The unfortunate part with enterprise software, is that in the end its all bought on tender basis. Every minute the programmers and architects spend participating in the actual job, they do not spend creating new features. So your product will either have less features than the competition, or your product will cost more than the competition.

Dogfooding enterprise software is mostly impossible, since the people developing it are pushed to deliver features, because that is what administrators/managers use to choose between software. They don't care about usability, because it is not something tangible you can check.

I agree. It’s just hard sometimes if you don’t experience the problem you’re trying to solve in your daily life. How can I dogfood construction software if I’ve never worked construction? Get a construction job?

Who are you making the software for?

If it's your own company or a client, it's not too hard to get them to let you tag along with someone who will be using the software.

If you're launching a startup then you should have customers lined up before you start writing code. Ask them if you can watch them work to better understand how you can help them.

Most people are happy to share when you show interest in their work and it doesn't take much. A couple of days can really break your false assumptions and give you a good idea of what the workflow looks like. You can pick up on details like the order in which information is available.

It's worth doing more than once too, presumably your software is going save time and impact how they work so you'll need to refresh your knowledge.