Summary of ESOMAR Fusion Day 1

Coming to you live from Madrid! It’s been a great first day of ESOMAR FUSION, the conference that mashes together the latest and greatest in data, tech, quant, qual; pushing us to consider the future trajectory the insights and marketing science profession is taking and how we can propel it along into new areas. With delegates from 35 countries, we’re expecting content around A.I., data fusion, privacy and security, and organisational transformation to harness the world of data. And Day 1 delivered just that.

Today, Pete & I have seen great papers from the likes of Microsoft, on the topic of transforming the insights supply chain and data democratisation, Molson-Coors on the topic of how to push data science and data assets to their limit in extracting high value consumer insights in the beer categories in Canada, and Netquest on the implications of machine learning for privacy – a topic very close to our hearts here at Nature given our significant recent research on the state of privacy, consumer trust and perceived data security in Australia. On top of this, we saw tons of great content from European and North American tech and insights suppliers, which really provoked thought. So much great content, so much jet lag to deal with!

Before the conference even began, Pete and I worked off the jet lag along with our colleague and friend Andrew Therkelsen from The Lab, walking the streets of Madrid on a cold late autumn Sunday. In between and during our visits to the famous Prado and Thyssen Bornemisza Museums, we talked about where we thought the consumer insights industry would be a decade from now, HOW we would we conducting research in a decade, and what the key undercurrents are driving change right now. In this era of tremendous innovation and tech change, we couldn’t help but keep coming back to the thought that for research to have impact and drive change in our clients’ organisations, that regardless of HOW the work is done, the human consulting skills involved in problem formulation and framing, relationship formation, communication, relating analysis to objectives, and delivery … are all king. These skills were key when we moved from the now near dead methods of CATI and face-to-face or door-to-door data collection to online; and they will be key as we gradually blend in new conducting research now and in the future. Ironically, even though were we about to attend a conference all about tech and innovation, we’re strongly of the view that at the end of the day, it’s important to not sweat the technique and to remember that data and technology are after all just a means to an end. The human side of what we do is therefore oh so important – and without these high value skills, technology won’t get us very far. The key to use the right technology in the right way, at the right time, and to neither ask too much nor too little of it.

On Day 1, the main take away themes for us were:

1.      The first theme to emerge from Day 1 is entirely consistent with what we ourselves had felt on our Sunday walk – that human relationships are still key, and will arguably become more so in the future. This was central in many presentations, with a very good case in point being that of Barry Jennings from Microsoft, who talked of the centrality of human relationships in extraction of value from data – which is his case had heavily shifted toward toward A.I.-based and away from traditional research. In this case study, the ‘human’ element referred to setting objectives and working in an iterative style – which each implicitly involve people working with people, leveraging data and insights. One can argue that to truly collectively advance organisational knowledge, data and evidence alone are not enough; instead it is contingent on working effectively with people – at the beginning formative stages of insights engagements, during the insights derivation stage, and on delivery. It’s human interaction and communication that unlocks value to the same or equal degree as data and technology. Indeed, without effective human involvement at several key points, data and technology are commodities, and the importance of this cannot be understated. 

2.      The increased availability of third party and social media data opens up new possibilities of fusing traditional panel data and new data sources, to provide incremental value. A great case in point from Day 1 came from Coca-Cola Turkey’s presentation which talked of the criticality of understanding consumer motivations to their business, which they are increasingly exploring by complementing traditional survey measurement with social media-based insights fuelled by A.I.   

There are clear gains to be made in both efficiency and effectiveness through such fusion / complementarity. Leveraging social media data offers the means of keeping a finger on the pulse of shifts in motivations through the course of time in a passive and authentic way (effectiveness). And no surprise, A.I. delivers efficiency. Coke Turkey’s team talked of examining 11.3 million social media posts using A.I., negating 10 years of manual work. 

And it is this simultaneous ability for new technology to deliver uplift in both efficiency and effectiveness that is what is turning heads, and understandably so. 

What was particularly impressive about Coca Cola’s work was their acknowledge of the need to use both new forms of data AND traditional research.  Note that this signifies that we are a time when the way research is being done is creeping forward and evolving; not that so-called ‘traditional’ methods are being rejected, discarded or left behind. Indeed, the power lies in fusion, and layering in multiple sources of data to speak ultimately to the same topic simply from different sources. 

3.      But is there a price to be paid through the increased integration of A.I. in the research process? What cost does efficiency come at?  What must not be forgotten is that all machine learning involves a human coding element, and therefore doesn’t overcome the element of human bias that may come with more traditional methods. The role of the human isn’t therefore just about problem framing and delivery – even with the rise of technology, the role of the human is peppered through the entire process, for better or for worse.  As Ray Poynter put it today “Everything we do is biasing the data”. As such, whilst technology can and will produce uplift in efficiency, it can’t yet be viewed as a panacea. Just as the quantitative marketing scientist can directly affect the results of, for example, a regression analysis, through the addition or removal of an independent variable from an equation; so too can the data scientist by choosing alternative definitions or taxonomies in driving machine learning routines.

4.      One matter to consider is this … are we at a time when the distinction between qual and quant is diminishing? While they are likely to continue to play a complementary role to one another, new technology stands to blur the until now rather clear distinction between these disciplines. By way of example, A.I. and machine learning can extract meaning from large volumes of data unlike not before. This means qualitative researchers have at their disposal much larger volumes of data than ever before, and quantitative researchers have the ability to derive meaning from vast volumes of unstructured data more than ever before. This is about leveraging user-generated data, often alongside other forms of data. So this doesn’t mean qual & quant will converge, but it does mean that in certain arenas, one can expect a blurring at the hand of methodology. 

To wrap...

On Sunday, Pete, Therks and I enjoyed amazing art painted over 400 years from the mid-1500s to 1900s, interspersed with our discussions on the future of research! The variation in the style of painting we took in was staggering … which is of course the product of the tremendous change, development and innovation in the techniques and styles used by artists across these 4 centuries. While the expression of the creative idea, and the techniques used, clearly evolved markedly through the course of time, the constant throughout 400 years was the artists' conceiving of the idea (the brief / problem at hand), and their commitment to the production of a product and end result in the form of visual ‘story’ for the intended audience of the day. These are human skills and traits that transcend time and the prevailing innovations and ‘ways’ of the day. In this, we see an unavoidable parallel to the evolution of the research industry.