent confusion’. This may be combined with several other factors including:
– That many brand and media choices are founded on emotional drivers, which can never be uncovered in a verbal survey conducted by a complete stranger;
– Readership is a low-interest category and notoriously difficult to measure. There are over 160 identified factors that influence claimed readership; and
– All respondents will tell interviewers exactly what they think the interviewer wants to hear. This ‘halo effect’ is especially problematic in LSM 4 and below.
We then get research data that is not only unbelievable, but also impossible, and therefore must be manipulated in order to become a more accurate representation of reality.
Adapting the numbers
If your media planner is not adapting the numbers to suit the human condition and ever-changing market conditions, you may as well save yourself three percent and get a computer optimisation programme to plan your next budget.
Let’s begin by looking at the base numbers from AMPS without applying any changes. Now bear in mind that unlike media owners, who have billions of rand worth of ad spend at stake, I have no vested interest in either newspapers or the internet. I am simply searching for something closer to the truth; figures that can be corroborated by other sources, preferably audited ones. Most importantly, advertisers are booking ads for next week and they deserve to know what audiences and return on investment they will get.
So let’s apply reality checks to the AMPS numbers above for each medium and, where the numbers are patently wrong, we will apply calculations that change the data to match audited sources. Unless stated otherwise, the source of the data is AMPS 2012.
A total of 86% of homes have at least one radio, and when we add listenership via taxis, train and bus stations, cellphones, public places, offices and neighbours’ houses/communal sets in the rural areas, we see no problem with radio reaching 96% of adults in a four-week period.
TV household penetration has grown consistently by around 1.5% annually for the past 15 years, and in 2011 television households overtook radio homes and now stand at 88%. Instead of using TV licences to verify AMPS penetration, we can use them to prove that less than 20% of people pay theirs. When we add out- of-home viewership at other people’s homes, bars, clubs, shebeens, restaurants, offices, taxi ranks, airports and hotels, it makes perfect sense that TV could reach another six percent and arrive at 94% monthly reach. Increased household penetration will be off-set by reduced viewing at friends so we will keep penetration constant.
Sure there are over 52 million sim cards in South Africa, but as at the mid-point of the AMPS 2012 fieldwork (Jan 2012) there were 29 million adults or 83% with cellphones. Cellphone growth for the past three years has been around 170 000 new users each month, or two million a year. The penetration of adults in March 2013 will be
31.5 million or 87% of the 36 million adults. INCREASE 4% FROM 83% TO 87%
With over 16 000 sites countrywide we have no issues believing that 79% of adults would have seen one in the past four weeks.
For print we have Audit Bureau of Circulations (ABC) figures. So we have audited numbers of newspapers sold to compare to AMPS readership. Forget the fact that Q4 (fourth quarter) 2012 newspaper numbers declined by 144 000 versus Q3 2012, or more versus last year; we have more meaningful calculations to do before we begin reducing readership for declines in circulations.
A far more important calculation is to establish the true readers per copy (RPC) calculation. This refers to the claimed AMPS readership of newspapers divided by their audited circulation. As seen in the table (on the left) for daily newspapers, these show that RPC varies considerably, but on average there are 6.2 readers per copy of daily papers.
This is for dailies. The biggest weekend newspaper, the Sunday Times, has 7.8 RPC, and Bona has over 30 RPC. But for this article, let’s keep it simple and just stick to dailies. So, 89% of copies bought are “bought by self or a family member”, but there are only 2.5 adults per household in South Africa. Have you ever passed your newspaper over the fence to your neighbour? If the answer is no, you have to read further and change the figures. Actually to get to six readers, your neighbour would also have to pass it to her neighbour as well!
But what about out-of-home readership? Publishers always claim this pushes up the average but when you see the numbers required, you will agree that this is also totally impossible. The other 11% ‘no family’ copies would then need in excess of 36 readers per copy to get back to an average of over six. This means that not only do all 16 people in the taxi have to read it, but so do another full taxi of 16 people have to read the same copy as well.
You can’t have four friends coming over to your house to read it, or else 89% could not claim it was “self or another household member’s copy”.
There are other possibilities in that newspaper reading households are not average; for example, they have more adults in them.
So I produced a frequency distribution curve with an average of 6.2 RPC. This appears below and just reinforces that this readership figure is statistically impossible. When looking at this data remember that 74% of households have fewer than four adults and 92% have fewer than six. So, to get to the overall average of over six, the centre of gravity, at around 20%, has to be at seven RPC, but only 8% of homes in South Africa even have this many adults. This proves that these AMPS RPC levels are not simply overstated, they are impossible to achieve.
Echo (global research firm) has developed another more realistic distribution, that would at least be statistically possible. With an average of 2.5 adults per household, one would think the centre of the curve would be in the two to three range, but readership of out-of-home copies pushes the average up to 4.2.
So, 4.2 RPC is 68% of 6.2 RPC. This is the ratio by which we must reduce newspaper readership, which would mean that one in three South Africans reads printed newspapers versus one in two. A massive but realistic decline.
A corollary and commonsense validation of this follows: 42% of South Africans live in small towns or villages of fewer than 8 000 people, where newspapers are hardly circulated. We cannot get the ABC certificates by region anymore, but they showed, for example, that in Mpumalanga the Sunday Times had 44 RPC. Let’s say 10% of these 42% of rural folk do read newspapers, to get back to the 50% average of the total population would mean that 80% of the 58% urban population would have to read newspapers.
Do you see 8/10 cars stopping to buy newspapers from vendors, or stopping at the garage shop to buy a paper or subscribing? I think not.
REDUCE NEWSPAPER READERSHIP BY 32% FROM 50% TO 34%
To keep this article brief, we will just reduce magazine readership by 29%. To do this precisely we would have to examine the frequency of distribution for every single print title on AMPS. We are merely trying to show data adjustment principles here and encourage debate and data manipulation, not do an actual media plan.
REDUCE MAGAZINE READERSHIP BY 29% FROM 48% TO 34%
The AMPS code for the past four-week internet access records this as 7.8 million people or 22% of adults. There are a registered seven million Facebook users in South Africa. If this figure were correct, 85% of all adult internet users would be on Facebook, but common sense tells us this is not the case.
The major problem with AMPS internet research stems from respondent confusion, among certain sectors of the population. They consider the ’internet‘ the same as accessing from a computer. In the cellular section of the questionnaire, there are questions regarding mobile surfing, ‘Facebooking’ and downloading. If one combines this mobile access with computer access, one arrives at a figure of 10.2 million (29%) or 12.3 million (35%), depending on what mobile activities one includes.
To determine which figure is closest to the truth, we have to look at Effective Measure’s (EM) research, which all major sites and members of the DMMA use to determine website visits. But these measurements are based on ‘time stamped devices’ not people, so a short explanation of audience currency exchange rates is required.
Every device that accesses the internet has a unique identifier known as an IP address. When you visit a website it records your IP address. This is not a static address, however, so is not an audit of devices, since every time a person clears their cache (removes cookies), they will be assigned a new IP. Every different IP address that accesses a site, is known as a UB (unique browser). Advertisers are interested in reaching people (P) not these machine identifiers, so we need a formula to get from UB to P. This is P = (UB x 0.77). In other words, because people access sites on multiple devices, actual people are around 77% of UB figure. This formula could be the subject of its own article, but simplistically it is derived from the EM panel of over 8 000 respondents and has been corroborated by the Echo multiple access channels AMPS analysis.
Another problem is that UB also include children, so we have to remove under-15-year-olds and then we arrive at the ‘best fit’ figure between EM and AMPS. As at the first quarter 2012, the internet adult penetration that best matched the 14.7 million UBs reduced to
11.1 people was a 29% penetration or seven percent above the AMPS internet (PC) penetration of 22%.
But the internet is the fastest growing medium in history and UBs have grown by 37% year on year. For February 2013, they now stand at 19.98 million monthly browsers, which equates to 15.4 million people, less the 1.25 million kids means 14.1 million adults. So the internet currently reaches around 39% of adults in the country.
INCLUDE MOBILE ACCESS TO INCREASE TO 29%. THEN INCREASE 37% TO GET TO 39%.
This is small and we have attendance figures. NO CHANGE
So after these adaptions, we arrive at what Echo believes is a better and more accurate reflection of ‘reality’ or ‘the truth’ regarding South African media penetration in March 2013.
Everybody loves AMPS, but we cannot accept the figures blindly. We need to make changes to the data that match other sources based on audited printed copies or IP addresses, not tainted by human emotions. In most cases the audiences of titles have remained static – they are just migrating from the single printed word to a broader repertoire of digital ones.
This data adaption needs healthy debate and input from all. I am just starting the ball rolling by saying to the industry, you can no longer accept impossibly high readership. Interrogate the data until you tame it, then manipulate it.
Every person in the industry, from media owners to agency people, owe this quest for greater accuracy and truth to our advertisers who trust us to do what’s best for them as they fund the R20 billion ad spend, without which there are no media, jobs or advertsing industry. n
Peter Langschmidt has spent over 30 years in advertising, media and marketing. In 2006 he won the prestigious Media Innovator of the Year award. For more information, visit echo.za.com or mail Peter Langschmidt at firstname.lastname@example.org