How does tracking work?
Basically, tracking works by setting a cookie in the user's browser and following that cookie as the user moves from page to page. This is like getting a ticket stub at some event where you can show the stub on demand to claim a doorprize or complimentary drink, or to otherwise prove you're the same person who gained admittance at the door.
Extending the metaphor, now imagine that you are asked for the stub every time you enter a new room at the venue, and the doorman in each room logs which ticket numbers entered which rooms when; in that way, the hosts could anonymously track the movement of each specific ticket number through their venue and figure out generally where people generally went, and when, and in what sequence.
The tracking code on each page just requests the tracking cookie's ID from the user's browser, or sets a cookie if there isn't one, and tells the tracking service (such as Google Analytics) to log that ID as having visited that page. Tracking just "follows the bouncing cookie" as the buyer goes from page to page, where the tracking code on each page just asks the user's browser for a cookie that's already been set, or sets a new cookie if none was found.
How can tracking get thwarted?
Users who choose to block "third-party" cookies (cookies which are either set or read by a domain other than the page you are viewing) also present a problem for tracking conversions, because although GA sets their tracking cookie as a "first party" cookie (set for the same domain as your page), our thank-you/download page is on e-junkie.com, so then that same cookie becomes a third-party cookie with regards to the tracking code on our thank-you page. See this help page for instructions to work around this by using a thank-you page on your own site
Our thank-you/download page is the same page whether the buyer arrives there directly following checkout or via the link we issue in their thank-you email. However, if they arrive there via the link in their thank-you email from a different computer than the one they used for placing the order, or on the same computer but using a different browser program, or in a new browser session where they had set the browser to clear cookies between sessions, then the conversion cannot track back to their original purchase-related activity.
What is tracking actually good for?
For these reasons, Google Analytics and other tracking tools are mostly useful as a relative metric for more/less and before/after types of comparison, rather than an absolute precision-headcount metric. Across a large enough number of visitors, the variables of cookie settings and such discussed above will be a fairly consistent factor regardless of any variables you control in your site or marketing. Even though the numbers are not precise, the degree of imprecision or "margin of error" will be fairly consistent, so you can compare when changes to your site or marketing efforts result in significantly more or less traffic or conversions than before, or when more or less traffic and conversions appear to come from one part of your site vs. another.