Keep it simple. Write a proposal with an estimated number of days to completion and a total price tag for the whole engagement. This is not a fixed-price proposal, but it has the same look-and-feel as one. If you go over the number of days you estimated, you bill in additional day increments. If you go under by at least a whole day, you dock days.
It does not matter when on the calendar or the wall clock you get the work done.
This basic technique works just fine when juggling multiple clients. When you're managing multiple projects, don't deliberately bill people a full day for 2 hours of work (it is just fine to bill a whole day for 2 hours work if they're the only client you have during that day, and the project simply takes less than a day: you have a 1-day minimum), but don't obsess about tracking hours.
Real clients, the ones worth keeping, care first and foremost about determinism. They want to pay a predictable amount of money for something to get done in a predictable amount of time, and that's it. They do not care about you minimizing the number of hours you bill, unless you ask them to care --- and a great way to beg a client to care about minimizing billed hours is to bill hourly.
True in the micro; false in the macro. If I deliver under budget and on time, I get more work from that client. Keep doing that and they find you more work. You get to keep billing, perhaps for years.
Changing the unit of time does not change this. You're just making it easier for yourself to justify charging a full day even if it took you 2 hours. But you could just as easily apply your methods using hours or half-days or weeks.
If you're working 2 hours and charging 8, you should simply raise your rates, and for the next project, give a more accurate time estimate.
The first couple years especially, freelancers should be doing this from project to project. It negates the "broken incentive system."
If not, then why stop at hours? Why not bill by the minute or second?
The answer to why hours are superior to minutes is the same as why days are superior to hours. It's just a useful level of aggregation to mitigate tedium.
Regardless, maintaining an incentive to bill make- or fake-work is not the only reason to avoid hourly billing. I have something like a dozen more arguments in the links I posted, and elsewhere on the site.
For me there are many aspects of a project/client I try to select for before the billing increment. I would not turn down a cool project that sounds great just because they do not agree to daily billing.
Of course in those 20 days I also worked on other stuff, worked for other clients etc.
It would seem totally unfair to me to bill them 20 days instead, so how would you handle this?
Even when I only have a single client and do nothing else, it would seem unfair to count a full day if I worked longer or shorter that day. Unfair to me if I worked longer, unfair to them if I worked shorter.