Join us for networking & quality resources to help you and your team succeed in digital transformation.
I’m not afraid of machines taking my job away. Here’s why.
I’m a marketing researcher and marketing science, by various definitions, is my specialization. Much of what I do is now called data science, though this term means different things to different people.
During the thirty-plus years I’ve been in marketing research, technology has affected what I do and how I do it, but the central parts of my role have not changed dramatically. I am a customized marketing researcher and, though there are significant learnings from other projects I’ve worked on, each project is unique.
Customized research is primary research, tailored to specific client needs at a specific point in time and very different from syndicated research, which is essentially the same from client to client. I am a researcher and do not sell products. For those of you who are not marketing researchers or just starting out on your marketing research career, Why Bother with Marketing Research? is a brief overview of the many reasons why clients conduct marketing research.
Many of my clients, like me, are marketing researchers and work at MR agencies. They often need specialized analytics but lack the internal expertise or capacity to do this themselves. These needs, in turn, are driven by the needs of their clients – the end-clients. I work with end-clients too, though less frequently. I am using “needs” imprecisely here and should really say perceived needs. As anyone working in marketing research knows, decision-makers typically have a vague sense of the type of marketing research they think they’ll need to make certain decisions.
To do my job competently, I need to understand the background and context of these decisions, as well as who will use the research results, how they will use them and also have a rough idea of timing. There may be other decisions they should be contemplating, in addition, and this may emerge during discussions when the research is being designed or when the results are being “socialized” internally by the client. Moreover, an end-client may ask for conjoint analysis (for example) but actually require another method or a different sort of conjoint – most statistical methods have many variations.
None of this can be automated, nor does any AI exist that can stand in for me. If it did, most likely clients and consumers would also be AI, and “reality” then becomes a Sci-Fi comedy.
Statistical modelling is built on mathematics and computer technology but requires judgement and gut-feel decisions. I’ve touched on this in Taking Quantitative Marketing Research to a Higher Level and other posts. Only some kinds of analytics can be done mechanically in an automated fashion, as data analysis requires human decisions, many of them based on experience and intuition, as noted. Moreover, communicating the result of customized research in general is hard to automate.
See AI, Big Data and Decisions and Who Cares About Evidence? for snapshots of what I’ve learned about how we humans make decisions. Much of what I’ve written applies to data science as well – you might find Data Science and Data Science Fiction of interest. Are Data the New Oil? addresses some of the hype about data science.
Over time, AI and automation will have more impact on what I do and how I do it, mostly for the better. But it cannot replace my role entirely without a radical transformation of human reality. I am not afraid of machines. Not yet.
Article by channel:
Everything you need to know about Digital Transformation
The best articles, news and events direct to your inbox