What I tracked was net change in words. If the manuscript was 200-500 words longer than the day before that was a win. If it was 200 words or more shorter, that was also a win, because I tend to have a rhythm of spewing a lot, then going back and cutting half or more.
The metric broke down a little if I was engaged almost exclusively in rewording-type revisions rather than cuts or new progress. Then I would just have to estimate whether I'd done an equivalent amount of work.
Keeping train of thought was easy as long as I didn't skip a day. Best was really to stop in the middle of a sentence, so I'd be yearning to complete it. If I skipped a week, though, the trail was cold and I was in trouble.