We pull traffic numbers and statistics via Google Analytics' API (What is an API? Read more here). Google chooses to calculate the numbers that it displays in its interface differently than how it calculates the numbers that it makes available through its API.
For example, if you look at the "Unique Users" metric in your native Google Analytics account, the number that you see does not include duplicate users when reported for an aggregate time period. Google strip these out before they report the aggregate number in their interface. As a result, you'll notice that the graph that they provide for this metric reports a different number than you would get if you added the unique users reported for each day.
On the other hand, the number that Google make available to developers via their API does include duplicate users. Google does not strip them out. As a result the graph that we display for the "Users" metric does equal the total that you would get if you added up the unique users reported for each day. Note, however, that this is a daily unique user count: If one of those users returns on a different day, Google chooses to count them as a new "unique" user when sending your data via their API. The user is not flagged as a duplicate. This is why those numbers can be different when looking at weekly or monthly data passed through Google's API, compared to what Google choose to display at analytics.google.com.
The bottom line here is that Google crunch the numbers differently when they report them internally, compared to when they make that same data available to developers.
Although "Users" is used as an example above, this can also apply to other metrics for the same reason.