Filtering Column A with a text value from column B

Ok, cool.

I don't think we need a conditional column. The following query turns this:

...into this:

In Power Query, create a new blank query, open Advanced Editor from the Home tab, then paste all of this over the default code so you can see each step in action:

let
    Source = Table.FromRows(Json.Document(Binary.Decompress(Binary.FromText("i45WCklMykk1MDBU0lFSitVB4Ts6OaMLubi6IQsZoekC8SMiowgpQeYbA/nuHp7oQkqxsQA=", BinaryEncoding.Base64), Compression.Deflate)), let _t = ((type nullable text) meta [Serialized.Text = true]) in type table [Id = _t, Data.Column2 = _t]),
    repBlankNull = Table.ReplaceValue(Source,"",null,Replacer.ReplaceValue,{"Data.Column2"}),
    groupIdAllRows = Table.Group(repBlankNull, {"Id"}, {{"data", each _, type table [Id=nullable text, Data.Column2=nullable text]}}),
    filterTargetTable = Table.SelectRows(groupIdAllRows, each List.Contains([data][Data.Column2], "XYZ")),
    expandDataCol = Table.ExpandTableColumn(filterTargetTable, "data", {"Data.Column2"}, {"Data.Column2"})
in
    expandDataCol

--Summary--

repBlankNull - You can ignore this step, it just swaps blank values for nulls.

groupIdAllRows - Group the table in [Id], but use the All Rows aggregator to keep all the other rows in nested tables.

filterTargetTable - Evaluates the [Data.Column2] column in each nested table to see if it contains "XYZ", and keeps the [Id] row if any of them do.

expandDataCol - Just expand back out any of the nested columns that you need.

Pete