Getting Smart On Data


Asia Research recently concluded its annual review of the online and research technology business in Asia. This review involved conducting interviews with a range of industry experts who are the heads of some of the leading online panel and research technology companies operating in the region.

For 2017, we interviewed Martin Filz, CEO of Lightspeed; James Burge, Managing Director of Research Now; Ludo Milet, Managing Director of Toluna; Fumiya Nakajima, Global Business Director at GMO Research; and Mark Lepine, Managing Director of SSI.

Some of these companies are redefining their market position having moved from being ‘online panel managers’ to ‘survey technology providers’, and today positioning themselves as broader ‘data management companies’ whose work includes automation of insights and passive data metering. This demonstrates how far and how quickly these organisations have developed their product offering to include more holistic data collection, analytics, data delivery, and reporting solutions.

As we conduct these reviews, there are particular developments in the industry that define the year under review. In the opinion of our industry experts, one of the key developments in the industry in 2017 was the greater use of ‘combined data sources’. This includes fusing primary survey data from standard Q&A surveys with both passive data collection (online metering and geolocation) and data that might already be held on the respondents in the panel. For the latter, a lot of information can be retained about respondents from screening questions and past surveys, which can then be built up within a data bank to give far more insight into each respondent. For customer-specific surveys, clients can append additional data on customers (e.g. transaction history) to provide more insight into each customer, and there is the facility to combine this with information on customers from their social media (e.g. pop-up groups that they have liked).

All this additional profiling can help to identify segments more easily within panels, so that online surveys can be more targeted, with less screening out of respondents. This can bring down the costs of online research for very targeted samples.

This approach somewhat replicates the principles of ‘insight communities’, where data held from past surveys can be used to provide greater insight for subsequent surveys without the need to ask so many questions. As consumers get increasingly impatient with long surveys, this approach helps to reduce survey length and hence promotes greater participation in and completion of surveys.

While it has been slow progress to reduce the length of questionnaires, clients are becoming more aware of the limits of PC-based surveys. Martin Filz from Lightspeed comments that APAC has both a disadvantage and an opportunity, with local clients still transitioning from offline but jumping straight to mobile online and skipping PC. However, both established online markets and growing Asian markets face the challenge of cutting down either lengthy online trackers or offline questionnaires, and this is no simple task.

Lengthy tracking questionnaires can result in nearly all surveys being completed on PCs rather than mobiles, with obvious implications for lack of representation. Younger generations may only entertain surveys on mobiles, and this generally requires surveys of 10 minutes or less.

In order to reduce survey length, some organisations are opting for chunking (i.e. getting a respondent to do surveys in stages), or are just running multiple shorter surveys with balanced samples, with each survey addressing different elements of a study but with different respondents.

Fumiya Nakajima from GMO Research comments that mobile surveys are particularly important in Asia, and many solutions have looked at cutting up surveys into chunks, but looking at all these chunks as one whole survey.

Martin Filz comments that, for practicality’s sake, clients might opt for ‘least fill questions’. For example, within a broad U&A survey on financial services, when the survey encounters a respondent with an unusual financial product, the survey would focus just on this product so as not to bog them down with questions on the mainstream products, for which there will already be enough data from others taking the survey.

Martin also states the need for Modern Surveys which address length and compatibility issues, as well as incorporating passive and third-party data to limit the questions but maximise the insights clients need to answer the business objectives.

Mark Lepine from SSI comments that the move to shorter surveys cannot come quickly enough. Researchers are having to compete more and more for the online attention of consumers, so not only do we have to think about creating shorter surveys, but also more engaging ones by providing a better survey user experience for consumers in the same way as other online interface providers are doing. Aside from technological developments, some comment that another defining characteristic of the online research business in 2017 is greater adoption of online research in general. Mark Lepine observes increased use of online methods in SE Asia, which has been traditionally associated with the dominance of offline data collection. Local, Asia-based panel start-ups are also helping to develop this business with homegrown panels.

Change in buyer types

A consistent finding in this year’s review is the emerging range of new buyers of online research. Increasingly we see marketing personnel from client organisations bypassing their own insight departments to access consumer information directly from the panel/technology companies. Most of these requests are for very short, simple surveys, where the client need is for very fast feedback to aid tactical marketing decisions.

Ludo Milet from Toluna draws a distinction between the business they receive from end clients and that from research agencies. The end clients have ‘questions to answer’ (and very quickly), whereas the research that is channeled through insight departments and agencies tends to be ‘projects’ requiring considerably more thought behind the design and the analysis. Ludo feels that this ‘dual demand’ will continue, and the request for answers to questions from end clients will not kill the need for projects from insight departments and agencies.

For example, Toluna provides weekly feedback to a telecoms firm to understand the main motive of consumers who happened to buy a mobile phone that week in choosing their phone. This very simple survey tracks the hot selling points of mobiles at that point in time so that communications can be adjusted accordingly, or simply to provide insight for the client’s next marketing meeting.

In 2017, James Burge from Research Now observed increased use of video-based research. While online quantitative research traditionally maintains a certain distance between the client and the respondent (compared to qualitative research), greater use of video – such as through Voxpopme – provides clients with another window into consumers’ lives. These videos can almost be mini ethnographies, as we can actually see the respondent in their real-life situation. BDRC is using video to provide audits of mystery shopping through their BDRC Vision approach, as well as linking mini qualitative studies with standard online segmentation surveys, like with their research into TV news viewership with the BBC.

Ludo Milet from Toluna comments that another development this year has been far more automation of research. For example, product and concept tests that follow a very standard test protocol can be piped through programs with little human involvement in the survey. Toluna’s PowerConcept and PowerPack, which run on their own technology and panel, can deliver an end-to-end solution in 48 hours.

Mark Lepine comments that lower costs and more accessible surveys are opening up research to companies or institutions who previously could not afford it. They are providing similar end-to-end solutions, where clients can undertake a simple survey, select a sample type, and have it analysed without having to interact with any researcher. Similarly, Martin Filz comments on key developments for Lightspeed’s end clients, including user-friendly online reporting and access to data in the field which connects the brand to their data.

But again, it’s horses for courses – these solutions will not replace the full agency-based concept test where more complex purchase paths and nuances might have to be assessed.

What happened to communities?

Asia Research has frequently commented on the relatively low adoption of online insight communities in Asia, as reported in our client surveys earlier this year. Initial objections to insight communities from clients have included the costs, particularly in the set-up, but competition and price restructuring have addressed some of these concerns. However, Mark Lepine from SSI comments that insight communities are being compared directly with other online methods and are still viewed by some clients as expensive. Mark comments, “Clients need to be informed that insight communities require a lot of time in management and moderation, making it quite labour intensive. But the value from insight communities is enormous.” He continues, “Perhaps what is missing are local case studies that can really point to the benefits of communities over traditional research, including qualitative methods. With some experimentation and growth in expertise, I see no reason why communities cannot be as popular in Asia as they are in the West.”

For communities to work in Asia, Fumiya Nakajima at GMO Research again comments on the importance of mobile.  For example, asking panelists to perform tasks on PCs will not work in Asia, but if the online community platform can be adapted to mobile then it can work well in Asia.

The challenge in Asia is that many clients’ budgets are simply not large enough to keep a panel engaged. We also spoke to Karen Schofield from Join the Dots, who are active in this area. She points out, “for an ongoing online community to really be a ‘community’ rather than a more traditional panel, we need to be having regular conversations with our members, talking to them at least a couple of times a month, and ideally more often”. While client-side researchers who focus on a specific market often have sufficient research needs to generate this level of engagement, clients who have responsibility for insights across multiple SE Asian markets can find this more of a challenge, since they focus on different markets at different times, depending on the business’s needs.

In this scenario, short-term ‘pop-up’ communities in the markets of interest can be an alternative, as they are set up to meet a specific set of objectives over a fixed period (e.g. 1–4 weeks). Karen comments that communities are actually a very flexible tool and can be scaled up or down as needed, and a short-term community can be a great way of showing the value of an online, longitudinal approach to internal stakeholders, without the need for a long-term commitment.

Martin Filz from Lightspeed comments that a potential solution is in the sharing of insight communities across different clients. For example, an insight community can be developed for a specific category (e.g. sportswear), but different brands can tap into this community for a range of studies.

However, as touched on at the beginning of this article, the online panel companies are now using their own panelists as a form of managed community (e.g. learning more about them across various types of survey), and hence these panelists can be utilised in the same way as insight communities. In Australia, for example, Martin states that Lightspeed’s own panelists are being invited to take part in online bulletin board research over a two-week period, making them in a sense a type of short-term community.

This could be the trend for insight communities in Asia, where they are not brand specific and will be ‘owned’ by either the research agency or the specialist panel companies.

The B2B opportunity

While still in their early days, B2B panels are gaining some traction in Asia, but Mark Lepine comments that there is a bit of a “chicken and egg” issue here. In order to grow and maintain B2B panels there needs to be the demand, but there is low demand due to lack of B2B panels. Currently, much of the demand is coming from outside the region where global B2B studies need to be extended to Asia.

Indeed, it was global demand that helped develop consumer online research in Asia, often cross-subsidised by Western investment, and this trend seems to be repeating itself for B2B panels. These panels are sometimes developed from consumer panels, where job titles and functions can be verified through LinkedIn profiling as an additional quality check.

Artificial intelligence

A bit like ‘big data’ five years ago, artificial intelligence (AI) is one of those buzzwords that the research industry is talking about. But what does it really mean and what are the implications? We put this question to our industry experts. Martin Filz from Lightspeed feels that AI fits in with appended data insight on the target audience so that it does not require additional questioning. This can include looking at correlations – so, for example, a particular consumption habit can imply other habits without even needing to ask the question (e.g. wine drinkers are almost certainly also cheese eaters), and unless you need an exact figure on the incidence, the correlation itself can be enough information for some clients.

AI can mean understanding consumers through other data sources, such as social media profiling or information held from past surveys. Future surveys could be a series of ‘micro-surveys’ made up of two questions asking a ‘what’ and a ‘why’, and the AI will tell you the rest about who they are. Ludo Milet from Toluna sums up AI as “analysing data without even collecting it”, but human judgment is still needed to determine whether these are quality insights or not.

James Burge from Research Now thinks that AI can power more standard analysis and insight generation that can feed into end-client tools and dashboards, replacing the more mundane tasks covered by researchers today.

But will AI kill the industry? Overwhelmingly, the answer is no! Fumiya Nakajima at GMO Research says that, while there are digital and automation providers, “the knowledge and expertise of extracting the ‘why’ is our asset held by the traditional research industry, so I believe there is a way we can mutually find a way to collaborate – the power of asking and analysis is the only way to obtain real insight from the enormous data”.

James Burge comments that some agencies are growing towards more strategic research, and this opinion is also held by Martin Filz, who states that the industry will still need “the best people for seeing the nuggets in the data, and researchers are best placed for doing this”.

First published in the Q4 2017 edition of the Asia Research Magazine