Thought Leadership

AI and Market Research: Compatible or Combustible?

April 24, 2019
Author: Rob Darrow
AI & Market Research

Artificial intelligence (AI) has increasingly been in the news as a technology that will radically change the world around us. Autonomous vehicles, virtual assistants and medical diagnostic systems are just some AI-based services that will alter how we live. However, whether the impact of these changes will all be positive is an open question. To that point, some individuals focus on the efficiencies and insights that AI will produce, while others are more concerned with a loss of control, a vulnerability to meddling and the loss of jobs that might result.

With the potential ramifications AI presents across all industries and fields, what are the implications for market research? Specifically, will AI serve as a competitor and undermine the need for specific research services and tasks? Or will AI serve as a powerful complement to market research and enable enhanced services and insight?

The most likely answer is both. Some research is likely to decline as organizations start to fully pursue AI initiatives. However, in other cases, AI is expected to create new opportunities by providing greater focus and additional options to leverage.

Market Research Still Has a Role to Play

Despite the impressive range of new capabilities and insights offered by AI, it is not all-powerful or infallible. As a result, market researchers would be wise to both:

  • understand the current limitations of AI and the role that market research can continue to play in filling client needs, and
  • recognize the genuine advancements that AI is making in business analysis and then enhance and extend these new insights through focused, in-depth market research initiatives.

AI Constraints Are Real

There are several reasons AI is not an existential threat to traditional market research, and they relate to some of AI’s current constraints and limitations:

  • “Fragile” systems: AI is only as good as the trainers and/or the data that are used to train the models. Deep learning models, the most advanced form of AI, are able to train themselves but only through exposure to a huge amount of data. If the initial construction of models is not robust enough or there is insufficient or incomplete data to adequately train the deep learning models, AI inaccuracies will result. The Wall Street Journal article “Who Comes to the Rescue of Stranded Robots? Humans” speaks to the fragility of AI systems by chronicling how humans have needed to “rescue” autonomous food-delivery robots found stuck in gardens and snowbanks.
  • AI models as a “black box”: One significant problem with AI models is that it is often difficult to understand the “why” behind the model. Even those algorithms programmed by humans can be difficult to follow. In turn, algorithms and the underlying logic behind the answers produced by deep learning systems can be unfathomable. This makes it difficult to determine what the key drivers of a decision might be or how to troubleshoot output that may not be sufficiently accurate. Most important, many find it difficult to have confidence in output from a “black box” they don’t understand.
  • Resource constraints: Data scientists are highly trained professionals and are in high demand. In addition, the development of algorithms and the training process can be intensive both in terms of labor hours and computer processing equipment/capabilities required. Given that AI is frequently resource-intensive, it is unlikely to be a practical solution for addressing all business questions and decision-making.

AI Will Increasingly Influence Marketing Decision-making

However, it is clear that AI is a powerful tool that will play a huge role in optimizing business operations and marketing decision-making. As an example, many organizations want to use AI to better understand and use the vast amount of data they already generate and own. By processing and analyzing these data, AI is able to identify key variables and important relationships between these variables that might otherwise go unrecognized. AI is also able to predict certain behaviors by determining how these and other variables will likely impact decisions. Therefore, AI will increasingly be used by marketers to evaluate different options to determine which will produce the best outcomes.

The Ability to Leverage AI Insights Is Also Real

Rather than worry about the negative impacts of AI on market research, a better approach might be to focus on the ways that market research can leverage and benefit from AI. As AI applications develop and evolve, so too will market research opportunities that leverage AI findings and insights. A few such ways might include:

  • Digging into the “what”: Using AI’s evaluation of social media and other customer feedback systems, market research can further probe the themes, messages and needs identified by AI, as well as explore new product ideas, pricing, promotion and distribution ideas developed in response to these insights.
  • Examining the “who”: As AI defines different consumer segments and high-potential prospective customers, market research can conduct additional in-depth research to gain a more complete profile of how these individuals think, feel and act.
  • Explaining the “why”: While AI identifies key trends and relationships that predict consumer behavior, market research can work in concert to more fully explore what is driving a particular trend, determine the rationale behind key triggers, and understand how key influences come into play.

AI will clearly advance the ability of organizations and their marketing departments to better understand their markets and key factors for success. At the same time, market research can and will play a critical role in building on this understanding to create even greater and more extensive insights. It is in this way that market researchers will continue to add value in the era of AI.

Let's Connect

Rob Darrow
Rob Darrow
Senior Director, Qualitative

Rob is a seasoned qualitative research and marketing professional with over 30 years of experience in qualitative research, product marketing, and product management. Rob has served a variety of clients in banking, wealth management, and health insurance including Bank of America, BlackRock, Blue Cross Blue Shield, Capital Group, Franklin Templeton, Merrill, Quicken Loans, United Healthcare, Vanguard, and Wells Fargo. Among other targets, he has worked with financial services executives, advisors, small business owners, health insurance professionals, and consumers. Prior to joining Escalent, Rob held research positions at Ipsos UU and King Brown Partners. In addition to his research background, Rob has 16+ years of experience in the tech industry, and held senior Marketing roles in both Fortune 1000 corporations and start-up firms including Motorola Computer Systems, Plantronics, Pixo, and Vocera.  Rob received a BA from Stanford University and an MBA from UCLA Anderson School of Management, and he is a RIVA trained moderator.