Getting Around An Inept Leader In A Business That Doesn't Care About Data Science
Happy Wednesday and thank you for being part of the community! I appreciate everyone’s support for part 1 of this series. Let me know if these remain helpful and I will keep the series going.
When businesses hire data scientists just to say they're doing data science, they don't typically invest in its leadership structure. Data scientists can find themselves as an afterthought buried in the technology organization. They could be part of a product team or the business unit they support most frequently.
In any of these organizational structures, the person above you is probably not a data scientist. It's unlikely they understand how to advocate for you or explain the value that you can deliver. In some cases, they may impede progress.
As a result, you need tools to help you bypass that layer of leadership. This advice may sound risky on the surface, but most leaders in this position are disinterested in what you're doing. You'll be surprised by how much latitude you have. The reality is you can get away with a lot and need to use that to your advantage.
Think about it this way. Today you're not producing very much value for the business, and your ROI is probably underwater. You still have a job in this era of tightening belts and increasing fiscal accountability. That means you aren't judged in the same way that everyone else is. It's a harsh reality, but you and your fellow data team members are trophies. That comes with some perks.
Find Out Where Your Mandate Comes From
Unless you go way over the top, you have a significant amount of freedom, even if you don't feel that way. It's unlikely your boss has any say over whether you keep your job. That sounds ridiculous, but I've seen this happen in multiple businesses. The CTO or CIO is very interested in AI, and that's why the data team is there.
Step one is figuring out where your mandate and demand for the team came from. Do some poking around and figure out who developed the idea for a data team in the first place. The company line is probably somebody in middle management. Don't trust it because middle managers rarely can sign off on a decision that big. Your mandate likely comes from the top or close to it.
You could also leverage the significant hype around things like ChatGPT. Between exuberance and fear of missing out, there is a lot of perception that you can take advantage of. Even if you didn't start with a C-level mandate, you might be able to talk your way into one.
Listen for executive or C-level leaders suddenly interested in large models and what AI can do for the business. Work to position yourself as a trusted advisor for all things AI.
It's also worth holding open training sessions for anyone in the business who is interested. Teach prompt engineering, basic walk-throughs, or offer non-technical explanations about how the models work. Large models have caught the public's interest and attention. Use that to start conversations.
Takeaway #1 is not to feel stuck under the leader you are reporting to.
Start With A Product Approach
In many companies, data science has a branding problem. The job title might be the sexiest of the 21st century, but what the rest of the business thinks you do is far less glamorous. It's called reporting. As long as the rest of the company looks at the data team as over-glorified Excel jockeys, they will never rise to achieve their true potential.
Rebranding takes the focus away from reports and data points and pushes it toward products. Just the terminology is enough to begin the shift in thinking. The purpose is to get the rest of the business to look at the data team as a value-generating business unit. Reports and data aren't seen in that light, but products are.
This begins the shift from a cost center to a growth driver. Data and model-supported products can live up to the substantial hype, but not if the data team is stuck fielding data requests and getting pulled away from more valuable work.
It's not just a change in language; it's a shift in focus for how the data team operates. Adopt the motto "Production grade from day one." That should become the team's new focus. Start initiatives with two questions. What is the smallest increment of value that this product could deliver to users? How fast can I/we get that product built?
You're aiming to discover initiatives that can deliver in four to six weeks. That's typically all the buy-in you'll get at a company that doesn't care about data science.
It doesn't seem like a lot of time, but you can produce significant value in short order when you change from thinking about delivering data to delivering products. Products are built incrementally, not all at once. The concept of a minimum viable product is critical for data teams to adopt. Think of a team as a miniature startup within the business.
Startups need to bootstrap and work on shoestring budgets. The data team needs to function the same way. In this case, your budget isn't money; it's measured in time. Think of C-level leaders like VCs looking for significant returns on investment. Look for products that are home runs, not incremental gains.
The new focus should establish the data product as a central artifact for the team. This will help you significantly with your leader. The person you or the data team reports to is interested in maintaining their role in leading the team. They probably care far less about leading the product or strategy. This can be your opportunity to drive a compromise.
Once you have established the product as the central artifact, talk to your leader about taking ownership of product strategy and management. The organizational structure doesn't change at all. What you've gained is control over prioritization and your value creation. It's a power you can leverage to go from cost center and afterthought to significant cost savings and revenue generators.