Hi Aysen Mustafa, hopefully I can add some further insight here.

The '% in Quotes' score indicates how much of the document is within quotation marks or block quotes.

A student may try to get a lower similarity score by placing quotation marks at the very beginning and end of their assignment, then colour them white so that you can't see them. Of course, this only works if Quoted Materials are excluded from the similarity matching. 

We try to surface this through the '% in Quotes' data and the 'Hidden Text' data. In the above scenario, the Authorship report would show % in Quotes = 100% and Hidden Text = Yes. This also isn't necessarily an indication of contract cheating, but it does indicate that the student has engaged a cheating behaviour.

Having said all that, you've already ruled this out through your re-colouring of all text!

There may be some other rogue formatting which is throwing the numbers. I'll email you directly about this as the team would like to investigate the document to see if there is something can be surfaced in the report for all.

*this response was edited 04/05/2020 to remove some incorrect advice that may have mislead others

For anyone following along! 
We were able to uncover that the document in question contained a large section of text that was randomly indented. 
Turnitin identifies quoted materials as text within " " or block quotations. 3 or more lines of indented text will trigger a block quotation.

In this paper, 18 lines were indented together, and this equated to 12% of the overall submission. This is why the report showed 12% in quotes, whilst only two words were in quotation marks.