select v1.amount,
    sum(case when v1.ConfirmedAt is null then 1 else 0 end) as ConfirmNullCount,
    sum(case when datediff(day,v1.confirmedat,getdate())<365 then 1 else 0 end)...

Extremely vague.  But in general, rather than actually SELECTing the data, using EXISTS() to check for it is more efficient.

If you wanted to always keep the original temp table, that's fine.  I don't have a problem at all with that.

But write the q that way then. 

Float values are not exact, they are approximate.  Internally, float 0.55 might be 0.550000000001.  CEILING goes up to the next int value no matter how small the amount above the previous...

I'm genuinely confused.

Your q states"How can I combine the three queries to one".  We write queries that combine the 3 queries into a single query. 

But then...

It could be more efficient to get only typeid from table1 in the first pass, like below.  This depends on indexing and the cardinality (how many t1 rows have typeids...

Since you're only keeping ~1/6 of the data, you may want to consider a different approach.

Backup up everything (of course!).  You might even want to make another copy...

SELECT [Product Line]
FROM dbo.[Table]
GROUP BY [Product Line]
HAVING MAX(CASE WHEN [Business Unit] = '' THEN 1 ELSE 0 END) = 1 AND
    MAX(CASE...

If there is any (non-filtered) index keyed first on id, then MAX(id) will do an index seek/lookup.  Internally it could officially be a scan, but it will be a scan...