How a business can get started with "doing data science"

How can a business get started once they've decided they want "data science"?

What should a business do if they believe they "want data science"? Most businesses recognise that they have data lying around and they're not doing much with it. Perhaps they already have some accountants doing financial analysis, but want to expand this capability.

What are the options?

Option 1 - hire a consultant to deliver a specific project

Sometimes a project is so urgent that it cannot wait for an internal capability to become available; a short-term fix by an expert is needed to deliver it. This model is tried and tested and has its merits. It is plausible that a single project is important enough to warrant the higher outlay, and the absence of a long-term vision or data strategy means hiring a permanent is not yet the right choice.

Cost: high

Impact: short-term

Pros:

  • Quickest impact
  • No long-term hiring needed

Cons:

  • Will need to repeat for the next project and ongoing maintenance
  • Costs scale linearly with the amount of work needed

Alternative: have a consultant on retainer to either do prototyping work from time-to-time or in an advisory capacity.

Option 2 - train their own employees with the data skills required to do the work

One drawback of hiring an external consultant is they might need time to get up to speed on the specifics of the industry and business in question. Domain expertise is a key skill in data science; you can't make an impact without it. An alternative therefore to hiring a consultant to deliver a project is to empower the company's existing domain experts to deliver some of the work themselves. Depending on the complexity of the project it may be best to leverage the subject matter expertise already present. There is an initial investment in upskilling, but for more strategic, medium-term projects, this may be the preferred option, as it will permanently increase internal capability and require fewer external resources for future projects.

Cost: medium

Impact: medium-term

Pros:

  • Tap into existing domain knowledge that external consultants or new hires may not have
  • Investing in existing staff by upskilling

Cons:

  • Training (and embedding of knowledge) takes time, so impact may come along later

Alternative: hire a consultant on retainer to work alongside domain experts while they upskill and collectively deliver the first project at the same time.

Option 3 - hire a data scientist or a team

For the biggest, most long-term impact, the strategic choice may be to actually build a data science capability by hiring a data science team. The size and make-up of this team will be strongly dependent on the needs of the business, but an important thing to remember is data science cannot exist in isolation. There must be a data strategy of some description which supports this function. There needs to be a data infrastructure, ideally even a data engineering team, which supplies data for the data scientists. There also needs to be a vision for how exactly this team will make a measurable impact. Without these, a data science team alone cannot fulfil its potential, and this should be considered when embarking on the creation of a team.

Cost: high

Impact: long-term

Pros:

  • Permanent in-house resource for data science problems

Cons:

  • Impact realised long-term, since hiring + onboarding takes time
  • Requires additional support, e.g. a data strategy and infrastructure
  • Company may not need a permanent data science resource

Conclusion: every company is different

Those are some of the main options for delivering a data science project, but when deciding how to make the move into data science, naturally the right answer will be specific and tailored to each company's needs. Here's a handy breakdown of what I just discussed.

Impact Option Investment
Long-term Hire data scientist(s) £££££
Medium-term Train your employees £££
Short-term Hire a consultant ££££

If you're interested in a chat to explore your options, let's talk!

About David

I'm a freelance data scientist consultant and educator with an MSc. in Data Science and a background in software and web development. My previous roles have been a range of data science, software development, team management and software architecting jobs.

Related articles tagged Data science

Citizen Data Science

Data science

What does "citizen data science" mean beyond being a buzzword for "everyone should do data science" (which they most certainly shouldn't)?

How a business can get started with "doing data science"

Data science

How can a business get started once they've decided they want "data science"?

How to use your impostor syndrome to learn anything

Data science

How can you harness your impostor syndrome? Get good at being able to learn anything.

Contact me

Please give me a brief idea of what you're looking for help with, and I'll get back to you very soon. If you're having trouble with the form below, you can also email me at hello@davidasboth.com

Twitter LinkedIn