Back to top

By Walt Dickie, Executive Vice President

In June, The Pew Foundation published some very interesting data on cellphone based internet use that packs some worrisome implications for a lot of online marketing research.

Some 88% of U.S. adults own a cell phone of some kind as of April 2012, and more than half of these cell owners (55%) use their phone to go online ... 31% of these current cell internet users say that they mostly go online using their cell phone, and not using some other device such as a desktop or laptop computer. That works out to 17% of all adult cell owners who ... use their phone for most of their online browsing (my emphasis).

Pew also finds that 5% of cell phone owners use their cell phones and some other device equally for online access, 33% mostly use some other device, although they also use their cell phones to get online, while 45% of cell phone owners don't go online at all using their phones.

So, let's do a little back-of-the-envelope calculation: based on these stats, how many cell phone users should we be finding in our general-population online survey samples?

We have to make some assumptions. Pew asks their questions in terms of the respondent's device choice for "most" online access. Let's say that "most online browsing" means something like 75% of all browsing. In other words, let's estimate that the 17% of adult cell owners who "mostly" use their phones are actually doing so for about 75% of their browsing. Similarly, let's assume that the 33% who "mostly" use some other device are actually using their phones 25% of the time. Finally, we'll assume that those who split their online access equally between phones and other devices are splitting the time 50/50.

Using those numbers and Pew's overall cell ownership data, we should expect .88*((.17*.75) + (.5*.05) + (.33*.25)) = 20.7% to show up as using cell phones in a general population sample.

If the people who "mostly" use another device actually use their cell phones for only 10% of their online access, then this proportion would drop to 16.3%. In the extreme case, in which people who "mostly" use cell phone for access do so only 51% of the time, and people who "mostly" use another device actually choose their cell phones only 1% of the time, we would still expect to see cell access making up about 10% of a general population sample.

So, based on Pew's data, the incidence of cell phone access to general population surveys should be in the 10% to 20% range.

If that sounds problematic, the trend data that Pew offers seems even more so.

Pew doesn't give tracking data on "cell mostly" users but they do give data on the growth of cell-based online access overall. It's not unreasonable to assume that the "cell mostly" segment will grow at roughly the same rate as cell access as a whole. Here's Pew's data re-drawn and projected forward.

Image removed.Pew's data shows that phone-based internet access is growing at just about 10% per year. At that rate essentially 100% of a gen pop sample "should" be using a mobile device in about 10 years. If MR samples continue to under-represent people who access the internet via cell phone by 65% to 75%, then the "standard" MR sample sources will shrink by a comparable amount.

Of course, this is a crude estimate - whatever happens, the trend line won't be linear - and ten years is not tomorrow.

But still, assuming that these numbers are anything like inside the park, this implies some big problems. We need to know what is keeping cell users away from online MR surveys, and we need to find ways of changing our approaches to make our research more amenable to mobile access.

Pew's report doesn't directly address the question of how to do this, of course, but it does have some strong hints about what might be involved in the reasons given for choosing cell phones as web access devices.

"Cell mostly" users say that their cell phone is a "simpler, more effective choice for going online" compared to other available options (18%); 7% say that they "do mostly basic activities when they go online"; and 6% "find their cell phone to be easier to use than a traditional computer."

Would these people consider marketing research surveys simple and basic? How long can an online activity take and still be simple and basic. How complex? What kind of engagement can be involved? Is going through a battery of a dozen or so attributes and rating each one on a 1-to-something scale either simple or basic? Is viewing a succession of concept statements with accompanying images - over a wireless connection - simple and basic?

We know from many sources that cell phone use is dominated by short "sessions": a quick text message, a visit to Google Maps for directions, checking Yelp to find a good restaurant nearby, a fast check of incoming email. This isn't to say that people don't sustain long periods of engagement - playing Angry Birds on the bus to work, reading the news, even reading a novel using the Kindle app. But many aspects of cell phones, from screen size to data plans to spotty coverage urge short bursts of use that generally don't mesh well with anything resembling even a 10 or 20 minute questionnaire.

Although I don't know for certain where the cell phone users that are missing from our general population samples have gone or why they're not in our samples, I do have a hunch. They're not in our samples because they're not even in our world, which was built, sample panel on sample panel, river source on river source, on a PC/laptop model of online engagement and interaction. The "cell phone mostly" web users have simply moved on to something simpler and more basic.

I've blogged about this issue before and will again, I expect. There is a conflict between "marketing research" understood as "the collection of data designed for statistical analysis tailored to the needs of standardized corporate decision making procedures," which is what drives a major portion of client MR activity, and marketing research defined as "collecting as much data relevant to marketing issues from as many sources and in as many modes as is possible via available technology."

The incorporation of MR into corporate decision-making happened during an era when the technology at hand - phone and mall interviewing, then online surveys - created a certain style of research that demanded high focus and a fairly large time commitment from respondents. That kind of research is still quite possible, but its days may well be numbered.

By Walt Dickie, Executive Vice President

In June, The Pew Foundation published some very interesting data on cellphone based internet use that packs some worrisome implications for a lot of online marketing research.

Some 88% of U.S. adults own a cell phone of some kind as of April 2012, and more than half of these cell owners (55%) use their phone to go online ... 31% of these current cell internet users say that they mostly go online using their cell phone, and not using some other device such as a desktop or laptop computer. That works out to 17% of all adult cell owners who ... use their phone for most of their online browsing (my emphasis).

Pew also finds that 5% of cell phone owners use their cell phones and some other device equally for online access, 33% mostly use some other device, although they also use their cell phones to get online, while 45% of cell phone owners don't go online at all using their phones.

So, let's do a little back-of-the-envelope calculation: based on these stats, how many cell phone users should we be finding in our general-population online survey samples?

We have to make some assumptions. Pew asks their questions in terms of the respondent's device choice for "most" online access. Let's say that "most online browsing" means something like 75% of all browsing. In other words, let's estimate that the 17% of adult cell owners who "mostly" use their phones are actually doing so for about 75% of their browsing. Similarly, let's assume that the 33% who "mostly" use some other device are actually using their phones 25% of the time. Finally, we'll assume that those who split their online access equally between phones and other devices are splitting the time 50/50.

Using those numbers and Pew's overall cell ownership data, we should expect .88*((.17*.75) + (.5*.05) + (.33*.25)) = 20.7% to show up as using cell phones in a general population sample.

If the people who "mostly" use another device actually use their cell phones for only 10% of their online access, then this proportion would drop to 16.3%. In the extreme case, in which people who "mostly" use cell phone for access do so only 51% of the time, and people who "mostly" use another device actually choose their cell phones only 1% of the time, we would still expect to see cell access making up about 10% of a general population sample.

So, based on Pew's data, the incidence of cell phone access to general population surveys should be in the 10% to 20% range.

If that sounds problematic, the trend data that Pew offers seems even more so.

Pew doesn't give tracking data on "cell mostly" users but they do give data on the growth of cell-based online access overall. It's not unreasonable to assume that the "cell mostly" segment will grow at roughly the same rate as cell access as a whole. Here's Pew's data re-drawn and projected forward.

cell owners graphPew's data shows that phone-based internet access is growing at just about 10% per year. At that rate essentially 100% of a gen pop sample "should" be using a mobile device in about 10 years. If MR samples continue to under-represent people who access the internet via cell phone by 65% to 75%, then the "standard" MR sample sources will shrink by a comparable amount.

Of course, this is a crude estimate - whatever happens, the trend line won't be linear - and ten years is not tomorrow.

But still, assuming that these numbers are anything like inside the park, this implies some big problems. We need to know what is keeping cell users away from online MR surveys, and we need to find ways of changing our approaches to make our research more amenable to mobile access.

Pew's report doesn't directly address the question of how to do this, of course, but it does have some strong hints about what might be involved in the reasons given for choosing cell phones as web access devices.

"Cell mostly" users say that their cell phone is a "simpler, more effective choice for going online" compared to other available options (18%); 7% say that they "do mostly basic activities when they go online"; and 6% "find their cell phone to be easier to use than a traditional computer."

Would these people consider marketing research surveys simple and basic? How long can an online activity take and still be simple and basic. How complex? What kind of engagement can be involved? Is going through a battery of a dozen or so attributes and rating each one on a 1-to-something scale either simple or basic? Is viewing a succession of concept statements with accompanying images - over a wireless connection - simple and basic?

We know from many sources that cell phone use is dominated by short "sessions": a quick text message, a visit to Google Maps for directions, checking Yelp to find a good restaurant nearby, a fast check of incoming email. This isn't to say that people don't sustain long periods of engagement - playing Angry Birds on the bus to work, reading the news, even reading a novel using the Kindle app. But many aspects of cell phones, from screen size to data plans to spotty coverage urge short bursts of use that generally don't mesh well with anything resembling even a 10 or 20 minute questionnaire.

Although I don't know for certain where the cell phone users that are missing from our general population samples have gone or why they're not in our samples, I do have a hunch. They're not in our samples because they're not even in our world, which was built, sample panel on sample panel, river source on river source, on a PC/laptop model of online engagement and interaction. The "cell phone mostly" web users have simply moved on to something simpler and more basic.

I've blogged about this issue before and will again, I expect. There is a conflict between "marketing research" understood as "the collection of data designed for statistical analysis tailored to the needs of standardized corporate decision making procedures," which is what drives a major portion of client MR activity, and marketing research defined as "collecting as much data relevant to marketing issues from as many sources and in as many modes as is possible via available technology."

The incorporation of MR into corporate decision-making happened during an era when the technology at hand - phone and mall interviewing, then online surveys - created a certain style of research that demanded high focus and a fairly large time commitment from respondents. That kind of research is still quite possible, but its days may well be numbered.