My team and I went through a great session today to kick off an effort to improve our data strategy. I thought I'd share a few thoughts on some of the things we discussed.
What's a data strategy?
The phrase "data strategy" has always sounded a little "CIO of Nike"-ish to me-- like it's the kind of thing you talk about when you're the CTO of a company that's so large that everything you do is a business buzzword with a Powerpoint slide that goes with it.
But a data strategy is a real thing that becomes more important as your company moves from the smallest start-up to even a mid-sized company. At that point, you begin to realize that you've got all this data going in and out of various places, some of it redundantly, much of it manually, insecurely, and in a way that's ultimately going to hamstring you from progressing until you get it fixed. Think of it like technical debt, but this is data debt.
Identifying the data flows
First, it's a good exercise to map out how your data flows.
- From where are you receiving data?
- How does it come in?
- Where does it go?
- Who's using it?
- For what business purpose?
How's that working for you?
Then, take a clear-eyed look at how these data flows are working.
- Are they automated and efficient or manual and time-consuming?
- Are they scalable?
- Are they secure?
- What happens to any PII or the like?
- What happens if they fail?
- Are they cost-efficient?
- Are you paying to store the same stuff multiple places?
- Are you holding on to data longer than you need to?
- What happens if a disaster strikes?
- What's your RTO and RPO (i.e., how long to get back online and how much data can you afford to lose?)
- DO YOU HAVE RELIABLE, TESTED BACKUPS?!
Design a blue-sky scenario
Once you've answered these and other questions about your existing data situation, imagine what the world would look like if you fixed the short-comings. Even better, what would it look like if you knew then what you know now about how the organization uses data?
The perfect data scenario is one in which you have fast, reliable, cheap access to whatever data, however much of it, you need to get things done, even if the sh!t hits the fan.
Ne'er-do-wells hack your mainframe? No worries, you've got hot and air-gapped backups. Intense analysis needed? Perfect for your analytics warehouse. Everyone needs a different access-controlled view of the same core data? That's why you made that amazing API. Just won a 10X client, auto-scaling will take care of that!
You get the idea. Imagine the perfect end-state, and then...
Design a roadmap to get you there. Or close. Closer than you are.
Every improvement is an improvement. Don't like perfect be the enemy of better. Figure out how you might tackle the most egregious problems with an eye towards getting you closer to your beautiful, utopian data world.
If you're afraid to break things, you've got your first project. Why do you feel that way? Usually, there's a way to solidify your understanding and migrate to infrastructure-as-code tools and automated back-ups to make your environment reproducible and resilient. Sometimes, after a closer look, though, you realize that you're just doing things you've always done, and the best route is to pitch it and start over. Make a plan for that.
Embrace the change
Change, especially when it involves overhauling an entrenched data strategy, can seem scary, but the rewards for your business can be huge. Improved efficiency, security, and cost savings are just the beginning. The ultimate goal isn't just to fix what's broken but to transform your data handling into an asset that enables you to move your company forward.
A solid data strategy means you're free to develop better applications with less wheel re-invention and general anxiety around how data is handled.
As you design your roadmap to a better data strategy, consider milestones that will mark progress. Celebrate each achievement, no matter how small, and keep learning from the setbacks. Implementing a data strategy is not a one-time effort but an ongoing process that evolves with technology and the business.