When working with a small dataset and an efficient database, you often won't notice inefficient calculations. With larger datasets, the efficiency of your calculations can start to make a fairly dramatic difference to the speed at which a view is rendered.

Here are some tips for making your calculations as efficient as possible:

• Boolean and numeric calculations are faster than string calculations. If possible, avoid string manipulation and use aliasing or formatting to provide user friendly labels. For example, don't write the following code: IF [value] == 1 THEN "Yes" ELSE "No" END. Instead, simply write [value] == 1, and then edit the aliases of the field and set True to Yes and False to No.
• Always look for ways to increase the efficiency of a calculation. If you find yourself writing a long IF ELSEIF statement with lots of conditions, see whether there are one or two conditions that you can check first to eliminate the checks of all the other conditions. For example, let's consider simplifying the following code:

IF [Type] = "Dog" AND [Age] < 1 THEN "Puppy" 
ELSEIF [Type] = "Cat" AND [Age] < 1 THEN "Kitten" 
END

• The preceding code snippet can also be written like this:

IF [Age] < 1 THEN
    IF [Type] = "Dog" THEN "Puppy"
    ELSEIF [Type] = "Cat" THEN "Kitten"
    END
END 

Notice how the check of type doesn't have to be done for any records where the age was less than 1. That could be a very high percentage of records in the dataset.

Row-level calculations have to be performed for every row of data. Try to minimize the complexity of row-level calculations. However, if that is not possible or doesn't solve a performance issue, consider the final option.

When you create a data extract, certain row-level calculations are materialized. This means that the calculation is performed one time, when the extract is created, and the results are then stored in the extract. This means that the data engine does not have to execute the calculation over and over. Instead, the value is simply read from the extract. Calculations that use any user functions or parameters, or TODAY() or NOW(), will not be materialized in an extract as the value necessarily changes according to the current user, parameter selection, and system time. Tableau's optimizer may also determine whether or not to materialize certain calculations that are more efficiently performed in memory rather than having to read the stored value.