So how do you avoid fast access mode? How can you eliminate sampling?
You need break your request to Google into smaller chunks, so each request is less than the amount of data that triggers sampling.
Use Analytics Canvas
The Analytics Canvas desktop tool lets you do this quickly and easily, by simply specifying how many date ranges you want to break your query into, and then automatically combines all the results into the final answer with all the data.
It then lets you save that data to a file (as an spreadsheet for example), or if you like, writes it into a database.
Analytics canvas is available for 30 days as a free trial.
Analytics Canvas tutorial
Here is how to do it in the tool. For more detailed tutorials and videos go here.
First, define the dimensions and metrics you want:
Then, select the “time period” tab, and select “Set specific time period”, then all you have to do is select the date range, and how many partitions you want.
The result will be the exact data set for the full date range as long as you have enough partitions to avoid ever including more than 500,000 visits in a single partition. In addition, the smallest partition is 1 day, so if you have more than a half a million visitors a day this will make the sampling more accurate, but won’t avoid it completely. At a half a million visitors a day, you might consider Google Analytics Premium – although it costs around $150k a year – its got big data capabilities and Analytics Canvas Premium lets you take advantage of all that data to the fullest.