No platform reports per-hashtag performance. What you get is aggregate reach and, on Instagram, a rough breakdown of how much reach came from hashtags at all. Everything beyond that has to be inferred, and inference requires discipline.
Change one thing
The whole method is this: hold everything else as constant as you can and vary the tag set deliberately. Same posting time, similar content type, similar format. Then compare across enough posts that a single viral outlier does not decide the result.
A workable structure
Build two tag sets, A and B, differing in one meaningful way — competition level, or specificity, or count. Alternate them across at least twelve posts, six each, over four to six weeks.
Record for each post: date, format, set used, reach, reach from hashtags if available, saves, follows. A spreadsheet is entirely sufficient.
What to look at
Not total reach. Reach is dominated by content quality and luck, and it will drown your signal. Look at reach from hashtags as a proportion, and at follows per thousand reached. The second one is the number that matters, because it tells you whether the tags brought people who wanted you.
Six posts is not a result
Social metrics are extremely noisy. One post that got picked up will swamp five that did not. Twelve posts is a minimum for a weak signal; twenty is where you start to trust it. This is why the method takes weeks and why most people skip it.
Check the tag page directly
The cheapest test available: post, wait thirty minutes, then open each tag page and look for your post in the recent section. If it is nowhere on a small tag, that tag is not working for your account size. This tells you something analytics never will.
Write down what you learn
The output of testing is not a number, it is a sentence in your notes: "sets weighted under 40 difficulty produce roughly double the follow rate for us." That sentence is worth more than any third-party benchmark, because it is about your audience.