The format
One point per line, with the text first and the importance after a pipe: Needs a 220V circuit I do not have |4. A tab works in place of the pipe, so two columns pasted out of a spreadsheet come through intact. Importance runs 1 to 5. Leave it off and the point is counted as 3 and flagged in the table. Decimals are rounded; anything outside the range is pulled to 1 or 5; anything that is not a number at all is counted as 3 and listed under "lines that did not parse".
Roughly: 1 is a point you would not have written down if you were in a hurry, 3 is a normal reason, and 5 is a point that could decide the whole thing by itself.
Why weight rather than count
Availability is the problem. The reasons that come to mind first are the ones that are easy to picture, and ease of picturing has almost nothing to do with importance. "The seller will help me load it" arrives instantly; "parts for this model were discontinued in 2009" takes real thought and might not arrive at all. Both occupy one line. The count therefore measures how fluent you were, not how the decision should go.
Attaching a 1-5 rating breaks that. The tool reports both figures side by side on purpose: when the line count says one thing and the weighted total says another, that gap is the illusion you would otherwise have acted on.
Points that a total should not be allowed to cancel
Some entries are not really weights at all. "I cannot test it under power before paying" or "I would have to borrow against the house" are conditions, not quantities. Summing them against four small advantages produces a number, and the number is meaningless, because no quantity of convenience makes an unacceptable risk acceptable.
The tool flags any 5-weight point on the Against side for exactly this reason. When you see one, decide that point on its own first. If the answer is that you cannot accept it, the rest of the list is irrelevant. If you can accept it, take it out of the list and score what remains — otherwise it will keep distorting the total.
Deciding not to decide yet
When the two sides land inside about 15 percent of each other, this list has done all it can. A single rating moved by one point would reverse it. The failure mode here is fiddling: editing the list repeatedly until it produces the answer you wanted, which teaches you nothing you did not already know.
The better move is to ask what you are missing that keeps it close. If the answer is knowable — call the seller, measure the doorway, check what a replacement fence costs — put a date on finding out and a date for deciding, so waiting stays distinguishable from stalling. If it is genuinely unknowable, more time will not produce more information, and the sensible tiebreak is to take whichever option is easier to undo.
Questions people ask
Is it unfair if one side has more lines than the other?
The arithmetic does not care, because only the weights are summed. But a very lopsided list is worth a second look for a different reason: eight pros against one con usually means the list was written by someone who had already decided. If that describes your list, force yourself to add three more entries to the short side and run it again. Sometimes nothing changes, which is itself useful to know.
Can I give several points a 5?
You can, but the ratings stop doing any work once too many share the top. Something like a fifth of the list at 5 and the rest spread across 1 to 4 gives a readable result. If everything feels like a 5, compare points in pairs — "if I could only have one of these two, which?" — and an order falls out quickly.
Should I follow the verdict?
It has no authority beyond the ratings you supplied, so no. The output is a mirror, not an oracle: it shows you where you put the weight. Its most useful state is disagreeing with you, because that means either the ratings are not honest or the list is incomplete, and both are fixable in about two minutes.