A guide to getting started and doing well at consulting and freelancing, and avoiding some speed bumps. This is aimed squarely at people in academia who are good at working with data.
I am an academic. I am very well suited to life in academia. I love every part of it, including teaching, and have very little interest in leaving it. This is true of my attitude and interests, and remains true despite the fact that I am unlikely to ever be published in a top journal in my field, and also that I have maintained, while in academia, a pretty lively string of consulting and freelancing assignments.
In this post I will share with you some guidelines and tips for pursuing consulting and freelancing work while inside of academia, either as a grad student or as a professor. This will be targeted at people with a skillset similar to myself: people with lots of technical and analytic skill when it comes to working with data. It will also heavily suggest doing things in the exact way that I did them, and in the United States, because that’s what I know. I will not pretend that this is the only way to do things. But it is one way.
I have decided that the most appropriate tone for a post like this is “world-weary” and I’ll do what I can to rise to this standard.
You’re busy as heck. Why should you consider doing any sort of consulting?
That said, yes, this will eat into whatever free time you have, and you’ll want to make sure you’re doing well at your main academic job as a prerequisite for spending time on other stuff. Reviews at my academic job suggest I’m not skimping there. If that started to slip I’d reevaluate.
In my case, it’s definitely true that I work all the time and have had to largely excise everything but work and family from my life. But I’m okay with that. I say that not as a brag or judgment on how much people should work, but just an honest statement that I am pretty happy with a work/life balance the internet tells me I should not be happy with. I enjoy working. If you enjoy working less than I do then you should work less than I do. Consulting, conveniently, does let you scale how much work you’re doing, as long as you’re willing to say no when you have too much.
If so, you’re good to go, at least conceptually. Technically you’ll be doing yourself a favor if you’re skilled in common private sector languages like Python and the various flavors of SQL (or, more importantly, can pick up new tools and languages on the fly). Or learning to do stuff without relying on fancy models. Or using Excel.
There are some parts of freelancing that have to do with changing the mindset you have around the work that you do (see the final point above). This is the kind of thing you’ll just have to get through practice, though, so I won’t bother trying to preview it.
I find jobs largely on upWork. There are other places with freelance job listings, too. If you can manage to get yourself hired directly by a company that needs someone on an ongoing basis and will pay you a decent rate, that’s fantastic. I have, on occasion, built contacts through upWork that later turned into far more lucrative contracts direct with a company that has deep pockets.
You might wonder about the value of something like upWork given that they take a relatively hefty cut (20% for the first $500 with a client, 10% afterwards). But it’s worth it. The platform provides a lot of clients. More than that, they make sure you get paid. If you work with someone one-on-one, who knows what happens. Wait a few months to get paid, never get paid, spend a lot of time chasing payment, etc.. upWork forces them to pay once you’ve agreed to something, I’ve had literally two contracts ever go south out of hundreds (and one of those went south in my favor, so hey!), and that’s absolutely worth 10% on its own.
upWork isn’t where I got started though, nor is it the only place I get work. I started out my freelancing focusing on education rather than data work. I did tutoring through Tutor.com and Chegg Tutors, I wrote a zillion practice questions for Albert.io, I wrote practice questions for textbook publishers. I generally don’t advise this path for people situated in academia. This work is rote, time intensive, and poorly paid relative to working with data. Not to mention exposure to actual students in a freelance online-education setting means you’ll spend the majority of your time dealing with people begging for you to just do their homework for them instead of getting paid (or giving in to those requests, and I hope you have more dignity than that). If you’re a grad student and don’t think you have the data chops yet, then sure. This stuff is at least low-commitment and can earn you a few bucks.
Better opportunities outside of a job board like upWork often require connections of some sort, or being known for something. I’ve had a number of great opportunities come from being referred by a client I met on upWork to another client not on upWork. I’ve gotten hookups from friends who have gone into the private sector. I’ve also had companies email me to offer work. When you’re a professor it happens! You might also come in contact with middlemen consulting agencies who link up freelancers with bigger companies. I’ve worked with a few such agencies, most of which hired me the first time off of upWork. All these experiences have been good. They have access to bigger clients that you’d be unlikely to find on your own. Good pay and longer-term projects.
Some of the most stressful and frustrating, but financially rewarding, work comes in the form of legal-case consulting. Remember the stupid fantasy money I mentioned in data work? Well, being an expert in a legal case, or just offering commentary and advice to one, is super-duper stupid fantasy money. If you’ve ever been on a jury for a trial with experts, you know lawyers love to ask the opposing experts what they get paid, and for good reason - what juror wants to listen to someone getting paid ludicrous amounts to be there? I’ve done two legal cases, one which came to me through upWork and one which was an email request, and each time charged several hundred dollars an hour. This is low for what this usually pays.
You do have to put in the work, though. Lawyers expect extremely fast turnarounds, there are often restrictions on the scope of what questions you can answer or what methods you can use to answer them (if you’ve ever read an expert opinion for a legal case from an academic you like and thought “wow, this is not as good as their usual work,” this is one reason why), and you might have to give deposition or testimony which can be a little terrifying. Plus, while this isn’t something I’ve experienced myself, you might end up in a situation where the lawyers really want you to find a way to make their side right, rather than providing your actual best expertise, also then also put your name on it. The legal-case work I’ve done has all been in the form of rebuttal to other experts which makes some parts of this easier but not all of it.
On a much smaller scale, I will, as someone with a mildly public persona via Twitter and YouTube, occasionally get requests straight to my email from individuals. Many of these are requests to help in general without pay (which I’m happy to oblige, up to a point), but others are offers of work. These might be worth it if you happen to have the time, but I don’t think I’ve ever had one blossom into a truly worthwhile contract (see the note about “small contracts” below).
So how do you get work, in sum? At least starting out, if you don’t have any connections your best bets are going to be a job board of some sort, or just waiting for the right opportunity to hit your inbox. It really does happen!
When you are starting out, you’re probably going to be starting small unless you have some sort of hookup to a major project. This means building a good profile on a job board like upWork and accepting small jobs that you only expect to pay a few hundred dollars in total. At this point you can start working pretty quickly, but it’s unlikely that you’ll be making this whole venture really worthwhile right away, given the opportunity cost of your time. But you’re building a profile with some hopefully satisfied clients, and getting used to how these consulting jobs work.
At this point you’re giving yourself a crash course in the parts of consulting that your academic and data skills don’t cover. You’re building negotiation skills and client-management skills. You’ll want at least a little of that under your belt before taking any real big projects. Try to focus on taking projects where your skills are handy (i.e. sure you could proofread a college kid’s essay for a few bucks but that’s not really teaching you much in terms of building to something bigger), and that you can turn around quickly. Small projects with well-defined goals are great here. By doing these you’ll be able to build a set of positive reviews (if you’re on a job site) and a general knowledge of how this work operates without getting bogged down by any one project. You’ll also get to know your own work habits and speed better. How long will that project really take you? Do you really have time to do it? You’ll develop a much better internal sense of this.
Once you feel like you’ve got a handle on it you can go for bigger projects, and be more likely to land them because of your track record and improved ability to sell yourself to this audience. Bigger projects are generally better once you know how to handle them (see How Can You Deal with Clients below).
How much should you charge starting out? Don’t undersell yourself. If you’re on a job board, check what other people charge. You may be surprised. If you’re just playing around and trying to get a few small contracts under your belt, sure, charge fifty bucks an hour or something. But once you’re actually trying to earn money, or being asked to take on something that will actually compete for time with the rest of your busy schedule, you should charge what the market will bear! And the market will bear a fair amount. $100/hour is a pretty low-end amount for an academic with data skills. I charge different clients between $90 (for long-time clients with consistent low-difficulty work) to $150/hour (for new clients) these days, but I already get more offers than I can handle and so that $150 could probably be higher. Keep in mind that clients will often take too-low wages as an indicator of poor quality. Maybe it’s just the upWork algorithm but whenever I raise my price I get flooded with new job requests. A high hourly rate will also deter people who do not actually want to pay you (see How Do You Deal with Clients? below).
Whatever hourly rate you land on, if you come to a project that pays for the final deliverable rather than hourly, you should think of that project in terms of how much time it will take you and charge based based on your hourly rate, plus a markup since you’re taking the risk of the project requiring more time than you expect. You can even be explicit about it. I will often tell people asking for project-based payment “My hourly rate is $\(X\). I expect this will take \(Y\) hours. So I will charge \(X*(Y+Z)\) to cover the possibility that it will take longer. Or if you prefer, we can do an hourly rate, and maybe you won’t need to pay for \(Z\).”
One last thing - in the “figure this whole consulting thing out and get established” phase, since the projects you’re taking are smaller you’re likely to stumble upon some odder or more fun ones. Go for it, this is your time. You can definitely afford to cut your rate for someone offering you an evening of entertainment. This was the stage of my freelancing when I was definitely happy to write assignment guides or check someone’s answers on their game theory homework, because I think game theory homework is fun and I miss being assigned it. This is the person you’ve chosen to read an article by and take advice from. (Note: when a student contacts you offering money to solve their homework or write their essay for them, obviously not something an academic should be doing, instead offer to write them an assignment guide, which includes no solutions but is a set of tips and nudges in the right direction to help them solve the assignment themselves. A surprising number of them will take you up on it).
Managing (and selecting) clients is a task that is really very different from something a typical academic is going to run into. Their expectations, desires, and ways in which they will take advantage of you are entirely different from the expectations, desires, and advantage-taking-ways of an advisor, boss, or purse-string-holder in academia.
And finally, as time goes on, shift towards avoiding clients with small one-off projects more and more. There’s nothing wrong with these, but new projects require setup, price negotiation, figuring out the goals, learning the ropes, etc.. You’ll be using your time more effectively, and also have much more predictable responsibilities and income, with bigger projects. These bigger projects can run from month-long projects that pay 1-2 thousand dollars to multi-year engagements as the resident data consultant for a small company. So once you have enough big projects to fill your clock, avoid taking on small ones.
There are lots of things to consider when handling clients - being kind and prompt, returning work quickly, not letting them take advantage of you, and so on. But the most difficult and important part about working with clients, and what I’ll focus on here, is handling their expectations. This includes what their expectations are of you and also what their expectations and desires are for the work you will give them. The reason this is so difficult is that private clients often have very different expectations and desires than you’ve been trained to think about in academia. So there’s a minor culture shock.
That’s your first goal, then. Figure out what it is they actually want. As a general rule, private clients are more interested in results than they are in methods. What are the takeaways? A reasonable data-driven hint as to what they should do is usually way more valuable to them than a caveat-filled conclusion that you’re absolutely certain has been supported by the data and buttressed against any possible criticisms.
With that in mind, you get a little bit of freedom to not be quite as rigorous as you would be in an academic setting. But this also means that you’ll be doing a little dance of figuring out exactly where the line is between “result that’s not quite as robust as something I’d try to publish in an academic journal” and “result I don’t believe at all.”
This leads us back to handling client expectations. Letting them know what you can and can’t do with the data that’s been provided is key, and will help keep them from expecting more of the data than can actually be achieved.
Doing all of this properly requires understanding how clients think about the kind of work you do, a topic about which I had a Twitter thread long ago. Many private clients think of statistical analysis as a product they buy in a store that solves a problem for them, rather than as a tool for unconvering some sort of truth.
This means that, at least initially, many of your clients will be expecting you to provide results that justifies a particular already-made choice, or looks good for them in some way, or provides a very clear and distinct answer to a super-important question.
Obviously, results from data do not always fall this way. This can lead to problems. Clients will ask for a zillion variations until some result implies what they want (whether or not the result is mush by that point), they’ll ask for a refund, or, worse, they’ll string you along with revisions until something pops out that they like.
In my experience, the best salve here is to simply be honest. Remind clients that if they already had a conclusion in mind there’d be no need for data analysis. Tell them that, yes, in your expert opinion this really is what the data says. This approach doesn’t always work, but it works sometimes, and I’ve yet to find a better alternative.
Letting clients know what they will, and will not, get from you is a pretty good idea in general. Yes, you can build a predictive model. No, you can’t guarantee it will boost sales.
There are a bunch of other skills that you’ll learn in the process of doing consulting work that will help you manage clients better, but I’ve been thinking through them and for the life of me I can’t think of any that I could adequately inform you about through text. So you’ll just have to work past those issues yourself.
one thing I have left out, though, is managing academic clients. You’ll actually get a lot of these, or at least I do. This makes sense - you’re an academic. They need academic skills. Academic clients range from students wanting help with their assignments up to other academics basically hiring a postdoc but with lower commitment.
Academic clients can be great clients! But some things to keep in mind:
That all said, academic clients can be great, and I will in many cases take lower payments for them. One aspect of freelancing I always enjoy is when grad students with terrible advisors come to me for help. I’ve never worked at a university with PhD students, but I’ve always wanted to be a dissertation advisor. This fills that lil’ gap in my heart. I don’t know where these grad students come up with the money to pay an advisor on an hourly basis (even at a discounted rate), but I’ve been the de-facto dissertation advisor to five students, four of whom ended up with PhDs (plus, served on the actual committee for a sixth, who for the record did not have an awful advisor and brought me in for coauthor-level work on their stats). Loved every minute of it.
Also, five of you out there are just truly awful PhD advisors, and have your name as dissertation advisor on a student’s CV when you didn’t do jack, I did. I do know who you are.
You’re a busy person who already has a job. Consulting isn’t going to work for you if you can’t manage your time properly so it doesn’t get in the way of your main job (or the rest of your life for that matter).
When thinking about how to manage your time, there are three big things to keep in mind. The first is in taking on the right amount of work, the second is in taking the kind of work that gives you flexible scheduling, and the third is in avoiding project creep.
Taking on the right amount of work. You should have a decent idea about the amount of time you have available to dedicate to consulting. Then, don’t take on more work than will fill that time. This would be easy, except of course that hitting the sweet spot of how much work you’ll take on can be very difficult.
This is not the most original of advice, but it’s a good idea to look over a project carefully before committing to it, thinking about how long you expect it to take you, and then realizing that it will almost certainly take you longer than that. Pad your expected contribution, because you have no idea where surprises will lurk for you! This is important both for managing your time and in quoting prices to clients - anyone asking for an expected amount of time something will take you (if they’re paying hourly) or a quote (for project-based payment) should get a number that is at the very least 20% above the time you actually think it will take you.
Taking on the right amount of work also means favoring projects where you have a better idea of where the endpoint is. Projects that change and expand as time goes on make it very difficult to manage your time (see avoiding project creep below).
Pursuing flexible scheduling. The easiest way to fit consulting work into a life that already has an academic job in it is to do work that can be done any time. This means avoiding work that requires lots of scheduled calls or that otherwise have to be done during business hours.
This also means limiting the number of projects you have with very tight deadlines. Outside of grant-writing, deadlines for academic work are often suggestions - conference submission dates get pushed back, referee reports languish, the edit you told your coauthor would be done next week is, uh, well. There are plenty of hard deadlines in academia, university admin deadlines and so on, but not so much for academic work. If you need more time to do it right, then take more time until it’s done right. Private clients do not usually see things this way. If there’s a deadline, you need to meet it.
Tight deadlines do not give you flexibility, though. If I have a week to do a four-hour project, I can do 45 minutes a night and be done easy. If I have until tomorrow, well that’s my day. Flexibility out the window!
So unless it’s a really good offer (like, say, a legal consulting gig), don’t take projects that you’ll have to crunch to finish. And if you have some control over the deadline, don’t feel bad about pushing it out. Remember the padding for how many hours it will take you to do the project? Well, take that padded number and think about how many days it will take you to cobble together that much worktime. Then pad that.
It does not feel good to have to push back working on a research project you’re excited about because you gave a client a too-short deadline. So don’t put yourself in that situation!
Avoiding project creep. Certain projects have ways of expanding as they go along. Sometimes this is fine. Projects that go well get sequels, so to speak, and if it’s work you like, and you can plan ahead for those sequels, then great! That’s just a return customer. But some projects expand in bad ways that can leave you frustrated and eternally unable to bid farewell to the project. Unfortunately it can be tricky to catch these up front a lot of the time, but there are warning signs.
One major warning sign is if you are going to be working with data that you can tell is super noisy or just isn’t very good, or are being asked to use bad methods. Data analysis works to some degree on a garbage-in-garbage-out principle. In my experience, clients are very willing to push back on “garbage-out,” and will tell you that results aren’t plausible and need to be fixed (good instincts born of deep contextual knowledge!), but they will often fail to accept “garbage-in” as an explanation (bad instincts born of a lack of data-analytic intuition). Thus begins the project creep where you are asked to try an ever-expanding list of alternate methods so as to find the one magic configuration that produces a believable result. All the while the entire project annoys you because you know the work you’re doing is very, very bad. Plus, if you agreed to project-based payment, you’ll quickly hit the point where the project is no longer a good deal for you, but the marginal cost of just one more variant is low enough that you keep just hoping it will end and they’ll pay up. Worst-case scenario is they try not to pay because they don’t like the result.
You can avoid some of this by building a maximum number of revisions into your project-based contract. This is harder to squirm out of if you’re on an hourly contract though. What you should really do is do a very careful assessment of whether the data is likely to produce anything believable, and back out if it’s not up to the task. You can try saying something to the client ahead of time like “hey I’ll run this if you want. It will probably be garbage though,” but almost every time I’ve done that they’ve said to go ahead anyway and the same cycle has occurred.
On bigger projects where you’re a consultant working with a team, you can avoid some of this project creep by being sure to talk extensively about the data directly with the person who knows the most about it. The number of projects I’ve had to redo because the middleman between the data person and me didn’t explain something properly and I had to redo everything at the last second!
As an aside, you will also often find it worthwhile to anticipate project creep before it happens, and prepare yourself for it. The only advice I really have to this end is to always write your code in a way that’s easy to expand upon. Private clients, just like Twitter commenters, love to look at your final result and say “neat, but how about you look at that same result in these twenty different subcategories?” At that point you’ll be kicking yourself if you haven’t already written your code in such a way that you can just loop over it to get results by subcategory.
This section is completely US-centric. It seems important for this section to point out that I’m not an accountant and this is not, like, professional tax advice or anything. So, really, double-check everything I’m telling you against an official source. My only credentials are that I have been doing my own taxes a long time and know my way around this stuff pretty well.
Consulting income is indeed income. If you’ve only ever worked, like, actual jobs before with W-2s and such it can be easy to forget that that one-time project you did in May that paid like $800 is also income that you have to pay taxes on. In fact, there are a few things that might change about the way you do taxes if you start doing substantial amounts of consulting. For one, you might not be paying taxes just once a year any more (so if you start getting real into consulting, don’t think you can just wait until April and tack on an amount and see what happens, you may be in for some late fees).
Things to note, especially if you are doing your own taxes:
Consulting! Freelancing! Try it out. If you have the time, sign up for upWork and do a few jobs, see if you like it. You may get a lot of interesting and educational experiences from it, too, in addition to being paid. It doesn’t even have to become an enormous part of what you do, since you can largely set your own limits on how much you work. I’ve found it pretty rewarding in many ways, and hopefully this article can help you avoid some of the headaches I ran into.
Don’t do more of it than you want to. But also don’t do less of it than you want to. Do the correct amount of it. This is useful advice.
For attribution, please cite this work as
Huntington-Klein (2021, Nov. 19). Some Thoughts: Consulting and Freelancing in Academia. Retrieved from https://nickch-k.github.io/SomeThoughts/posts/2021-11-19-consulting-and-freelancing-in-academia/
BibTeX citation
@misc{huntington-klein2021consulting,
author = {Huntington-Klein, Nick},
title = {Some Thoughts: Consulting and Freelancing in Academia},
url = {https://nickch-k.github.io/SomeThoughts/posts/2021-11-19-consulting-and-freelancing-in-academia/},
year = {2021}
}