We’ve read that “innovation happens in the space between disciplines”; and as part of our ongoing quest for meaningful innovation, we’re always keen to seek the latest thinking. As such, we were delighted to be invited to the Innovation and Ideas Exchange (IIex) European conference in Amsterdam in February.
Whilst mostly a consumer-focussed conference, there were a couple of key themes repeated throughout the conference that are very familiar:
- Media overwhelm for consumers / audiences – the challenges of standing out in a complex marketplace and being ‘heard’ in a loud media environment.
- Increase in complexity, volume, format of data and challenge of data integration. Melding qualitative with quantitative and making more links between bits of information.
In order to tackle these themes, the presentations and vendors at the conference used a couple of common tools:
We all know decision-making is not a rational process; that humans are subject to social influence and make snap decisions. At IIEX, most of the presenters and vendors applied Behavioural Sciences almost exclusively using ‘implicit’ methods (Eye tracking facial coding, Implicit association tests and Neurometrics (EEG, fMRI scanning)). Across the presentations on this topic, there were a mix of perspectives on the role of these methods: some claiming all you need are ‘implicit’ metrics, versus those who firmly positioned it as complimentary (to give a different angle or sharpened insights, not expected to match up with direct responses.)
Our perspective: We strongly believe in applied Behavioural Science at the core of tackling most of the challenges in the modern pharma landscape (we’re presenting on the power of Behavioural Science to help with ‘information overload’ at the EphMRA conference in June). We also believe in the power of implicit methods as a complimentary method; however, we have very strong reservations about the emphasis of implicit methodologies alone as ‘applied Behavioural Science’
- The first is the challenge and critical importance of finding validated approaches to gathering implicit metrics. We’ve run several studies, both self-funded and in partnership with our clients, using these methodologies (and have been speaking on the topic since 2014). We’ve regularly encountered variation in the specificity and reliability of the different tools and different providers. As Michelle Niedziela from HCD research said in her IIEX parallel; “different methodologies are only validated to respond to particular research questions, which with the wrong tool, the wrong application, or wrong interpretation could give completely spurious results”. Certainly, in our own validation studies, there were many situations where we were unable to demonstrate meaningful links with behaviour. So labelling these approaches ‘implicit’ as a class without gradation and emphasis on accuracy can lead to ‘trendy’ applications without appropriate rigour, and potentially introduce the risk of brand teams making significant decisions on the basis of implicit data- that is little more than random chance.
- The second challenge is that we believe limiting the applications of Behavioural Science to particular methodological approaches is overly simplistic; sure, methods that help access unspoken responses are a critical part of uncovering the full reality of human decision making, but with our internal expert team HRW shift we’ve been passionately speaking for the past three years about the power of applied Behavioural Science expertise as a lens across the whole project (quantitative and qualitative), and as a foundation for targeted and evidence based recommendations, rather than a ‘behavioural’ method tacked on.
Social media collection
At IIEX there were a lot of presentations on the value of social media insights. From social media analysis as replacement for brand tracker (digital MR) to the shift from topic-related to audience-related data as the core value of social media data (Pulsar), the insights from social media had extensive applications for consumer products (and, indeed, we’ve seen excellent results from social media analysis in consumer healthcare).
Our perspective: Social media research has potential as part of desk research or a broader literature review to deliver an initial understanding in a therapy area; the terminology spontaneously used, interesting case studies, key social influencers, and a sense of the public dialog in the area. Yet we don’t often recommend it for our pharmaceutical clients (depending on therapeutic area), because of the small universe sizes, the risk of adverse events, challenging source validation, and the difficulty in finding specific answers to questions. Furthermore, with GDPR imminent, explicit consent procedures and data ownership have become even more tricky in this area. It is an approach that can have a lot of value but predominantly as a means of question generation, rather than question answering- so needs to be treated with caution amongst marketing teams.
Artificial intelligence (AI)
AI was a big topic and many speakers were developing their own artificial intelligence powered algorithms to make better use of data (identifying hidden data patterns, computer open-ended coding, or simple sentiment recognition). As computing power continues to grow, this is becoming a more accessible and cost-effective approach for every agency to access through in-house or partner expertise. However, the majority of the presentations acknowledged the current limitations of AI-driven algorithms with nearly every presentation touting the benefits of ‘intelligence augmentation’ (i.e. using AI that is corrected/cross checked by humans; saving time at the front end but training it further and not relying on the algorithm alone.). In the consumer space, potential applications are enormous, as one particularly compelling presentation by Shane Skillen from Hotspex demonstrated, using AI to recognise the tone and nature of the content being viewed on YouTube videos allowed advertisers to match their brands to the feelings generated and created more click-through.
Our perspective: Artificial intelligence is an area we’re following very closely. Its potential to increase the sophistication and speed of our analysis is huge, and we value the potential benefits of smart algorithms helping to identify ‘lurking’ tendencies across multiple parameters and bring these to the fore. As such, we have been trialling our own initial forays in to the area as well as exploring offerings with select expert partners. We were cheered to see that our consumer cousins are in a similar position in the application of these approaches, but we’re also impatient for the analytical and image recognition capacity of these approaches to become good enough that we can really start to trust it as an ‘off the shelf’ approach, as our data sets are not usually of the size and repetition that we can currently justify the value of creating and training a bespoke algorithm to help us with the analysis… but watch this space!
By Katy Irving