Put a dax formula in M

Hi AxelKAp 

There are different ways to do it, but it depends on your data. So it is better you can provide some sample data in a format which people can copy.

For example, the same PO number can contain P2P and S2P, but can it contain some other different values without S2P?

One way to group by PO number, check whether it contains S2P and merge back with the original table

let
    Source = Table.FromRows(Json.Document(Binary.Decompress(Binary.FromText("i45WMlTSUQowClCK1cFkB0PZRkjiuNjGeNglqcUlSrGxAA==", BinaryEncoding.Base64), Compression.Deflate)), let _t = ((type nullable text) meta [Serialized.Text = true]) in type table [#"PO Number" = _t, #"PO Type" = _t]),
    #"Changed Type" = Table.TransformColumnTypes(Source,{{"PO Number", Int64.Type}, {"PO Type", type text}}),
    #"Grouped Rows" = Table.Group(#"Changed Type", {"PO Number"}, {{"new", each if List.Contains(_[PO Type],"S2P") then "S2P" else null }}),
    #"Filtered Rows" = Table.SelectRows(#"Grouped Rows", each ([new] = "S2P")),
    #"Merged Queries" = Table.NestedJoin(#"Changed Type", {"PO Number"}, #"Filtered Rows", {"PO Number"}, "S2P", JoinKind.LeftOuter),
    #"Added Custom" = Table.AddColumn(#"Merged Queries", "new", each try [S2P][new]{0} otherwise [PO Type]),
    #"Removed Columns" = Table.RemoveColumns(#"Added Custom",{"S2P"})
in
    #"Removed Columns"