The Future of Behavioral Science in Business: Part 1
The “Jobs to be Done” of applied behavioral science
Introduction
The turn of the year allows us to reflect on our discipline's past and future. Let’s examine Behavioral Science through the lens of two comparable "science/research" fields with explosive growth in business in the last few decades: data science (DS) and UX research (UXR).
In the first of three installments, I'll examine the evolution of DS and UXR job descriptions to inform us how a professional discipline establishes itself: consensus on role names, tasks the roles require, and background and skills. I am a schematic thinker, so I've summarized my thought process on the evolution of data science and UX research positions in the image below:
There is a lot to unpack. Let's look at each stage. We'll start with the Wild Wild West in the upper right corner, move to the upper left corner, head down to the lower-left corner, and end with To Each Their Own in the lower right corner.
1. The Wild Wild West
This stage has no structure, and many job titles and backgrounds characterize it. People with wildly different skills do the same job with other job titles, and people with duplicate job titles do very different jobs. Due to the lack of explicit or formalized training paths, this stage's poster child is the "accidental" data scientist or UX researcher, someone in the right place at the right time.
DS job postings for the same tasks called for data analysts, business analysts, applied statisticians, machine learning engineers, etc. Typically, requirements included Java, C++, or another software development language (because they work on computers, I guess?).
In the UX field, if your job title was "UX designer," you might have been expected to use surveys or conduct interviews, possibly focus groups. One could be titled an "HCI consultant," an "ergonomics specialist," a "consumer researcher," or a plain "UX consultant."
2. The Unicorns
In this Second Stage, the variety of backgrounds remains high. Companies expect people to have the same job title doing the same work in two adjacent cubicles, even if it is as non-descript as a "designer." Still, we see a progressive convergence on 1 or 2 job titles, due, in part, to the rationalizing influence of Human Resources (HR). Large companies with HR departments have career paths and "job families" (i.e., structured job descriptions for new job openings).
The market also tends to follow the trends set by the most admired tech companies (FAANG, etc.) as they create DS or UX teams and proselytize the importance of using data or being user-centric. On the supply side of the labor market, candidates vie for prestigious and marketable job titles. For example, everyone now wants to be a "data scientist."
Businesses are hunting for unicorns and complaining about the lack of skills in the market. Companies haven't yet rationalized the actual "jobs to be done" or preferred backgrounds, leading to an "everything and the kitchen sink" approach from HR and inexpert hiring managers. On the DS front, companies hope to find someone to set up their data infrastructure, clean up their data, analyze it and draw business conclusions. Meanwhile, companies want a UX designer/strategist/researcher.
3. The Right Tool for The Right Job
In this Third Stage, the left side of the graph becomes more streamlined. The lessons learned from early adopters lead to a decrease in the number of backgrounds companies will accept (especially more prominent companies), and tasks become split between differentiated roles. This change bodes well for people starting their graduate studies as they have a more straightforward path ahead of them. At the same time, the door starts closing for new graduates and professionals already in the workforce who don't have the "right" degrees. Their best hope remains in start-ups, who tend to stay longer in the first two stages.
In the data world, it's the era of the "Ph.D. in physics turned data scientist" and the "software engineer turned data engineer." In UX, there's a cleaner separation of duties between researchers and designers. Researchers converge on qualitative research as the core foundation of their discipline, while designers rally around prototyping/wireframing tools.
4. To Each Their Own
We get a perfect linear mapping of backgrounds to tasks to titles in the Fourth Stage and lower right quadrant. This final stage translates into more job titles with few acceptable experiences, specific skills, and explicitly defined tasks.
This quadrant is structured well for "jobs to be done." It created corresponding job titles that higher and continuing education can now use. Individual experts and consulting firms early adopt and develop seminars, workshops, and boot camps. More slowly and decisively, universities and business schools get their piece of the action through continuing education and creating specialized master's programs. This evolution legitimizes the fields. I once joked that a technical field becomes legit with its first O'Reilly book…but graduate programs also count!
As the education arena becomes crowded, tensions emerge: "accidental" professionals with a general Ph.D. (respectively, physics and anthropology) compete for individual legitimacy with graduates of shorter but specialized programs. And now we ask…does a 12-week boot camp an expert make?
Where does Behavioral Science fit in the picture?
While the four stages I outlined above are not perfect, this model helps explain the forces at play in the progressive emergence and organization of a new discipline in business.
Where is behavioral science in that picture? My opinion is it's somewhere around Stage 2, The Unicorns:
Where is behavioral science in that picture? My opinion is it’s somewhere around Stage 2: The Unicorns.
- There are still a variety of job titles out there, but "behavioral scientist" or "behavioral science consultant" lead the pack.
- There is a growing consensus on the core jobs a behavioral scientist needs to perform, like implementing nudges and measuring their effectiveness through A/B testing.
- The diversity of backgrounds is still wild though we already see specialized degrees. I hypothesize this relatively fast reaction from higher education institutions builds on existing capabilities. Institutions that offered degrees in economics, public policy, data science, and UX have moved laterally to provide degrees in behavioral science without starting from scratch.
What does this mean for the future?
It was not a coincidence that I compared data science and UX research. Applied behavioral science is joined at the hip to both. Behavioral science's future will be similar to the current state of these disciplines. But how will it look?
On the one hand, sometimes behavioral science is called "the last mile of data science," which makes sense because data analysis is indispensable for A/B testing. On the other hand, deep behavioral research on customers' mental models is indistinguishable from UX "foundational research." Albeit UX vocabulary is slightly different, UX practitioners paradoxically focus more on observed behaviors than behavioral scientists who zero in on mental models and cognitive processes.
Like UX, we're beginning to see a differentiation between "behavioral scientist" and "behavioral designer," and I think this trend will continue. A scientist can design, and a designer can research. But ultimately, there are distinct skillsets and mindsets involved in both. Behavioral scientists should have exposure to UX design thinking because it is such a critical framework. However, if you want to be a researcher, there's probably not much reason to learn UX design if you're not using the relevant tools. The same goes for behavioral science and behavioral design.
It's less clear-cut what tools or roles from data science leak over to behavioral science. Behavioral scientists design their experiments and handle their data analyses in most companies. However, in my last role as behavioral science manager, I created a behavioral science team in a company that already had an experimental design team with superb statistical skills, offering me the opportunity to try out more qualitative behavioral science roles.
Creating the proper infrastructure for A/B tests at scale requires a lot of work, most of which is much too IT-focused for behavioral scientists. Most businesses are not aware practitioners need an experimentation infrastructure, though some have drawn attention to this need for some time*. I'll go out on a limb nonetheless, if only for the sake of symmetry, and venture a guess: We'll have "causal inference engineers" supporting behavioral scientists as data engineers support data scientists.
What about required backgrounds? As I often tell prospective behavioral scientists, "You don't need a Ph.D. to do the job, but you might need one to get the job." I often lament the elitism implicit in requiring a Ph.D. for an open position. I don't regret getting a Ph.D. in behavioral economics, as it gave me both depth and breadth in thinking. However, I learned a lot of economic principles entirely irrelevant to an applied behavioral scientist.
I don't expect the gap between the number of jobs in academia and the number of new Ph.D.'s to close any time soon, meaning a continuing supply of bushy-tailed (if sometimes disappointed) bright new doctors outside of academia. The gap suggests the persistence of a two-tier labor market with haves and have-nots, similar to the role of "badge of honor" played by the MBA in business.
Conclusion
Let's recap the key points and what I'm reading in the tea leaves:
- Data Science and UX Research offer us a glimpse into what the near future of behavioral science might look like;
- As the field matures, there will be an increasing streamlining of backgrounds, tasks, and titles;
- The nascent distinction between behavioral "researchers" and "designers" will deepen, and coherent skillsets/experiences in one or the other will be at a premium;
- A Ph.D. will remain a de facto requirement for many positions, especially higher up the food chain.
Again, this is just my take. Feel free to argue otherwise in the comments!
In the second installment of this series, I discuss how an academic field becomes a business discipline. In the last installment (coming soon!), I’ll consider if there’s a future for behavioral science outside of nudges. Follow me if you don't want to miss it when it comes out!
Big thanks to Ruth Schmidt for her insightful comments and suggestions on an early draft of this post! All errors and foolish statements belong to me.
References
* Linnea Gandhi & Erik Johnson, "8 Things to Do Before You Run a Business Experiment", Harvard Business Review, 2019.
Feel free to read my other stories on Medium: https://medium.com/@florent.buisson.
I also have a book about behavioral data analysis.
