Moving to interfaces requires a culture change

Responding to yesterdays post on LinkedIn, Anna Bergevin wrote:

I agree and interface makes things less tightly coupled. But in order to change this we’d need a big cultural change at the company to seeing publishing data for secondary use cases as a core function of the job. Part of why data teams go in and get the data vs the other team publishing an API is because it’s the data team’s whole job to make data available and takes minimal effort from the software/system team. If the team is expected to make and maintain an API (including change management) that is a LOT of extra work. […]

I’m not sure how to change this culture. I guess leadership needs to see it and make the shift?

I agree, it’s a cultural change, and Anna is right the leadership needs to see the benefit if they are going to support this.

One argument for this change is that the work is being spent anyway, so it’s not extra work, it’s changing where the work is being done. And doing that work earlier reduces work/cost overall and leads to better outcomes for the organisation.

For example, can you measure the time spent in data teams pulling data? Plus the time dealing with unexpected breaking changes. Plus the time wasted by data users because they can’t access the data when it’s down. Plus the impact if they don’t want to use the data because they can’t trust it. Plus the impact of not being able to deliver on strategic goals that depend on reliable data (product features, automation, growth, AI, etc).

How much is all of that costing the organisation?

How much would it cost to create the interfaces upstream, and how much would that reduce costs downstream? If the numbers are in your favour you have a good chance of creating a business case for these changes and getting it approved.

Then it becomes just how you do data, and part of your culture.