CATÉGORIES : Sondages Internet
18 février 2014

Do researchers really know what they are getting into when they use this tool? I think not 

DIY solutions are all the rage these days (DIY home renos, DIY hairdos, DIY jewelry, etc.). Human beings are looking for more and more ways to do things on their own. 

Our industry is no exception. In 2012, a “new” DIY market research tool threw its hat into the ring… GOOGLE!

We at Ad Hoc just had to try out Google Consumer Surveys

How does it work?

To access “premium” content on certain websites, web surfers are asked to answer questions (for a max. of 2 questions per person).

Google claims the following as advantages:

  • Speed (n=1,000 collected in no more than 48 hours);
  • Cost (can be as low as 10¢ per complete) and;
  • Validity.

Without getting too much into the nitty gritty, here are…

Some of the good…

 + Short and sweet! Respondents answer a maximum of 2 questions, which means validity is high and there is less respondent fatigue.

 + Researchers can focus on the insights and spend less time boggled down with the mechanics.

 + It’s Google!  The brand is so well-known that the name itself is seductive.  Not to mention, they also have access to data for all those who use their search engine.

Some of the less good…

 – Short and sweet! 2 questions per respondent is sometimes just not enough.

– Socio-demographic profiles are deduced according to IP addresses and double click cookies. So, for someone like me who visits (frequently), it is entirely possible that they have me tagged as male (!).

– It is a BIG BLACK BOX.  Google does not divulge with which sites they are partners and by osmosis on which site(s) your questions are posted. They do not provide the details on if and how your data is weighted, etc.

– For us, this is probably one of the biggest factors: QUEBEC. For now, questions can only be asked in English and are only posted on English sites.  Upon inquiry, Google confirmed that this tool is representative only of the English-speaking Canadian population. 

If your study is being done in Quebec…forget it!  It is not at all representative.

If your study is pan-Canadian, you are neglecting 16% of the population (Quebec represents 20% of the country and within the province 80% are Francophone).

– Google claims the sample is probabilistic; we see it more like a convenience sample (which means a margin of error cannot be calculated).

 For more information about our test of Google Consumer Surveys, contact me! I’m always up for a good discussion.