List of modifications

Modify Field transforms a value before it is stored or sent on. Reformatting a date, translating your vocabulary into the advertiser’s, deriving one field from another, or dropping a field from the request entirely.

It is used in Structures, in integrations, and in tracking. The operations are the same everywhere.

How a modification is built

Each modification targets one field and is made of steps. A step has four parts:

• Source, where the value comes from: this field, another field, or the result of an earlier step.
• Condition, when the step should apply. Use Always when it should apply unconditionally. See Condition Operators and Field value.
• Operation, what to do with the value.
• Parameter, what the operation needs, if anything. Some take a pattern, some take a value, some take neither.

Steps run in order

Steps are executed from top to bottom, and a later step can take an earlier one’s result as its source. That is how you build a value in stages rather than hunting for a single operation that does everything.

1.  {nin}    first regex match   ^\d{2}
2.  step 1   prefix              19

The condition on a step is checked against that step’s source value, as it is at that moment, not against what the field held when the modification started.

Two ways to use several steps

Chained, where each step builds on the last. Good for a single transformation with stages.

Independent, where several steps read the same original value and only one of them matches. Good for mapping a list of values, since each step has its own condition and only the matching one fires.

If you need several genuinely separate transformations of the same input, it is usually cleaner to create a helper field for each, then combine them at the end. That keeps each chain short enough to follow.

The operations

Changing the type

• to-int, to-float, to-bool, to-string

Text

• prefix and postfix add something to the front or the back
• replace set the value for a fixed one
• regex-replace rewrites part of it by pattern
• first-regex-match pulls out the first match

Dates

• format-date changes how a date is written
• modify-date shifts it forwards or backwards
• unix turns it into a timestamp

Encoding

• to-base64 and from-base64
• urlencode and urldecode

Calculating

• math evaluates an expression, using {value} for the current value

Not sending

• do-not-send removes the field from the request

Getting it right

Test with real data, not with what you expect the data to look like. Most Modify Field problems are not wrong logic, they are a value that arrived in a shape nobody planned for.

Keep chains short. Three steps you can read beats one clever step you cannot.