BD #1 - New Beginnings

Welcome to the first issue of Beyond Data - actionable advice for building data and analytics teams for the week ahead.

Firstly, thanks for joining me - in this newsletter I'll deliver some thoughts and tips that you can hopefully turn to good use. I'll also highlight some of the more interesting stuff that's passed through my little corner of existence.

This week: 4 essential steps to building a successful data and analytics team

The old way to do this seemed to be something like hiring in a data scientist and giving them some poorly defined ML goal. They'd then be set loose on your disaster of a data landscape and both sides would get frustrated until they left.

Don't do that.

If you're that data scientist or you're going to be building a data team, I'd advise you check these steps off at the very least.

1: Engage with experts

Before committing to expensive hires or risky projects ask yourself this - do you actually know who the end users and domain experts are? What is the business question they're trying to answer? Why is data going to make a difference?

So many people outright don't consider this when they start.

You'll find people that have lived and breathed the problem have incredible insights into what will and won't work - before you ever talk about the tech.

Make sure you involve them in the kick-off process at the very least (then hold them close throughout if possible!).

2: Develop foundations

Don't start with ML either. Most of your business questions can be answered, at scale, by analysts (sometimes needing the support of a data engineer).

Machine learning is hard to get going, difficult to scale, challenging to put into production, and not guaranteed to work.

You're much better off getting a talented analyst in to get to the bottom of your business questions and explore the data. That person can then start all the hard work of figuring out how to get your data landscape in place and identifying where things like ML might have an impact.

3: Incremental > Monumental

This is probably the thing most people get wrong. Overenthusiastic folks see the vision of how a large-scale ML project might save the day and conquer the world.

They plan some huge projects and over-commit. After a few months, things get off track and ultimately, little value is delivered.

With any data project, always ask - what's the smallest step I can take to prove some value? Then deliver that. And increment to big!

4: Establish cross-disciplinary teams

Don't don't don't silo your data team.

It's mad how many organisations put all the data folks in a cupboard and occasionally slide some project requirements under the door. This, inevitably, leads to a lot of unnecessary back and forth due to miscommunication.

The better approach is to help both sides become more familiar with the other. Your techies will appreciate developing their domain knowledge and your subject matter experts will love learning data (well most will). Do this by making the team cross-disciplinary or at the very least force regular interaction and feedback so they can stay close.

So what next?

I recently left my role as Head of Machine Learning Engineering at Origami Energy.

I had a great time at Origami. They're tackling the data problem for the ever-growing renewable energy sector.

Being a part of that journey has been incredible and my time spent as Head of MLE has taught me a lot.

I'm taking the experience and learning from there and starting a new venture.

As of this week, I'm officially taking on consulting projects. I have several projects lined up and a few things on the horizon.

I'll save you the sales pitch just here though - essentially a data strategy, data science, and data engineering consultancy focusing on building strong teams and getting the most out of an organisation's data.

Watch this space.

Learning

It should come as no surprise that I spent this week side-tracked a little learning about newsletters!

In the data and analytics space though I've been reading Agile Data Warehouse Design by Lawrence Corr. It was recommended to me by a friend for proposing some brilliant approaches to getting data models together with the stakeholders in the room.

I have a customer just now that needs support with this and I've been wanting to brush up for a while. It's a great book and I'll likely do a review at some point in the future.

Check this out

There are so many incredible people in the data space, with a wealth of super valuable content landing every day.

This week I'm just going to point you at Mehdi Ouazza's amazing site:

This place captures the profiles of a load of - you guessed it - data creators and points you to where you're most likely to find them. It's a great tool that's picked up a lot of traction in the short time it's been going. I also know there are plenty of other features in the pipeline - well worth checking out to see if there are creators in the data space that are aligned with your niche.

Final thoughts

Thanks for jumping in here.

If you have thoughts or suggestions for what you'd like me to write on but I don't normally cover elsewhere, please send me a message on LinkedIn, Twitter, or Medium and I'll see what I can do.

Best of luck for the week ahead!

Adam

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

• Sponsor this newsletter: email [email protected] to discuss reaching 2000+ data and AI leaders that subscribe here.

• Consultancy: for help with your data and analytics initiatives, get in touch via email to [email protected]

• Free, daily insights: on LinkedIn here or Twitter here