This content, written by Daniel Mintz, was initially posted in Looker Blog on Jun 27, 2019. The content is subject to limited support.
Every business has data. Most businesses use their data. But not enough use their data well.
Why? In large part, it’s because as the supply of data has exploded and the demand for access has grown, the existing systems simply couldn’t keep up with the squeeze.
Data analysts were trained to be a service department. Request comes in. Analyst prioritizes and responds. Requestor has a follow-up. Analyst helps. In a world of small data and limited data usage, it works great.
But in today’s world, this unleveraged model doesn’t scale. There are too many requests for the analysts available, and there’s too much data for analysts to remain familiar with all of it. What’s needed instead are data products. And to build successful data products, you need new skills on your team.
The situation isn’t that different from how software engineering evolved in the recent past. Companies realized that engineers should build and maintain reusable products—customer-facing apps, CRMs, CMSs, and ERPs—rather than just hand-coding responses to one-off requests. And to do that effectively, they needed partners who could shepherd the product through its entire lifecycle: research, scope definition, prototyping, early releases, refinement, user testing, launch, bug fixes, and upgrades.
Enter the product manager.
Businesses face a similar challenge today when it comes to data. Taking full advantage of the benefits of today’s data-rich landscape requires a shift from a service model to a product model. But too many companies are still leaving data-product-building to data engineers and data analysts.
Those roles are critical, of course, but it’s time to embrace data product managers. Building effective products is simply impossible unless someone takes the time to understand which teams will use the data and how, to research how to meet each team’s unique data needs, to test out different solutions, to observe and onboard people with the data tool, and to gather feedback and iterate on the data product.
Too many data professionals continue with a service mindset, building the data tool they think teams need or exactly what business users request. Then they wipe their hands and move on. They may not have time or motivation to find a champion for the tool, teach people how to use it, answer questions as they come up in the day-to-day, or field additional requests for improvements.
And too often, data products built this way fail. Products are built with good intentions, but they’re quickly abandoned because they’re too hard for business users to use, or don’t actually meet the intended need. Or worse, they may present authoritative-seeming answers that are just plain wrong because without an analyst constantly sanity-checking the data, bad data may slip through.
I can’t count the number of times I’ve seen people take bad data and run with it because “the tool told me this was the answer.” Whether it’s filtering some critical data incorrectly, or misunderstanding what a metric actually means, or taking a user’s request too literally, the impact is the same. And while it’s easy for data professionals to blame users, the reality is, when you’re building products, it’s your job to think about error handling and how to prevent problems proactively.
That’s why deeply understanding the product’s intended usage upfront is so important—and why having a data product manager is such a key advance. In fact, on the very best data teams, it’s not just a product manager, but a host of other new roles who collaborate to make data products great.
A data translator works on the business side, but has a deep understanding of what the data means and how it’s structured. She might be engaged to teach the product manager and data analysts about the data. A data ambassador—the person on a team who uses data most often and most comfortably—might be tasked with coming up with an onboarding and feedback plan for a new data product. And having an executive as a data champion is a crucial lever for getting people to give new data products a try.
But if you’re only going to make one change to the way you approach data, the data product manager is the place to start. She can take responsibility for the entire lifecycle of the way that data is used (or not) within a department or company. And that focus won’t just improve your data culture, it’ll also make your data teams more efficient.
Freeing data analysts and engineers from the need to squeeze product functions into their already packed schedule lets them focus on their job: building tools well and efficiently. And it makes sure that they’re building the right things and anticipating all the possible ways people will use their tools.
That’s the only way that companies will get the full value of their data. Because in today’s environment, with so much data and so many people clamoring for access, the service model of yesterday simply isn’t feasible. The transition to a world of data products is happening, the question is whether we’ll take the lessons from other disciplines and expand our teams’ competencies now, or whether we’ll stumble along with the status quo and learn the lesson the slow way.