Yes, handset WiFi is important - but dodgy surveys don't help make the case

Like many smartphone owners, I use WiFi quite regularly - although it's almost entirely for use at home. I find that iPhone users tend to be the most WiFi-savvy group, especially if they have come from the ranks of Mac laptop owners.

That said, I generally don't use WiFi hotspots on my phone - if I'm at a hotel or airport, I'd rather use my laptop if I've got 30-60 minutes. And if it's only 5 minutes, I'm certainly not going to sign up & pay money via a small-screen browser.

The solution is to have various automated auto-login options for devices, either via some for of embedded client (yes, even UMA has a role here), or an aftermarket app like DeviceScape's or Boingo's. On the other hand, this happens so infrequently, I'm certainly not prepared to have a separate paid-hotspot subscription for my phone - I don't even have one for my laptop, which would certainly be a higher priority.

Over time, this will continue to evolve, especially given roaming rates and the desire of operators to start offloading traffic from their macro networks.

On featurephones, it's pretty pointless having WiFi for all sorts of reasons - data consumption is almost zero, so the offload argument doesn't fly in particular.

In general though, I'm definitely in favour of smartphones having WiFi - it's quite possibly the case why the BlackBerry Storm isn't doing as well as many hoped, especially outside the US.

But what I do have a problem with is dodgy surveys to try and drum up interest. Honestly DeviceScape - what's the point in surveying your own user base (ie loyal WiFi smartphone users) and asking them if they like WiFi, and want it in more phones? Quite frankly, I'm curious about the 14% of people who *don't* want more WiFi in handsets, given the bias inherent in the survey base.

This just underscore a point I've made before. Any survey which is clearly being done specifically for PR purposes is generally not worth the paper it's written on. My standard expectation is that "90% of respondents don't know what they're talking about, and the other 10% lie". And that's where it's actually a representative sample, rather than "we asked people we could find cheaply and easily". In political polls, there is a branch of science called "psephology" that analyses election results. That's also an imprecise area - but at least it's treated with some level of rigorous oversight.

Surveys which are being done for real, internal, management information and product development purposes, but which have some data of external interest as well, can be a different story. Even then, you should always look carefully at the selection of the sample, and the "neutrality" of the question wording, as well as the number of responses.

Maybe there should be something like a "Technology User Survey Council" which sets down best-practice recommendations for sampling and questionnaires?