I Spent (another) $716.46 Talking to Data Scientists on Upwork — Here’s what I learned
Lessons from the top 1% of data freelancers
I paid freelancers to talk to me (again)
6 months ago, I interviewed 10 top data science freelancers on Upwork and wrote an article summarizing my key learnings. While this might sound like an expensive way to learn, I’ve found it to be an unreasonably effective way to accelerate my journey as a recent full-time entrepreneur.
There is no course or textbook on “data science freelancing.” Therefore, someone trying to break into the space mostly has to learn the hard way, i.e. by doing it and failing. While one can learn many important lessons from this approach, much of the time (and pain) it requires can be avoided by talking to someone a few steps ahead.
Was it worth it?
Data science skills are highly valued. For instance, it’s not uncommon for a data science freelancer to charge 75–150 USD/hr (more specialized consultants may even charge 200–300 USD/hr).
This means that even a 10% improvement (e.g. 10% more leads, closing 10% more clients, etc.) can easily translate into $10,000 of value over a year. From this perspective, the knowledge from these interviews is worth 10X what I paid for them — so it was worth it.
However, that was not the only upside. Here are other key (and unexpected) benefits of these conversations.
- Connections — I have maintained relationships with many of the freelancers I interviewed, which have been an enormous resource and support for me.
- Community members — Many freelancers joined a community I run (called The Data Entrepreneurs) and have led workshops on their expertise.
- Earnings from content — My previous blog has generated $470.76 in earnings (as of writing this).
Round 2
Given this laundry list of benefits, doing another round of interviews was an obvious decision. However, this time I went for “quality” over “quantity”. Which means I spent more money ($716.46) to talk to less people (4 total).
Similar to Round 1, I structured the conversations into 3 parts: Past — how’d you get started?, Present — what are you up to now?, and Future — where is this going? While many key takeaways from the last round were reinforced, new points were raised, and nuances for the past takeaways were revealed.
I review these points and nuances below.
What’s the #1 reason freelancers fail?
One question that generates insightful responses focuses not on what I can do right as a freelancer. But what can I do wrong? This brought up a wide range of responses, which I will summarize into 3 key points.
Misalignment — One of the biggest challenges when freelancing is a poorly defined business problem or project scope. This leads to miscommunications and project failures.
These seemed to (especially) be a risk when working with “non-technical” clients. The core dilemma here is clients with little to no data science knowledge often have inaccurate impressions of what is possible. Namely, the sentiment seems to be one can “just use AI” to solve the problem without a deeper appreciation of the investment required to implement AI solutions in practice.
Commit too early — Another point of failure is new freelancers may feel compelled to “give a number too early.” In other words, they might commit to a client’s desired outcome when either 1) they do not fully understand what the desired outcome is or 2) they do not fully know what it will take to get there.
A tactical tip one freelancer shared was saying, “I don’t commit to something I can’t do,” when clients press for a commitment prematurely.
Unrealistic expectations — While freelancing can provide incredible freedom and income, it is not easy (especially in the early days). Freelancers who expect too much too quickly set themselves up for failure.
This reinforced an insight from the first round of interviews: new freelancers should focus on repetitions and reviews, not money.
Find a niche*
A key takeaway from my previous article was to “find a niche.” Niching is an effective strategy employed by many freelancers. It helps give their services greater clarity to prospective clients and enables them to charge premium prices for their specialized expertise.
However, some nuances to niching were not fully highlighted in my first round of interviews. One freelancer said, “Don’t pigeonhole yourself into a single tech stack or solution. The more adaptable you are, the more valuable you become in a freelance capacity.”
This sentiment was echoed by another freelancer who stated, “A diversified consulting business is more robust at this point.”
This all boils down to the inherent risk of niching.
It works great as long as people need that specific service or expertise. However, if that specialization is no longer needed, niching can be catastrophic. Just ask any Blockbuster executive.
The moral of the story is this — niche to differentiate yourself, but don’t lose sight of other opportunities to expand your consulting business.
Learn the full tech stack (before you outsource)
Another key takeaway from my previous article was to “form alliances across the full tech stack.” The key point is that data science skills alone (e.g., training an ML model) can be limited in business impact and value generation. It doesn’t matter how good your R² is if you can’t deploy the model into production.
This is why many freelancers (from Rounds 1 and 2) advised me to not only form alliances but also learn the full tech stack for myself. While this may sound daunting, one freelancer soothed my apprehension by saying, “You don’t need to learn everything. You just need to know enough to containerize your script.”
In other words, even if you aren’t an “expert” in everything, having a working knowledge to get the job done is enough.
This is illustrated by the end-to-end framework and tech stack of one full-stack data scientist, which is given below 👇.
- Architecting Data Backend — AWS (RDS-postgres, S3)
- Data Pipeline — ECS, ECR, Kubernetes, Airbyte, Docker, R-modules
- Building Infrastructure — terraform
- Writing DS/DA Code — R
- Making Web Apps — R Shiny
2 Paths Toward Scaling
A question I posed to all the freelancers in Rounds 1 and 2 was: where is this going? For those interested in consulting long-term, 2 paths toward scaling emerged, which mirror the two career paths for data professionals I saw doing data science at a large enterprise.
- Path 1 — Manager/Leadership: less technical work, more people work.
- Path 2 — Technical Expert: less people work, more technical work.
Both Paths can be rewarding and come with increasing compensation (to a point). Through these conversations, I realized there are two very similar paths in data freelance and entrepreneurship.
Freelancer’s Path 1
One freelancer embodied a freelancer version of Path 1 above. Their ultimate goal was to scale their business into an agency where many consultants served many clients. While technical expertise is still helpful (and necessary to some extent), this approach to scaling requires more business acumen, managerial experience, and communication skills.
Freelancer’s Path 2
Conversely, another freelancer was on a Path 2-type journey. They had tried Path 1 but realized it wasn’t for them. They preferred to keep doing the technical work and not worry about the challenges of managing employees, sub-contractors, and clients.
Scaling Each
While scaling in the freelancer version of Path 1 is obvious (more clients = more $), scaling Path 2 might be less obvious. This latter approach boils down to simply increasing one's rates.
An interesting observation is the freelancer’s Path 1 is more aligned with not committing to a niche and going after demand. On the other hand, Path 2 often requires one to find a strong niche in which they can make a compelling offer at a premium price.
Going from Freelancer to Founder
For many, including myself, freelancing is a means to an end, where the ultimate goal is to build a product-focused business. This was the sentiment shared by two freelancers I spoke to this time.
While I’ve gotten mixed feedback from (successful) founders on whether freelancing is an optimal path toward product development, it does check two important boxes for those who want to launch a product.
The first is flexibility. One can turn up or down client workloads to accommodate product development time. The second is immediate cash flow. Freelancing is a straightforward way entrepreneurs can translate their skills into cash.
However, one freelancer warned that freelancing can easily “become a treadmill.” This means one can get so caught up in the consulting cycle that they don’t go anywhere long-term. That is why one freelancer (and former founder) recommended that I reserve time for strategic thinking about what I want my entrepreneurial journey to look like.
My 3 Takeaways
There aren’t many things in business that give an unreasonable return on investment. However, I feel the 14 conversations I’ve had with data freelancers near the top of the field have been one of those 10X investments.
To expand on the 4 Takeaways from my last article, here are my key learnings from this round of interviews.
- Clarity of Scope > — don’t commit until you have clarity. This will help avoid many potential pitfalls of freelance work and ensure that work provides value to both sides.
- Never stop learning (both technical and entrepreneurial skillset) — whether learning a new technology or improving how you communicate expectations to clients, continual learning is required in data freelancing.
- Find a niche, but know the risks and have a back door — don’t niche yourself out of a job, and don’t try to be everything to everyone. Find the right balance that matches your goals and market demands.
Resources
Connect: My website | Book a call | Ask me anything
Socials: YouTube 🎥 | LinkedIn | Twitter
Support: Buy me a coffee ☕️
Keep up with The Data Entrepreneurs:
👉 Join the Discord | 🎥 Subscribe on YouTube | 📅 Events Calendar
