5 key tips for data scientists in mixed-method teams

How to bridge the gaps and work well together

Chris Williamson
Data science at Nesta
5 min readDec 6, 2022


A team of people participating in an ideation session, sticking post-it notes on a wall
Photo by Jason Goodman on Unsplash

If you’re a data scientist, it’s rare that you’ll be working alone or just with other data scientists. It’s more likely that you’ll be working in a team with people using other methods. This inevitably means that it takes a bit of effort for everyone to get on the same page and work together effectively.

At Nesta, data scientists work alongside practitioners from other disciplines (such as design and behavioural science) on projects with positive social impact. Often we’ll do this in partnership with other organisations and people we’ve not worked with before, so it’s important that we get collaboration right. I asked my fellow data scientists about the key things they’ve learned from working in this way and compiled their answers into five key tips.

Every organisation is different when it comes to team structures, the types of projects they work on and the different disciplines that are involved. But the following advice should be relevant to any data scientist working alongside non–data scientists and help you to collaborate effectively and achieve your goals.

1. Identify your teammates’ needs

Your organisation may have paying customers, but if part of your role is to provide internal insights then your teammates are your customers too. Consider their needs just as a product owner would consider a user’s needs when prioritising product features.

Rather than just ploughing ahead with the data work, consider how to structure it so that you can feed into what the rest of the team are doing. For instance, completing a quick piece of analysis could provide insights to help a designer start prototyping, or nailing down what form your outputs will take might unblock a developer.

Try to be involved throughout the whole process by taking part in planning meetings, design sessions and wider conversations. This will help you spot opportunities to apply impactful data science that you might otherwise have missed. And remember to stay open to new ideas from your team and be ready to adapt your approach when needed.

2. Set expectations around working styles

As a data scientist, it’s important that you have uninterrupted time to get your head down and concentrate on technical work. But it can be really frustrating for your teammates if you’re never available to respond to their questions or participate in discussions.

Rather than shutting yourself off completely, have an open conversation with your team about what you all need to work effectively and how you can achieve a good balance between “focus time” or “deep work” and “available time”. You may want to agree on specific times when you might be less responsive and make these visible to your teammates, or schedule synchronous co-working time with your team to maximise collaboration.

3. Don’t sacrifice rigour

Even though your team might want to work quickly, there are a few things you shouldn’t let slip. Methodical data cleaning, documentation, code reviews and model testing are all crucial for making sure your results are reliable and reproducible. Data science is different to some other disciplines in that a small error or an incorrect assumption might lead to major changes in your outputs, so it’s especially important that your team understands that these tasks are key parts of the process.

Make sure that you leave enough time to be confident in your outputs before committing to deliver them. If you want your team to have some rough and ready insights, make sure they’re clearly caveatted — and perhaps ask for them not to be shared outside the team until someone else has reviewed your work.

4. Teach others…

Team discussions are a lot easier when your teammates understand what you’re doing. Spending a bit of time laying the groundwork makes it easier during planning meetings to explain what you’re working on and how long it will take and you’re more likely to get high-quality outputs when your teammates understand the strengths and limitations of your approach. What’s more, in my experience non–data scientists are often very interested in learning about data science methods.

But some things can be difficult to explain — an explanation that’s clear for one person may be confusing to another, so keep in mind what sort of technical backgrounds the individuals in your team have. It’s often better to describe concepts with analogies, examples and pictures, and without using too much complex terminology. For instance, you could explain a clustering method by drawing comparisons with how similar items are grouped together in online shopping recommendations.

While it can take effort to properly explain a concept, it can be rewarding to get a different perspective from someone who’s new to the topic and they’ll often ask insightful questions. It might even help clarify your understanding of it as well, helping you to spot some misconceptions or knowledge gaps that you didn’t even know you had.

5. … and learn from them

Finally, one of the things I personally enjoy most about working in a mixed-methods team is getting the opportunity to learn from others.

Many skills and methods cut across team boundaries and are applicable to data science as well. For instance, observing a designer’s process might inspire you to take a more iterative and human-centred approach to modelling. Rather than just passing a trained model on to a developer, you could learn about what it takes to deploy one in production to make the process as smooth as possible.

As agile expert Emily Webber points out, developing skills from other disciplines can lead to better collaboration, helping you to empathise with your colleagues and understand the challenges they face. And having skills from outside of your direct area of expertise can make you a valuable asset, setting you apart from the competition when it comes to career progression.

Do you have any other tips for data scientists working in mixed-method teams? Feel free to comment below.

With thanks to Julia Suter and Emily Bicks for sharing their insights.