A note to academics wondering how to collaborate with companies or trying to work at one

Over the last 3 years I’ve worked to bring social science insights to large and small companies. This includes doing new research within companies and helping disseminate existing research to the hearts and minds of product people.

What did this look like?

It looked like throwing massive conferences to connect behavioral economists and startups, being a founding member of the Google Behavioral economics team and manually connecting top tier academics with companies that may be able to use their research.

Before this I launched the first in-product A/B testing platform at Intuit’s QuickBooks Online and led customer research at a startup called Lytro.

How many papers have we published through this work?? Guess. Go ahead, guess.


Why was this? We produced interesting studies with nice hypothesis. Our team ran dozens of MTurks studies, dozens of Google consumer studies and dozens of in-product experiments with real users.

While we would have ideally wanted to test more (more testing is usually always helpful), we really can’t complain that companies are unwilling to experiment. Companies are very much willing to test, as evidenced in the market by the high demand for tools like Optimizely, Mixpanel and impressive engineering investments in building custom testing tools.

The reasons I believe we haven’t published are less about the research idea or the methods, but are fundamental problems with the relationship between social science academia and companies.

What is the main evidence that the industry/academic relationship is broken?

Companies are regularly re-inventing the wheel with every feature, doing new research without consulting existing academic insights. When was the last time the product team requested a lit review before designing a feature? Either companies are disregarding social science research as invalid…or more likely they just aren’t reading.

And academics are not helping. They are choosing research topics by studying previous papers others have done. Many conference paper talks start by citing the gaps in other people’s research or dive into the tiny nuance left open in very well studied domain. At SPSP (main social psychology conference) there seem to be almost no collaborations done with companies. This is, even for an academic field, fairly shocking.

What’s at stake?

For companies…one could argue nothing. Right now there is very little at stake for ignoring social science Academia. They get no points for publishing a paper. This is not in their job description. On top of this, companies are hiring UX researchers who don’t have a PHD in social science (however immoral this actually is) and they are still launching successful products in the market. And, even though companies aren’t doing interesting experiments that result in core mechanism learnings, they are learning if what they tried works or doesn’t work. This is short term satisfying and confidence inducing.

For Academics…a lot is at stake if the field chooses to ignore the influence of industry. The world is changing. We have tech tools to do real world testing (and I’m not talking Mturk). We have massive and pervasive social networking platforms to measure social interactions. We have very fast test cycles. Companies are doing tests and getting results in a month (some in a day)! We have smartphones to collect actual vs. self reported social interaction, location and movement data. And most interestingly, it’s almost becoming expected that companies are experimenting on their users. It may be not be too far fetched to project that classic academic research methods will go out of style.

This means that the burden is on Academia.

Here are 6 things (social science) Academia can do to systematically improve the relationship with industry

  1. REQUIRED company internships for PHDS
    This does two things. It inspires your research agenda with market problems. And it helps you identify the actual experiment you should pitch to the company that matches their roadmap vs. force feeding your research agenda on them.
  2. INVEST in technology (apps, FB ads, smartphone sensors) as a testing platform. Hire an engineer and make something. Are you studying why and how we collaborate with strangers? Build an app that let’s people actually share an item. By creating more realistic, fast and flexible testing environments you make your research more relevant/interesting to companies that could adopt it or test it for you (and btw make yourself more desirable on the job market).
  3. SOLVE THE MOST IMPORTANT problem. Companies won’t run experiments that don’t advance their number one priority. They care about the outcomes not the learning. We have to remember that experiments take resources and may risk their brand. In order to collaborate with a company at any level you can’t come with the solution. You have to start with the problem.
  4. CREATE INCENTIVES for companies to open source their learnings. Companies are testing at an incredible rate however there is no incentive civilian marketer, researcher or data scientist to publish these findings in top journals. In order to incentivize good methods and replicable results at companies we need to provide a platform for these industry researchers to share their results. This platform should acknowledge field testing constraints, lower barrier for follow up studies and set new standards for feasible and recommended methods.
  5. SIMPLIFY your (and other’s) research to small consumable chunks. 
    Product Manager, marketers and designers are craving insights on human behavior. Jawbone and One Medical have recently distributed our behavior workbooks and Aetna has sent out our Daily Bits email to their sales staff. These companies have seen something crazy happen: their employees want more! People are asking questions about the research, questioning their intuitions, they are proposing experiments. By democratizing the literature, companies may look to academia more for collaborations vs. reinventing the wheel.
  6. Academics UNITE. Suppose a company wants to run an experiment, in partnership with Academia? How is a company supposed to find the academic who is thinking about the problem they are trying to solve? Currently there is no easy way for a UX researcher at a company, without in depth knowledge of the academic scene, to identify you. There should be an up-to-date database of academic research agendas that companies can reference.

Here are 5 things you, as an academic, can do to move into Industry.

See above. Seriously. In order to be be desirable to companies the above list applies: you must have some industry experience and awareness, be able to solve social science problems using your toolkit and drastically simplify your communications to the most important pieces.

But here are 5 tiny tips:

  1. Pick a research topic that is market viable. It is not OK to study “egocentrism in the volunteer’s dilemma” and think that you’ll be desirable on the industry job market. Try to pick a domain focus that has obvious and clear appeal in the market.
  2. Figure out your focus. Are you going to be technical? This means you know light coding, are great at R and can use mysql. Are you a UX researcher? Get experience with qual research. This is a required skill for a researcher at a company.
  3. Educate the company on your value. Maybe add the word, “behavioral” to your title. Or you can proactively propose different research methods that the company likely hasn’t explored. Teach the company how to think about you.
  4. Work with a company. Show that you can execute. Show you like applying your work to real problems. Academic hires are risky to companies (not only because you have no job experience) but also because of the stigma that researchers move super slowly.
  5. Practice explaining your work. Your research is only as good as your ability to explain it to a civilian. What is the most important learning you got and what are the implications in the world? Always talk about implications. Companies care very little about the future studies you could do, instead they care about how they could apply your work tomorrow.

…Good luck.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.