Alternative way to make this lighter/faster

I hope I have successfully attached my excel file. Anyway the main concept is to sum the values by LEVEL. (e.g. "Level 1" is the sum of all items labeled "Level 2", "Level 2" is the sum of all "Level 3" and so on) if at all possible, to be all in the same column. I have a perfectly working workbook but the problem is that this particular document can easily reach 5000 to 15000 rows and it gets REALLY slow. I have to turn my Calculation Options to Manual every time I need to edit but it still takes at least 3 minutes just to add another row. Thanks in advance. any form of help is appreciated.

just to clarify, the performance hit is tied to your extensive use of COUNTIFS ('helper' columns as referenced by @bebo)

COUNTIFS, contrary to common belief, is not very efficient, nowhere near as efficient as SUMIFS counterpart so, when used in large volume this can cause significant overheads.
(conditional formatting will not help matters either, in terms of volatility)

Originally Posted by bebo021999

How about in AA2:

Please Login or Register  to view this content.

Drag down

AF2:

Please Login or Register  to view this content.

With first SUMIF is sum below codes those higher than B2, the second one is minus duplicate code of B2

Now all unneeded helper columns in the left can be removed.

wow this is quite perfect thanks!
but this kind of "not aligned with the same row" formula are the types that get all over the place when i start adding rows in between.

what i do normally with this is I use =indirect(address(row()+1,column()))

is this okay or is it very volatile?

Okay here is the file with updated formulas and different data

nevermind the aforementioned issue as it is not really important. anyway, using a different data, i have seen a flaw at row 126 (please see attached image).
it seems to include the other "Level 8" items even though those items should belong only on the new "Level 6" items.

PS. using your formula made the file a lot lighter to use but I still spend about a minute on "Calculating (8 threads)". do you think with having 17000+ rows
I will have to expect it anyway?