KG 01 - See comments inline On Wed, Jun 6, 2012 at 8:00 PM, Rob Weir <robw...@apache.org> wrote:
> Keep in mind that we already have a large recent survey of sorts, > based on Google Analytics data from those who have visited the website > and downloaded AOO. > KG01 - Do we ask for email addresses of people who download. How might we engage downloaders to complete survey? > It won't tell us some of the detailed stuff, like whether they use AOO > at home or at work, but there is more info available here than might > be generally known. > > KG01 - Re-use is great. Can you send along a link to where I can review the specific data we capture? Yes, I think we can eliminate many questions form our demographic questions and pull the data from other sources, such as the download info. > For example: > > - what countries users are mainly from. Can also get detail to the > level of what cities are most often downloading AOO. > - what languages > - what operating systems and versions they are using > - what screen resolution they have > - what browser they are using > - if they found our website from searching Google, what were the most > used search strings > - if they came to our website via a link from another website, what > were the most common "referring" sites > - what social networking sites lead them most to the website > - what pages on the website are most frequently read > - what paths through the website most often lead to a download > > and any of these can be correlated against download conversion rate. > So for example we can look at what % of visitors download AOO based on > country, or language, or OS or browser or whatever. > KG01 - Despite the overlap, I suspect that we will need to include some basic demographic questions in the user surveys to ensure we can correlate the results. For example, if users from a certain geography, or users of a certain role have issues, we need to assocaite their task prioritization and satisfaction ratings against their demographic data. Data from disparate data sources would not support such analysis. > > Obviously this is not a replacement for a survey that looks at the > habits and preferences of the user's in-application behavior. But > this information is "low hanging fruit" that is based on data already > collected. > > KG01 - Indeed, task assessment research is another category all together. > -Rob > > > On Tue, Jun 5, 2012 at 1:51 AM, Graham Lauder <y...@apache.org> wrote: > >> KG01 - see comments inline. > >> > >> On Tue, Jun 5, 2012 at 8:26 AM, Graham Lauder <y...@apache.org> wrote: > >> > > > Hi. > >> > > > > >> > > > Questions relating to research! > >> > > >> > [....] > >> > > >> > > Perhaps the first survey we should conduct is a survey about what > sort > >> > > of surveys our users would respond to. > >> > >> KG01 - Thanks for your feedback and interest in the user research > effort. > >> While I agree we could deploy different types of surveys to gather > >> different types of data, I feel that a survey of surveys might induce > >> premature survey fatigue. > > > > Survey fatique has already set in, that is not a new thing, that is > > recognisable simply by those surveys conducted by SUN. We haven't caused > > that, it is a factor of the modern marketing malaise. The cost of > incentives > > these days, that one needs to hand out to get a significant sampling in a > > timely manner is huge. > > > >> User research, especially surveys, consumes > >> people's time and energy. > > > > Indeed as I myself pointed out earlier in this thread > > > >> Rather, I propose we work from the other > >> direction. If the goal of the research activity is to gather data that > will > >> help us build insight and drive informed design and development > decisions, > >> then we should focus the surveys on the information we need to do that. > I > >> have captured some comments in the wiki discussion page. > > > > Indeed, however if the sample of respondents is ridiculously small, as > has > > historically been the case, then the data is useless. > > > > You cannot use corporate methodologies in an open source environment. > We have > > no ability to offer incentives, we therefore need to make the survey > process > > as pleasant and enjoyable as possible or we need to find out from people > what > > would encourage them to participate. > > > > That requires research, I doubt it will require as big a sample as a UX > survey > > but that is only because there are a limited number of answers needed. > > > > Every good research organisation I have worked with does short surveys > to find > > out what they're doing right or wrong. For the most part they do these > at the > > end of another survey, but that is because the group of respondents they > are > > questioning will probably never do the same survey again. For us the > problem > > has been getting respondents to finish. Lose them once and they won't > come > > back again and we will need to talk to our user community if not often, > at > > least regularly > > > > I would prefer to do things right first time up so people will happily > respond > > to any surveys we need to put out. Remember that there are not only UX > > surveys to be done but Marketing as well. > > > > We know already know two things that get people to complete surveys: > > Brevity and Fun. > > > > If we do a light hearted, quick survey that gives us the reasons that > people > > will participate, I think that's a really good use of resources. > > > > The Surveys already put up are boring, generic and not likely to inspire > > people to complete them. > > > > OOo has a user base in the hundreds of millions a few hundred > completions is > > not a sample. We need 10s of thousands of responses across scores of > > languages, to get a easonable sample. > > > > So first we need to figure out how to get that sample. > > > > Cheers > > G > > > > >