← All nodes
Transformations

Type Conversion

Changes the type of one or more columns: text to a number, a number to text, text to a date, a date to text, and so on.

When to use it

Reach for it straight after an input that hands everything over as text — a CSV, a Google Sheet, the SQL Query node — so the columns downstream are the type they look like. Excel exports get real number and date cells, JSON stops quoting numbers, and a formula adds instead of joining.

In a pipeline

Settings

SettingTypeRequiredDefault or hint
Conversions (JSON)textarea—[{"columns":"Amount","type":"decimal","format":"2"}]
When a value cannot be convertedselect—blank
Read numeric dates asselect—auto

Example

An order export where everything arrived as text

  • Conversions=Amount → Decimal number (2); Qty → Whole number; Ordered → Date (DD-MM-YYYY)
  • When a value cannot be converted=Leave it blank
Rows from the CSV, every cell text
OrderQtyAmountOrdered
A-10412$1,240.002026-03-31
A-104218602026-04-02
A-1043n/a(312.50)not known
Result
OrderQtyAmountOrdered
A-10412124031-03-2026
A-1042186002-04-2026
A-1043(blank)-312.5(blank)

Money, thousands separators and accounting negatives are read as written, so "$1,240.00" is 1240 and "(312.50)" is -312.50. The two values that mean nothing are reported and left blank.

Worth knowing

  • One conversion takes any number of columns to one type; add another conversion for each further type. The columns change in place — use Calculated Column with CAST() when you want a converted copy beside the original.
  • Format means something different per type: a pattern for Text (DD-MM-YYYY, #,##0.00, 000), the rounding for a whole number (round, truncate, floor, ceiling — truncate by default, as SQL does), the number of decimal places for a decimal, and the output pattern for a date, date & time or time.
  • Blanks stay blank. Everything the type cannot read is counted and named in the node's messages, then blanked, kept, zeroed or made to stop the run, whichever you choose.
  • Dates written as numbers are read day-first unless that is impossible (13/04 can only be the 13th). Set Read numeric dates as to Month first for American data. A date that does not exist, such as 31/02/2026, is refused rather than rolled into March.
  • Dates in an .xlsx arrive as Excel serial numbers such as 46107; casting the column to Date turns them back into dates. Unix timestamps and 20260331 are read too.
  • A date comes out as text in the pattern asked for, which is how the Date Format node and the rest of FlowDoo carry dates. Sorting and filtering stay correct as long as the pattern is YYYY-MM-DD.
  • The same three fields are the arguments of CAST() in a formula, so CAST({Ordered}, Date, "DD-MM-YYYY") does in Calculated Column exactly what one rule does here.

Others in Transformations