I would estimate that 80–85% of my spreadsheets are throwaway. I create them for a specific purpose: to understand some data; to create an import file for one-off use; to prove someone wrong. (Ha!) After its creation date, I'll never use the spreadsheet again.
The other 15–20% of the time, I'm creating a spreadsheet that will last: a model that will be used to run scenarios; a template for analysis month-on-month; a risks and issues register.
Before you set out developing a spreadsheet, it's important to understand which of the above categories your efforts are likely to fall into. Are you creating it to address a burning issue that will be gone tomorrow? Or are you creating a piece of the furniture that needs to stand the test of time?
If you're doing the former, your formulae don't need to be particularly efficient. Error messages don't need to be hidden. Data doesn't need to be streamlined. And exceptions don't necessarily need to be dealt with.
It's easy to adopt a purist mentality, and make your spreadsheet sing and dance at every turn. But if you don't have to, don't bother. Life's too short.
On the odd occasion, one of the spreadsheets you think is temporary becomes strategic and long-standing. (Every week, I use a spreadsheet that I created in 2004 that I thought would be used once.) If this ever happens, you can rewrite or re-jig it in less time than the time you'll have saved cutting corners on the other ones.