Unfortunately, SQL Server does not have JSON_AGG . You need to use a combination of STRING_AGGJSON_QUERY and STRING_ESCAPE

SELECT
  Date = CONVERT(nvarchar(10), t.date, 104),
  bk = JSON_QUERY(N'[' + STRING_AGG(N'"' + STRING_ESCAPE(t.AnalyticGoodsT, 'json') + N'"', N',') + N']')
FROM #tmp t
GROUP BY
  t.date
FOR JSON PATH;

db<>fiddle

STRING_ESCAPE is necessary to escape any invalid JSON characters, and JSON_QUERY is necessary to prevent double-escaping.

You can only simplify your statement. I don't think, that you can build a dynamic JSON array of scalar values, so you need a text-based approach:

• Build the JSON array using a combination of STRING_AGG() , CONCAT() and STRING_ESCAPE() (to escapes possible special characters).
• Wrap the generated array with JSON_QUERY() to prevent the escaping of the generated JSON array. As is explained in the documentation, ... JSON_QUERY returns a valid JSON fragment. As a result, FOR JSON doesn't escape special characters in the JSON_QUERY return value.
• Return the expected JSON with FOR JSON AUTO .

T-SQL:

SELECT 
   CONVERT(varchar(10), [date], 104) AS [Date], 
   JSON_QUERY(
      CONCAT('[', STRING_AGG('"' + STRING_ESCAPE(AnalyticGoodsT, 'json') + '"', ','), ']')
   ) AS [bk]
FROM #tmp
GROUP BY [date]
FOR JSON AUTO

You can do it using JSON_QUERY and STRING_AGG to contruct the bk array and FOR JSON PATH as follows :

SELECT *
FROM (
  SELECT date,
         JSON_QUERY('["' + STRING_AGG( AnalyticGoodsT, '","') + '"]') as bk
          
  FROM #tmp
  group by date
) AS s
FOR JSON PATH

Demo here