Don't overcomplicate your data strategy
It’s easy to make things complicated.
It requires less commitment, less resolve.
You can hide your business’s lack of clarity (or your own) with confusing terms and countless stakeholders and unclear requirements.
The result? An abstract mandate from an exec to “figure out our data strategy”.
But if you ask 10 different people what constitutes a data strategy, you’ll get 10 different answers.
The quality of your strategy has nothing to do with the size of you business. But “good” data strategies generally follow the same themes:
A data strategy is not made up of tools.
A data strategy has less to do with “the data” than you think.
A data strategy is never actually finished.
An incomplete data strategy does not mean it is an ineffective one.
Good data strategies are not complicated.
Hell, good strategies are not complicated.
Effective data strategies:
Meet consumers where they live
Provide a consistent experience
Extend to all
Simple, but not easy.
Meet consumers where they live
If your account team spends their time in Salesforce, you’re not going to have a good time telling them to use Metabase or Omni or some other tool. Don’t push for something new. Put the data where they spend their time.
Do your customers want to export data for their own use? Data sharing can take many forms, from warehouse native integrations to cloud storage. The methodology is a one part of the a larger whole.
The entry points you enable are arguably as important as the data you provide. Keep this in mind; we’ll come back to this one in a minute.
Don’t lose sight of the bigger picture - the most important factor is whether or not the data you share actually provides value to those who consume it.
Provide a consistent experience
Consistency means a few thing in this context.
Behavior needs to be consistent across systems.
Data needs to be consistent across access points - for instance, those customer exports and that Omni dashboard. Is data a core component of your in-app experience? Great, you better make sure everything jives across these various points of access. The only way to do this is to ensure they are logical and well-maintained.
The “single source of truth” label everyone likes to throw around is a bit of a misnomer; there is seldom a singular place to go to get all the data you want. Instead, a well-structured and well-understood set of resources (ie tables, datasets, etc) are what you need.
A data platform is an effective way to eventually make this “single source of truth” a reality, but it can often become just a formal entrypoint into an ugly mess behind the scenes.
Beware the “lipstick on a pig” phenomenon. You still need to establish standards and adhere to them.
Remember, tooling does not make a strategy.
Data platforms are also never quite finished, even when you think they may be. The velocity of change may slow and shrink in size and scope, but there will always be change.
It’s simply the reality of providing one centralized point of entry into a much broader and heterogenous set information.
Iterations of work to build and harden your data platform for many use cases are a requisite. Don’t be afraid of this part.
Those iterations also serve another purpose - refining your data strategy.
Extend to All
It might seem odd to use platform development to inform your strategy. Shouldn’t your strategy inform your development work?
If you are an established business with stable revenue, stable tech, and well understood growth levers, sure. If you are not, prepare for a lot of iteration.
But one thing is for certain - your data strategy defines how everyone interacts with data at your business. This means your customers, internal users and the many teams across the organizations.
They may have different SLAs or require different sets of data. But your customers - both the ones who pay you and your coworkers on different teams - can be serviced by the same strategy.
If you got this far and found yourself thinking “ this is not an exhaustive list of criteria for a data strategy”, you’re right.
Itemizing every single thing that can fall under the umbrella of “data strategy” results in more noise than value.
There is an elegance in simplicity. However simplicity does not suggest oversimplification.
Some data problems are hard. They require a lot of work, take up many resources, and have many moving parts.
An effective data strategy does not need to be hard to design or implement.
In fact, it should not be.
The simpler your data strategy, the easier it will be for your team to deliver on it.