FEATURE1 July 2010
FEATURE1 July 2010
Everyone seems to want to have a go at the stuff that market researchers used to do. And some of them are offering to do it for free. Robert Bain surveys the competition.
Market research has never stood alone as a way for businesses to try to understand their customers and make better decisions. But with more and more ways to collect, crunch and make sense of information (a lot of which result from new technology), research seems to be facing more competition than ever, often from unexpected quarters. An enterprising client could quite easily cobble together their own solution by doing bits of it themselves, turning to the world of social media for other parts, and maybe even asking the bookies for help. Is this the dawn of a great era of openness, creativity and cross-fertilisation? Can researchers look forward to working with exciting new tools, new colleagues and exploring new business opportunities? Or is MR under threat from all sides? We decided to get to know some of these new market researchers (although they mostly wouldn’t call themselves that) to find out where the action is and what the future holds.
It’s easy to see why DIY surveys have got market researchers worried – anyone on the clientside can now use free tools such as Vizu, QuestionPro or SurveyMonkey to bypass MR agencies and save money.
“The market research department is frustrated that some of the marketing department are running their own customer surveys”
Dave Goldberg, SurveyMonkey
And they are doing so in droves. Dave Goldberg of SurveyMonkey says five million people have been using his firm’s free and paid-for tools to ‘do the monkey’ (as it has become known to some), gathering more than 25 million responses per month. But even though Goldberg claims this makes SurveyMonkey “the largest survey business in the world” he still doesn’t see it as competing with market research agencies. He told Research: “We’re actually opening up a whole new market. A lot of our customers could never afford to hire a research firm – they’re academics, they’re schools, they’re non-profits, they’re local government agencies, they’re small businesses. So that’s a big proportion of our customer base who get a lot of value out of our tool, who really have no alternative.”
At the same time he knows that not everyone welcomes this democratisation of research. “Sometimes in the big companies where we have customers, the market research department is frustrated that some of the marketing department are running their own customer surveys. But it’s not anything that we can control except to just try to help people make better surveys with our tool. Once they’ve decided they’re going to use our tool they’re going to do whatever they want with it.” In this spirit of helping people do better DIY research, SurveyMonkey recently hired Phil Garland, formerly of SSI, as its head of methodology, to help develop the survey and question templates that it provides to its users.
Steve Gatt, who heads insight at the Volkswagen Group, says that DIY has its roots in the shortcomings of conventional MR, and that the industry needs to do a much better job of responding to it. “DIY would have come along in any case, but I would have hoped they’d be more competitive in the market research industry and actually offer alternatives rather than having a choice between either a DIY survey or a full-blown market research study as you have at the moment,” he said.
The ball, it seems, is squarely in market research’s court on this one. DIY isn’t going anywhere, so agencies can either try to help hone the tools and train the clientsiders who use them, or seek to demonstrate that there is a better way. If, that is, that they can still convince clients that what they do is worth paying for.
Social media analysts are becoming bolder by the day in pitting their methods against traditional market research. A Carnegie Mellon study published this year found that analysis of Twitter data could rival polls as a way of tracking public opinion over time. In the UK, Tweetminster, a site that aggregates tweets written by or about politicians, conducted an experiment to see if it could predict the overall result of the general election using on Twitter buzz. The answer, in short, was yes. Its prediction of the share of vote (Con 35%, Lab 30%, Lib Dem 27%) had an average error of 1.75 points – on a par with Ipsos, Populus and Harris, and closer than several of the other pollsters.
“What happens on Twitter doesn’t stay on Twitter – it influences what happens on mainstream media”
Alberto Nardelli, Tweetminster
Alberto Nardelli, Tweetminster’s CEO, says trying to gauge buzz is a natural step for a company rooted in social media. “We’re constantly experimenting in terms of technology, in terms of how you measure influence and how you measure reach,” he told Research. “We wanted to find out if there’s a correlation between what’s happening online in social media and something very tangible like an election result.” The election experiment showed how social media analysis, which until now has generally been placed in the qual camp, can go beyond ‘netnography’ and be used to make quantitative predictions – but Nardelli believes it only scratched the surface.
Those who claim social media analysis isn’t proper research are missing the point, he says. “It’s about doing something different. It’s not about representativeness, it’s about making predictions using a different approach which is trying to answer different questions.”
Critics might also point out that Twitter is a narrow sample of specific types of people, but Nardelli insists it is more significant than that because of the way Twitter relates to the rest of the internet. “What happens on Twitter doesn’t stay on Twitter,” he says. “It influences what happens on mainstream media and in other places – the majority of tweets contain links. It’s not like something like Facebook, it gives you an insight into what people are thinking and feeling.”
The firm is now looking at analytics and data-based services that it could derive from its work and sell to media owners or other buyers. And although Tweetminster’s name would seem to tie it to the world of politics, Nardelli talks persuasively about possibilities in the area of marketing analytics. “It’s not an area we exclude,” he says.
“The problem with panel in general is that we look at it from the perspective of what the market researcher wants”
Sean Case, Peanut Labs
Social networks now account for nearly a quarter of time spent online, and researchers cannot ignore them as a way of finding people,
especially young people. With users sharing huge amounts of detail about their lives online, social networks represent perhaps the most striking example of a free service offered in exchange for data. But that relationship can be fraught – Facebook found itself looking clumsy at best and sinister at worst when it has sought to change the rules on what its members share and how. However, the network has so far not done much to make its 400 million active users available for market research purposes, and while LinkedIn offered a b2b sample service for a short time, it was dropped after just eighteen months.
But others have turned survey sampling into a successful business. Habbo, the teen network run by Finnish firm Sulake, provides samples of its millions of regular users to research firms, as well as running its own global youth study, which reaches 100,000 respondents.
Peanut Labs, which grew out of social network firm Xuqa, has built a business sourcing survey sample from apps and games that run on social networks. The owners offer incentives to their users in the form of virtual currency, thus keeping people engaged and monetising their app at the same time. The reference to peanuts in the company’s name is increasingly ironic, as the market for virtual goods is already worth billions of dollars worldwide and growing fast.
It all raises the question of whether traditional panel providers have missed a trick. Peanut Labs vice president Sean Case, who came to the firm from GMI, says traditional panel companies see the firm as “the direction they need to go, but they don’t know how to get there yet”. He told Research: “I think the problem with panel in general is that we look at it from the perspective of what the market researcher wants. We’ve talked a lot about relationships and panel management and respondent engagement, but at the end of the day commercial panels only care about one thing, and that’s getting the completes that they’ve sold, and because of that I think we’ve abused a lot of our panels, which is why we have terrible attrition rates.”
“If you look at a traditional panel, most of them are going out and buying remnant ad inventory off giant ad networks for as cheap as they possibly can to drive people into their panel. You’re going to get a certain type of person that’s willing to do that, and that’s very different from the kind of person that’s willing to engage in market research as an activity as part of their overall social networking experience.”
The appeal of social networks is not just in reaching an audience that is decreasingly responsive to panels, but also in the ability to engage an audience in their own space and on their own terms – which happen to be quite favourable to all concerned.
This year’s Research Conference took place in the run-up to a general election, and concern was duly raised about the public’s trust in opinion polls. Many people, it was suggested, had more faith in Paddy Power than in market research.
The bookmakers seem happy enough to play along with this – both Ladbrokes and Betfair had a go at predicting the make-up of the new parliament. As it turned out they made roughly the same mistake as the pollsters (overstating Lib Dem support at the expense of Labour) only more so, and with some overstatement of the Tory vote thrown in.
So it seems the public can believe what they like – pollsters don’t need to worry too much about the bookies for now. Having said that, it isn’t such a crazy idea to look to the world of betting, especially exchanges such as Betfair where punters are matched against each other, without a middleman deciding the odds.
Since 1988 the University of Iowa has run the Iowa Electronic Futures Market as a not-for-profit research initiative. Participants can buy and sell contracts based on the outcome of an event. For example in the 2008 presidential election, participants could buy ‘shares’ in either a McCain or Obama win for between $0 and $1, which they would be able to cash in for a dollar if that candidate won. As prices fluctuate through the campaign, participants are free to trade their shares with each other = more than a quarter of a million dollars’ worth were traded on the exchange during that election. Predictions based on activity on the market rival those produced by opinion polls.
One of the firms to apply this idea to market research is BrainJuicer, with its Predictive Markets tool, which is used to test product ideas before launch. The company says the technique reveals clearer differences between ideas, making it possible to spot the breakthrough concepts that are ahead of the pack. Another advantage is that people like betting. Even if they’re not playing with real money, it has to be more fun than a straight survey.
Ray Poynter includes prediction markets as an example of “new MR” in his forthcoming book on online and social media research. In response to suggestions at the Research Conference that it would be a shame for market research to be beaten by Paddy Power, Poynter asked the very sensible question “Why would that be a shame?” If it works, run with it.
“A data-rich society will give the survey creators and focus group holders much better data to start with and much better questions to ask”
Doc Searls, co-author of the Cluetrain Manifesto
Ever wanted to know exactly how long you slept for, or how deeply? A Massachussetts-based firm has the answer: it’s called Zeo and it consists of a headband and a bedside display that measures electrical signals from your brain. You can even upload your sleep data to the web and keep an online journal. It’s not necessarily the type of information that most market researchers are itching to get hold of, but it’s just one example of ways customers are using technology to manipulate and manage their data in new ways. But what happens to research when customers take control of information?
Doc Searls, one of the authors of The Cluetrain Manifesto, has more recently been working on ‘vendor relationship management’, which he sees as the flipside of customer relationship management. The idea is to create software tools that allow customers to manage their relationships with vendors, sharing what information they choose to their own advantage. Searls points to tools like Foursquare, which lets you use a smartphone to track where you go and uses the data as the basis for social networking. He says we’re seeing the emergence of “a new ecology of data” in which the burden of collecting and managing information is shared by individuals, the devices they use and the companies and services that are interested in crunching the data and using it in new ways. “A data-rich society will give the survey creators and focus group holders much better data to start with and much better questions to ask,” he told Research. “I think there’ll also be more new kinds of research being done just because there’ll be more new kinds of data in the world.”
Similarly, e-commerce tools can offer the opportunity to collect better data on how purchases are made, by providing better ways for consumers to make those decisions. PrismaStar makes a system that allows customers to rate the relative importance of product attributes and narrow down their choices in a natural way. The firm also sells analytics services based on the data it collects, and expects this to account for more of its business in future.
How the politics of data ownership will play out is tough to predict. Research firms will be hoping that consumers will see the benefits of tracking and processing their own data in ways that can be useful to everyone. And they’ll face the challenge of convincing them that there’s more benefit in sharing their data than in hoarding it.
The internet offers extraordinary potential for targeting content, potentially removing traditional media research from the equation by providing a picture of exactly who is doing what, and giving them ads based on that data. But companies specialising in behavioural ad targeting haven’t had much success in winning over consumers.
On specific sites and networks of sites it’s usual for systems to remember you and offer you targeted ads and recommendations. Amazon, Facebook and Spotify do it all the time, and many of Google’s services are supported by ads based on your data. This seems to work for both companies and their customers – Amazon’s recommendation engine drives up to 20% of the revenues, depending on the product category.
But the potential for tracking people’s activity at a higher level has hit obstacles. Some firms have sought to team up with internet service providers, giving them the ability to track everything an individual does online. They claim that this would be done anonymously, but privacy groups say that’s an illusion because the companies have the ability to assign a unique number to each individual and build a detailed profile of their activity. So these companies face a challenge to win the public over, and so far they have a fairly dismal record.
In the US NebuAd closed down after facing a storm of controversy and scrutiny by government bodies. In the UK Phorm came under fire for testing its system without letting customers know, and was shunned by ISPs as well as many online publishers who said they did not want traffic on their sites tracked.
Phorm declined a request from Research for an interview, saying no-one was available, but its latest marketing materials reveal a move to present a friendlier face to consumers. The ad-targeting service comes alongside a free consumer-facing tool called Webwise Discover, which suggests sites and articles based on what you’ve visited online. And despite its problems in the UK, Phorm has struck partnerships with ISPs in South Korea and Brazil. Behavioural targeting may yet transform online advertising measurement, but it will happen on consumers’ terms.