What can we expect in 2024?

Rarely in data and insight do we see transformative changes over night. Artificial intelligence, big qual, data fraud and behavioural science have been gathering pace even before last year. However, 2024 is likely the tipping point in commercial application for these elements of the data and insight industry.

AI and Synthetic Data
In the first quarter of 2023 a lot of conference presentations on the topic of AI focused on what could be possible rather than demonstratable impact. With the growth of open source applications over the course of the year we’re sure to see the industry early adopters demonstrating more and more applications in 2024. One of those areas will be synthetic data. The advantages of synthetic data are clear in terms of speed and resource, however as an industry will need to be extremely careful to ensure that this synthetic data does represent and replicate the biases and irrationality of the human mind rather than the logical processes of artificial intelligence – if you want to know more about it you can see my recent presentation on the topic at last year’s IIEX Behaviour conference. A too quick adoption of unreliable synthetic data for commercial projects could potentially be quite damaging to the industry and the longer term adoption of AI.

We also need to manage our expectations of the impact of AI on the industry. The same voices that sensationally suggest that AI will “kill” research as we know it, said the same thing about automation. And while automation has provided a number of benefits and created a new space in the industry with new, and very successful players, it’s potential for disruption was overstated. While we will certainly see new and exciting applications of AI this year, we won’t be seeing peak disruption.

Big Qual
The term and application of qualitative research at scale, or “big qual” is not brand new, although recent. Dr Sarah Lewthwaite at the National Centre of Research Methods, was publishing papers on qual at scale as an evolution of big data back in 2018. And the idea of getting the depth of qual at the scale of quant research is an appealing one. Not just because of the quality and depth of insights you could uncover, but we also know that stakeholders can often feel that scale can make insight more legitimate. The application of AI in to qualitative analysis can also help expand what can be mined for qualitative insight; recorded customer service calls, reviews and complaints, as well as archived video and recordings of participants, and social media.

In parallel to technological innovations, methodological innovations such as System 3 and Narrative Research can, by changing our approach to how we answer questions, open traditionally quant surveys to qual at scale. Where we were once resistant to open-ends due to coding and resource concerns, these methods paired with new AI analysis support, can help unlock new insights close the gap between claimed and actual behaviour.

Data Quality
With the new Global Data Quality initiative from a number of national and international trade associations, data quality, or “data fraud” looks to be a key issue for the industry in 2024. This issue takes on additional significance in the a world of AI, machine learning, and synthetic data; bad inputs will lead to bad outputs. But let’s be honest with ourselves, this has been a problem for some time. If you take a look at the resources on the Global Data Quality, the 37 Questions to Help Buyers of Online Sample from ESOMAR, dates back to 2012. What Trade Associations can achieve in this area is limited, to really move the needle on data quality panel providers and agencies will need to step up, and that will be driven by research buyers. From our own research with research buyers we know that many assume that their agency partners are already mitigating for poor quality data. Unfortunately that’s not always the case. However this year as this becomes a more of a topic outside of internal industry discourse, hopefully clients will be actively making sure that agencies do have procedures in place to ensure their panel providers take care of the technical aspects in this area, and agencies are asking questions in ways that mitigate the issue. Only when there is a significant commercial impact in not taking care of data quality will we see commercial agencies do more in the area. In this area AI is a double edged sword it can both analyse responses and detect fraud, but can also be used by fraudsters to create more realistic fake responses. The big question that researchers need to ask this year is; do we allow the fraud arms race to continue or do we start looking at reforming the core reason for fraud in the first place – the incentive.

Behavioural Science Steps Up
Over the last few years behavioural science has gone from the fringe to a regular approach in the researcher’s toolkit. It is often used more for big complex questions rather than a universal approach, however, it is something that can be applied to every research method as an approach to help get closer to your customers and give you the “why” behind the “what”. Behavioural science provides insight that can drive decision-making and direction, rather than just a datapoint. In 2023 we could already see examples of behavioural science being applied in a broader set of circumstances. In the finance sector the FCA Consumer Duty regulations placed behavioural science at the centre of communications and product testing in improving outcomes for consumers. In November we took to the stage with Vodafone at the ESOMAR Art and Science of Innovation to demonstrate how they use behavioural science across the innovation function; while brands like Reckitt are already well known for championing the use of behavioural science in insight.

In the face of more consumer uncertainty I expect 2024 will be the year we see more brands catch up and apply behavioural techniques to more traditional research methods as they try and close the gap between reported and actual behaviour. What’s more, behavioural science can be the solution to the key issues mentioned above. Behavioural science can help us bake in human irrationality for more accurate synthetic data, and it can allow us to create questions in online surveys that improve the participant experience, mitigate biases, and can’t be answered by respondents flatlining their way through surveys.

If you’re interested in understanding how you can apply behavioural science to more of your research methods, check out our presentation on using behavioural science to close the say-do gap here.