person Guy Potter access_time November 29, 2017

Keep it short and get to the point. We talk about how to ensure quality data

So much has been written about the quality of market research data that I do not want to reiterate it all here, save to say (for those of you that care about these things), that we have conducted matched testing against other data sources, across time and to published and verifiable statistics and have always been a few percent away from whatever mean we were aiming for.  A quick Google search on poll performance at recent elections / referendums puts us ahead of the pack.  If you are interested feel free to get in contact with us for chapter and verse.

So what have we learnt along the way that helps us ensure our data is quality?  Several things.

Survey Length

Keep it short, and when we say short we mean short, not 42 questions that can be done in under 15 minutes, but five or ten questions, that can be completed in a few minutes.  We have found people drop off at two main stages in a survey.  At the very beginning, when they discover what the survey is about (financial brochures anyone?) and then as the length of the survey continues to grow.  This is such an obvious comment, but can be lost sometime on some market researchers.   Attention spans / demands on our time mean survey length must decrease.  Get to the nub of the issue quickly!

Keep it simple

Another obvious comment – More complex questions, grids, questions with many sliders, any questions that challenge respondents and make them think should be avoided, they cause drop outs and challenge people to think about how to interact with the page instead of how to answer the question.

Don’t bounce people out

Routers, the secret no one really talks about.  Their purpose is to take you to a survey and then if you do not qualify for that survey (which always seems to be the case) take you to another, and then another and another.  “The Terminal” was a great film starring Tom Hanks as a man stuck in limbo at an airport.  Perhaps someone should make a sequel called “The Router”, starring an exasperated person stuck in a series of questionnaires that they almost but never qualify for, though this may not get Hollywood funding.

Profiled respondents

Don’t re-ask the obvious

When (if) you sign up to a panel you get asked a profiling questionnaire that will usually be about 80 questions long, ranging from gender / age to how long you spend playing computer games a week and what kind of car you drive.  After the arduous task of completing the profiling survey your details are in a sql database somewhere, ready for you to be extracted and invited to do a survey.    All well and good.  However, if you are a male gamer, aged 28, who plays games for 4 hours a week, you are asked to reconfirm these details at the beginning of every survey.  I can understand the number of hours played a week, but gender and age are always asked again.  Shouldn’t software remember those details and automatically append those to a survey?  Or if they can change automatically re ask them at the appropriate time interval.

We at MaruUsurv remember these details about our people (and though many are regularly  refreshed) will not re-ask these every time we do a survey, it just bores people.



Guy Potter is maru / usurv's head of research and clients.