Effectively Managing Data Teams

Read this article
⸻ Beyond Data

5 practices to drive more impact

Prompt: Multitasking (made with Midjourney)

Data teams can be hard to manage due to their wide scope of responsibilities and required skills. There’s no understating the difficulty if you’re a data professional that’s been given management or leadership responsibilities for a data team. If you’re not a data person – a business leader or senior manager from another background – then I can’t imagine the confusion.

The thing is, many people and organisations are sold on the hype of data science long before they realise the reality. You hear about the big wins and advancements from such a company or the latest technology advances in some new blog and the FOMO kicks in.

How do you get in on the action?

For many organisations, that usually means jumping feet first into hiring a new data and analytics team – but aren’t really sure how to set them up for success.

Or maybe you’re part of an established team that’s bogged down in the business as usual, not getting enough breathing room to explore these new things and grow.

I’ve seen this time and again, both in companies I’ve worked for and customers I’ve helped. Over the years I’ve learned there are a few key things that can make all the difference.

Hopefully, they’ll help you.

Integrate domain experts

I think the most important advice I can give is to get subject matter experts into the team. Whatever the current project your data and team are working on the most vital information they need for their insights and outputs to be useful is in the head of someone else in the business that knows the domain inside out.

So many times I’ve seen super-talented teams struggle to make their new model or dashboard have any impact or waste significant time and effort wrangling data because the domain they just don’t deeply understand the domain.

Successful data and analytics is more about feedback than anything else.

If the current project is to deliver a dashboard to the accounts team, get someone from the accounts team into the project delivery. If it’s to build a new model for the senior leadership, get a representative from that stakeholder group to sit in on weekly feedback sessions where they can answer questions and steer further development.

The more integrated, open, and regular this two-way communication can be the better. Ideally, you want it to feel like they’re part of the team.

Bonus prize: you have someone outside the data team that’ll act as champion and translator for the project to the rest of the business.

Set clear goals

This might sound ridiculous but so many data projects kick off, destined to fail, with very loose, ethereal requirements and goals.

“Help the team become data-driven”

“Build a forecast model”

“Improve this dashboard”


These things don’t really mean anything concrete and are open to too much interpretation. Sure there’s the famous Jobs quote about hiring smart people and not telling them what to do – but you need to give them some idea of what the goals are and what’s important to the business.

Make sure your goals are concrete and easy to understand.

Failing to do so will lead to your data team guessing about what’s best and how important it is. When that guessing leads to a lack of results, managers tend to panic and then the micromanaging begins.

Ensure progress is measurable

Closely linked to the previous point – always make sure the progress on a project is measurable. This can feel impossible sometimes but that’s usually a symptom of the task being at the wrong granularity.

Think about how you can break your goals into measurable outputs and then make those measures clear for the team. This achieves two things:

  • it enables the team to deeply understand what good looks like and how to check they’re heading in the right direction without the constant need for wider input
  • the measure becomes motivating for those involved, like a high score in a game

Exploration, learning, and research tasks can feel very difficult to measure progress on. I’ve always been a big believer in artefacts to combat this. Make the measurable output a report or talk about what new insights or learnings have been made. Having a standard template can help. The measure then becomes delivering or contributing sections of these reports. Over time, the knowledge gained by those involved becomes builds into a shared understanding that the whole business benefit from – not just locked in someone’s head.

Link progress directly to ROI

Data professionals speak in terms of technologies, query speeds, precision, recall, ROC, confusion matrices, joins, semantic models etc. Use this language with some stakeholders and watch their eyes glaze over with boredom.

Every profession or domain has its own language and jargon that’s difficult for outsiders. There is a language that the whole business understands:

Everyone speaks money and time.

If at all possible, convert your progress directly into terms of how much additional money or time the project earns or saves. Do this in temporal terms that are meaningful to them on a personal level too – think per week/month/quarter/year for the most impact.

If you can convert everything to these simple measures, it becomes much easier to compare the impact of projects against each other.

Now a word of warning, this is probably the most difficult of these steps to achieve and sometimes it isn’t possible. If you’re struggling reach out to the community or myself (I’d love to hear), as I bet someone out there’s faced a similar situation.

Give them autonomy

The above points are really about getting the right things in place to enable the team to go at full speed, without getting lost or going off track. One thing that will significantly hinder their progress though, is the reliance on others.

Whether this be needing approvals from management or relying on other teams to get the technologies spun up to deliver the project – it all introduces friction. Every touchpoint can become another bottleneck, or worse, a well-meaning suggestion from that collaborator turns into a huge distraction that derails progress.

Getting full autonomy can be difficult – but remember this is for the data TEAM, not the individual. Need IT to be involved to get tooling deployed? Sounds like someone from IT should be in the team. Need a senior manager to approve access or budgets? Sounds like they also need to be in the team. Having some people tied to the team on a partial basis also helps with some of the points from the first tip above.

When the team have all they need at hand to get going and keep going – you’ll see a step-change in their pace of delivery.

Final thoughts

None of this is easy. And it should be said that it’s rare to find all of the above on one team.

If you’re on a team that’s struggling or just not going as fast as you know you could – I highly recommend petitioning your management to incorporate some of the above.

If you’re running a data team these tips might make the difference between good and great. Many of them have natural side benefits of making your team happier too!

Good luck.


If you enjoyed this, consider sharing this newsletter with others that would find value from it or follow what we’re doing at Hypercube


When you’re ready, there are a few ways I can help you or your organisation:

What do you think of this issue?

I’d love your feedback to ensure I’m writing about the topics you want to read, let me know if this one hits the mark

Want to write for Beyond Data?

Have something you want to share? Interested in writing an issue or working with our team?