Pulled the trigger on a budget handheld scanner last month after months of going back and forth, and I finally got a proper weekend to mess with it. First real test was a broken knob off an old cabinet, something I'd have spent an evening measuring and modeling from scratch. Scanned it in a few passes, got a usable mesh on the first try, and printed a replacement that actually fit the spline on the shaft. Genuinely satisfying moment.
The cleanup thing people always complain about is real though. The raw scans came in lumpy, with holes where the scanner couldn't see and a bunch of junk floating around the model. Fixing that up took longer than the scanning did, probly three or four times longer. Turns out matte spray helps a lot on shiny surfaces, and shooting from more angles means way less patching afterward.
So my take after one weekend is that it's worth it if you're already comfortable doing mesh repair in something like Meshmixer, and frustrating if you expect push-button results. The knob took me two hours total. Modeling it by hand probably would have taken about the same, but now that I've got the workflow down the next ones should be much quicker.