In a post last week, I shared an unexpected data issue with the “book library” MVP project. The book databases we planned to use have quality issues. Lots of other companies use the data, so it’s not terrible, but it’s not ideal. I initially saw two paths: we can use data with flaws, or we can build a pristine data set from scratch.

I thought about it, and I decided to do both. I want to keep things moving, so we’ll start with the data from the existing book databases. That will allow us to launch the MVP’s next version faster. The data won’t be 100% accurate and might limit what features can be built. But the starting database will have a limited number of records, and we’ll start with a small number of early users. It won’t be perfect (and doesn’t need to be), but we can get something out and start the feedback loop sooner, which will lead to improvements.

After that phase, we can tackle creating a pristine database (if it’s needed). The learnings from the prior phase should also help us figure out what features to build or, more importantly, not build, around the pristine data.