You create a page with various elements that comply with the standard. You test it in a browser that you think is up to standard (or at least close). Over time, you will correct your test because errors are found to more accurately reflect the standard. At least initially - assuming the browsers aren't very close - you have to rely on a lot of manual checking. As browsers get better and better, you can rely on browser performance as you make changes to make sure you're not introducing errors.
, , -, . , , , ACID3.