Simon Chadwick talks to Paul McDonald of Google Consumer Surveys
SC: We’ve read in the press that the primary impetus for Google Consumer Surveys was the result of a search on how to monetise your publishers. Is that fair?
PM: Basically, a friend of mine, who’s the lead engineer on my team, and I had a couple of conversations about how to help publishers monetise their web content. We were thinking how to incentivise people to share content and we hit upon this idea where you could pay for the content not with money but with time. The idea, at first, was to have people do things that humans but not computers were good at doing, for example, labeling Google searches over images. We did a test, we tried the market research, public opinion polling type of questions, and it turns out that users liked to do the opinion-based work and not the human computation tasks.
We have millions of advertisers that we work with through our AdWords programme, mostly small- and medium-sized businesses that don’t have access to professional-quality market research – it’s either too expensive or time consuming – and so we thought we could make something that’s professional quality and gets answers that reflect the population, but in such a way that these businesses can do market research for their own products. That’s when we came up with Google Consumer Surveys.
Is the real business publisher monetisation or does Google now regard Google Consumer Surveys as a real business?
Google has this maxim ‘focus on the user and all else will follow.’ So we really focus on the user experience, the respondent experience in this case, we give them their content quickly and without a lot of burden on them, plus a unique way to pay for this content, which turned out to be a much better experience than a pay wall or other form of payment. That is why you see some restrictions on the research side as we wanted to make it so that a question could be answered within 15 to 20 seconds, so the user could dig into the content they wanted to access as quickly as possible.
One thing that surprised people was the degree of thought that you put into the design of the questioning sequences and the reporting. Tell us about your research into understanding respondents and what good research would be tagged as being.
At Google we do a lot of market research, so we have a group of folks who focus on research for our products, or our ad business, we’re doing high-level research for some of our ad clients as a value add, so we had a great group of folks testing the product as we built it.
We built a tool that we wanted to use. The most important thing was speed, for the respondent – it’s quick to get access to the content; for the researcher, it’s fast to set up and field a survey, and generally you get responses back in 24 to 48 hours. Then we understood the power of being able to ask a question to random people and getting responses back and being able to iterate over that. Maybe someone had an issue with the way you worded your question– you just go back and 24 hours later have another answer.
Whether or not that lined up with the original one, you can understand what the differences were. And then we wanted to make it easy and fast and intuitive to cross-tab and segment the data, so we spent an enormous amount of effort and time perfecting the experience. It’s fast and simple to use and if you don’t want to use our interface you can export the raw data and use tools that you’re used to like SPSS or Excel or something else. We use it to do our own research, and we’re seeing, within Google, an incredible uptake of this tool in the product-development process, which validates what we’ve been doing, if nothing else as a tool for Google to build better products.
What about the question limitation? You’re essentially fusing little sort of surveylettes together.
That is a major limitation, and what we hear most from researchers in what they would like to be improved. If you look at how users are using the internet and the devices that they’re using, you can conclude that the way that research is done online today can’t go on the way that it is. We have these long-form surveys that take 30 minutes to complete, there are a lot of different question types, difficult types to comprehend like large grid questions, and respondents are getting fed up. They’re sick of getting solicited to respond to surveys, and there’s a lot of research that says the longer the survey, the more people are straight-lining through, the less accurate the data is, the lower their response rates.
Currently most online research is from desktop computers, but mobile is approaching fast and when we started designing the service we wanted the form-factor of the survey question to fit on the mobile device. So you can run it on a desktop or on a mobile phone and it fits on the screen perfectly. The design was a mobile-first design, it just turns out we went to the desktop first when we first launched. But we focus on mobile research, that’s what we see as the future, and you can’t have long surveys or multi-question surveys on mobile devices, because people don’t have the attention span, they don’t want to do it on their mobile device. So there has to be a shift in attitude and methodology from the research community to understand how to appropriately target and survey folks using mobile phones. This has us positioned to do mobile quantitative survey-based research.
How has the product been received by researchers?
There’s a lot of interest because Google is a big player online and there’s a precedent for opening up professional-quality tools to the small- and medium-sized businesses and slowly working to the Fortune 500 companies and bringing them onboard. I think the research community is interested and wants to find ways that they can leverage our platform to provide their clients research.
Then there’s start-ups, the community that Google most relates to, that is amazed by the power that they have. They can ask questions to random people across the US and get answers to questions they’ve always had. Usually technologists tend to go from the gut, trying to build products that they want to use or that they think are right for their audience. Now they can do gut-checks, they can understand incidence rates, their populations, and test their logos or their website design. So, they’re impressed and amazed and are repeat users.
I think there’s some concern with more traditional researchers – you’re Google, you’re getting into the market research business, to build up a huge team, steal client-side research and take the business away. That’s just not really what Google does. Google tends to try to build tools and platforms for others to utilise. So just like what you’re seeing with apps, will happen with this tool as well, where the agencies can take data and insights that we provide and use this tool just like they would any other. They can take the data out and use it how they’d like and provide more value on top of that to their clients. So I think they have misplaced concerns.
Google sits on enormous amounts of behavioral data. Is there not the potential to go beyond inferring demographics to inferring behavioral patterns, things that people reveal in terms of their online behavior, and put all of that together as a very powerful, integrated data and insights tool?
We have to put the user first. If combining users’ data or search data or browsing behaviour, with survey data leads to a positive benefit for the end user, better products or better advertisements for them – I think that makes sense, and we may do some of those things. But just to collate data and put it all in one place for people to use as a database is not interesting to us. I think you will see that we will use some of the data about users’ browsing behaviour just as we do to infer age and gender, we can also infer interests and other things based on the sites that they’re browsing. But combining a lot of behavioural data, at least right now, isn’t in the cards unless there’s a real reason to do so.
What is the make-up of clients that are coming to you?
The majority are small- and medium-sized businesses that are our average customers that we have relationships with are really our initial target. There is a small but growing number of Fortune 1000 companies using our tools, and I think they’re finding new and interesting ways to use that data. A lot of them are doing incidence testing. So with a single question, two questions, you can understand the percentage of the population that does one thing or another, and you can target them by screening other folks out. And being able to get to those small-incidence populations and getting answers back very quickly – it’s one of the ways that the larger companies are understanding what we can do and utilising that. Several research companies, large research firms, have used it.
How rapidly do you think you’ll expand this out into different markets?
There are several challenges. One is, we have to build both sides of the marketplace at the same time: publishers to run the surveys on, researchers to run our field surveys. There are also some regulatory issues that we have to resolve; then globalisation and translation. What you’ll see is English-speaking countries first – places like Canada, the UK, Australia, Singapore, Hong Kong – and then European regions, and then trying to get into areas that are harder to reach like sub-Saharan Africa or other African nations.
One of the reasons why we’re focusing on our mobile strategy and why it really started with the mobile form-factor first, is that it would be easy for us to go into some of these countries that really only have internet access through mobile phones.
Simon Chadwick is Managing Partner of Cambiar and Editor in Chief of Research World. Paul McDonald is project leader of Google Consumer Surveys.
