Annual Review of the Research Technology Business
Conducted with a range of senior industry stakeholders, the recent Asia Research Stakeholder Survey showed that most opportunities for the research business are in the application of new technology for greater speed, reach, and insight. However, we also see that the quality of data from online research could be a major threat.
To explore these issues in greater depth, Asia Research interviewed the heads of some of the leading research technology (ResTech) firms in Asia. We spoke with Maz Amirahmadi, Managing Director for APAC of the insight community specialist Insites Consulting, and a couple of panel/ResTech firms, including James Burge, Managing Director Asia for Dynata, and James Rogers, Managing Director for Asia of PureSpectrum.
In previous years’ reviews of the ResTech business, much of the discussion around ResTech has focused on panel and mobile research developments – e.g. expanding panels across Asia or further developing use of mobile, passive data collection, and ‘in-the-moment’ research.
Today, the hot topics around ResTech are in the application of artificial intelligence (AI), machine learning (ML), automation, and the metaverse.
Maz from Insites Consulting looks at the key developments in three main areas. Firstly, data collection: ResTech allows surveys to reach consumers in more ways, sometimes at crucial stages in the purchase path, e.g. within websites and apps. The metaverse is bringing new products to life in research through packaging and new experiences, and also making the research more interesting for respondents – hence improving engagement in the research itself.
Secondly, AI applications are aiding researchers in the analysis of unstructured data – e.g. producing summaries of large surveys with many open-ended responses, with very fast turnaround times.
The third area is in the activation of insights: ResTech is producing more data and insights in a shorter time, so client organisations need to be able to implement strategies more quickly. This is where more development in ResTech could be applied in order to keep pace with the volume of findings to which clients now have access.
James Rogers from PureSpectrum says that, coming out of the pandemic, there have been more developments in customer experience research, with many brands seeking faster turnaround in the research and the ability to reach wider audiences. This has contributed to faster adoption of programmatics, e.g. providing direct connection from the brand to suppliers’ panels, and the adoption of more tools for automation, not only for data collection but also for analysis and automated reporting.
James Burge from Dynata comments that ML has been applied well to panel management. Dynata has millions of panellists and implements hundreds of surveys each month, but allocating panellists to these surveys is highly complex, so ML has been used to optimise the panel, allocate panellists to appropriate surveys, and better manage recruitment and data quality.
The Asia Research Stakeholder Survey shows that the quality of online respondents is becoming a major threat to the industry. In the West, where panels have been established for longer and have had more time to ‘become corrupted’, the concerns are even higher and are now considered a leading risk to the industry.
Consequently, the industry is responding to this threat. Maz from Insites Consulting states that some clients have been encouraged to use more proprietary panels so that they will have more confidence in the identity of the respondents, as well as their relevance to the category being researched.
Dynata uses algorithms from leading survey experience platform inBrain.ai, a Dynata solution, and Quality ScoreTM AI to qualify real and unique consumers, as well as to filter out bad actors for the best possible respondent engagement and data quality reinforcement. On top of that, Dynata has also developed more specialist panels through loyalty reward programmes, or where different types of rewards such as charity donations are more suitable for certain types of respondents to ensure the target audience is fit for purpose.
PureSpectrum has undertaken a lot of work on quality, including developing a suite of ML-enabled solutions to assess the behaviour of individual respondents, e.g. how they behave in a survey, how well they respond, completion time, etc. It provides a high or low score for each respondent, and if they score low, they will not qualify for any customer surveys.
James also states that partnerships with publishers can help provide access to target audiences who have more of an interest in the category, such as using gaming publishers rather than traditional panels to survey gamers about gaming.
Piers Lee, MD of BVA BDRC Asia, comments that engaging respondents in topics that are of either general interest or of vested interest to the consumer is key to ensuring online survey quality. Piers says that BVA BDRC has developed its own panels of parents to assess their needs in relation to schooling and education. Parents make huge investments in their children’s education, want to get the very best for their children, and sharing their views via surveys is a means of crystalising their own thoughts and views on the topic, and for schools to take notice of parent’s preferences.
The future of AI
Clearly there will be more developments in AI for applications in consumer insight. James from PureSpectrum says that AI will be used more as a labour-saving tool, taking a lot of the routine work out of research. Maz from Insites Consulting concurs, but also points out that AI itself can pick up insights that human analysts might otherwise miss due to lack of time, tiredness, or their own personal biases.
Maz also highlights some limitations of existing AI – e.g. culture, nuance, tone, and sarcasm. Maz is currently working on how to embed cultural variations into AI analysis, including developing analysis of a set of 30 different emotions across a range of markets, smart probing on open-ended questions, and examining cultural and linguistic nuance.
Recently there has been a lot of talk around ChatGPT, a natural language processing tool driven by AI, which people can have more human-like conversations with than with standard chatbots. Google has recently launched Bard, another chatbot interface; instead of typing in keywords, Bard lets you have a full conversation as you learn new information. We will need to wait and see if such applications will have sufficient intelligence to replace human moderators.
Many see the role of AI as being ‘part of the project team’, working in conjunction with human researchers and providing the broad themes and some direction, but with humans still needed for the ‘subtle and clever stuff’.
This article was first published in the Q1 2023 edition of Asia Research Media