From what I can tell, the performance score on blog posts is the proportion of views (that aren't your own) where the user scrolls to the bottom of the page. This is a nice proxy for whether someone read the entire article but it leaves out the uncertainty factor.

That is, if 8 people read an article to the end and 2 don't read to the end does that really mean my score should be 80/100?

In stats we have a tool to measure this uncertainty.

If what we know is that 8 people read an article and 2 that opened it did not read the entire article, we can use a beta distribution to model the uncertainty. What if 100 more people were to read the article? How many of them would read to the end?

Our beta model would say that with 95% confidence, the actual number of reads to the end would be between 52/100 and 97/100. That's a pretty wide spread compared to 80/100.

If instead we knew that 80 people read to the end and only 20 people didn't, we could predict with 95% confidence that if 100 more people opened the article, 72/100 to 87/100 would complete the read.

I think that would be helpful to know.