BD #15 - Where Do We Want to Go? The Second Phase of a Solid Data Strategy

Wallaroo – the purpose-built platform for last mile ML - is the sponsor for this week's issue.

Wallaroo is a great end-to-end MLOps tool with some amazing features for complex deployments (I wrote more about this here).

Download the free Community Edition here, and there are a load of excellent tutorials here.

I will continue through the four-question framework I use to build actionable data strategies this week (introduced here). We've covered getting an understanding of an organisation's current data landscape in this article. Now we're going to talk about goals and ambitions - what does good look like in this context, and is it achievable?

The TL;DR for this stage is:

• ask the end-users and business stakeholders what pain points and opportunities they spot with data across their daily workflows

• filter out the ideas that don't align with the larger business strategy (and be wary of whole application/software replacements - that's not just a data thing)

• identify what key metric or value-driver these initiatives are going to impact and how.

But let's start with a pitfall.

Setting expectations

One of the biggest challenges many organisations have comes from the nature of best practices and industry news. It's easy to get caught in a bit of a trap about taking state-of-the-art solutions or processes seen at the latest keynote and trying to copy-paste the ideas directly into their own context.

The trap is that many of those ideas aren't written by organisations like yours. They're often written by industry leaders and big-tech companies with very high data and analytics maturity, with data and machine learning at the core of their value drivers. Jumping straight to these solutions, copied straight from a blog post without all the organisational learning and understanding that went with developing them can be dangerous and wasteful.

We need to take inspiration from the business users in the organisation at hand. And that's fine. It's not a lack of ideas that usually holds them back (although if it is, I'll cover that in another post soon).

A wealth of great ideas

People are smart. Most people working with data will understand that there are things that can be done. They've read the blog posts, tinkered with the tools, and spoken to former colleagues and industry peers. When you ask them what they want to do with data, don't be surprised if you're quickly swamped with ideas.

This can be great.

By listening to the stakeholders (over the techies), you'll be grounded in the real pain points and opportunities that drive the day-to-day of the business. Not just aligning with the latest technology trends or cult-like communities.

The technologists will get their time to shine in enlightening the end-users with the "Art of the Possible" or tempering their enthusiasm in reality.

But if you don't know the organisation or context as well as they do - how do you know where to start?

We need to find the balance between:

• what is going to deliver the most impact

• what activities carry an acceptable level of risk

• what is going to drive success and adoption across the organisation

• and what is actually achievable.

We'll do this in parts.

It's strategies all the way down

Now "strategy" is a devalued word and gets thrown about by consultants a little too much (sorry!). But I'm a strong believer in being explicit and giving simple definitions so here's mine:

A set of plans to achieve long-term goals and ambitions with measurable deliverables, designed within the limits of the practices, guidelines, and resources available to an organisation.

Each business should then have an overarching strategy to achieve its longer-term goals, and data strategies should directly feed into and align with that. In some cases, the data strategy will be aligned to some larger strategy of the organisation (most commonly, the data strategy feeds into the technology strategy, much like the sales and marketing strategies should feed the commercial strategy).

This gives you the golden filter through which all "great ideas" should pass.

How does this initiative drive a positive impact to the higher-level strategy?

If something isn't clearly aligned with the wider business strategy or its value can't be articulated well, then it needs more thought.

What's this metric for anyway?

You should now have a long list of ambitions, goals, and project ideas linked to real pain points and opportunities. These ideas should span the business (or at least the subsection of the business you're working with) and should be loosely aligned to the overarching strategies the business is pursuing.

But loosely aligned isn't good enough.

Essential for the last phase of building our data strategy, prioritisation, we'll want to get a clear understanding of how to quantify the impact of these ideas. How does implementing the above move the needle?

For this, get your stakeholders to link the deliverables from these projects to metrics. If your business has existing key metrics, results, and performance indicators (OKRs, KPIs, etc.) it uses then these are great first candidates.

Sometimes, a function or team will want something that isn't clearly linked to existing indicators. That's fine - part of writing a new strategy should involve challenging the existing ways of working. Have the team come up with something measurable and quantifiable for these cases. If they can link it to either money, time, or risk measures you're in a really good spot.

You might be tempted to get into estimating impact at this stage, but I've found it's better to do that later. Go through the next phase of the data strategy framework ("How do we get there?") and come back to impact at the last phase ("Where do we start?"). The added calendar time between ideation and impact assessment lets stakeholders review, update, and edit their ideas with a fresh set of eyes (spoiler alert, they'll want to change things).

Results

You now have a great long list of things the business wants to achieve to improve the day-to-day workings of the teams with their data. You'll have everything from data lineage and catalogues to better BI reporting to fully-fledged automated machine-learning solutions churning through streaming datasets.

The point at this stage is to get people excited and engaged, but also to understand what good looks like in their eyes - and align that vision with the actual business strategy.

There is a lot of information gathering and collaboration in this process stage. I use a mix of workshops (typically following the perfect workshop blueprint), surveys, and interviews to do this. I'll try and get some templates together and posted here for easy replication in the future if that's something readers are interested in.

We'll start to create a roadmap on what it would take to achieve this next. Then finally we'll look at what's possible and where to start.

Bit of a long one, but hope it was worth it. As always, any feedback on the issues or what you might want to read in the future is greatly appreciated.

All the best,

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 to this newsletter (booked out until Jan 2023).

• CDO-as-a-Service & Consultancy: I'll help you build and shape your data function for a fraction of the cost of a permanent Head of Data or Director hire; get in touch about this or consultancy via email to [email protected]

• Coaching: I offer 1-2-1 coaching; book a coaching session here

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