Alchemy comes in 3 distinct phases, but the destination isn't as fun as the journey.
Stage one involves smashing things together and hopefully finding something useful. This has some usefulness if you get lucky and find a hard to cure wound.
Stage two involves starting to spreadsheet properties, and semi-intelligently smash together pieces to start making weak buffs and wound heals that are notably better than most ingame firstaid options for hard to heal wounds.
Stage three involves moving from cataloguing properties to ordering properties and gaining capability to make "perfect potions".
So it's a spreadsheet game that not too many players want to directly deal with. That's not a complaint imo, the game has always been about finding niches and there doesn't need to be many alchemists to meet the community's demand for recipes. I've been begrudgingly told that my cancer juice is quite helpful for healing and sees regular use. Stages one and two I have no complaints about, it's a lot of work but it feels rewarding. My main criticisms lie with stage three and the general endgame of alchemy.
With the three potion types, you're given basically three options with very little wiggle room. Herbal swill potions are cheap and can be used to make 1 single boost/healing property potion at the cost of damage. Generally speaking, these potions have seen a lot of use since the actual damage dealt can be minimized and/or healed through other potions. But that's mostly because the other two recipes are restrictive in what they do. Mushroom recipes provide an interestingly cheap method of making perfect potions, but due to the properties and general limited number of mushrooms (15ish mushrooms vs 45ish ingredients that count as herbs) there's a very restricted and at least for this world mostly useless set of properties "affordable" mushrooms can make. And due to how alchemy works, "expensive" ingredients aren't even really worth testing. The effects of cattails or leaf ore are purely randomized and exactly the same potency. You almost don't want to know what leaf ore does, because you'll never get enough out of it to justify using it. The mercurial elixir removes that restriction at a significant cost of quicksilver, but it's rare that the amount of quicksilver involved is cheap enough to warrant using over an herbal swill with chills damage on it. Mushroom concoction isn't really salvageable in terms of utility without scrambling the properties. Mercurial elixir is just a balance issue, like lye if you 1/10thed the cost it'd be fine and come into use.
"What about mineral calcification and herbal grinds?" is an important question. Because both of those count as "alchemical ingredients", and the major limitation on what potions you can make is the primary properties of herb/mushroom/rock there's almost no point creating and cataloguing those. The vast majority of "serious" wound heals are available on raw materials, and there's no potion primary that requires them as a potential avenue for perfect potions.
Overall the limitation of "the endgame of alchemy is making potions that are grab bags of stats determined by the game for you" feels restrictive. The 45! potential herb grinds (the potential number of combinations is mind boggling) isn't an infinite playground because of there's no potion that requires it as a primary, and there's enough raw materials with the stats you need. The minor benefits of potentially needing a bit less lye to make a perfect potion aren't worth it. Notably: Brimstone potions aren't worth using at current due to cost compared to just looking for new combinations or eating a slightly bigger wound.
The other issue, mentioned by others, is that the stat boosting isn't useful at any stage of the game. Early game you're cataloguing potions and eating notable damage for small boosts of +20ish. By the time you've gotten more tools and a notable number of ingredients catalogued, the boost increases to 30ish but is no longer relevant. The duration is too short for things like combat or mining, and instead of +50% of your stats, it's closer to +10%. Soft cap wise you can't justify the cost, it doesn't last long enough for mining, and it doesn't last long enough for combat either (nor do you really want HHP damage going into a fight).
Personal takeaways to make all aspects of alchemy relevant to other gameplay: (this will make alchemy pots valuable in pvp if the % boost is valuable in mining, some people might not like that)
3 paired stats should give a % boost to stats to keep them relevant into the endgame.
Duration boost needs a buff to be useful. 2 pair should last 30 minutes, 3 pair 1hr.
Duration down is never worth using intentionally. Duration down reducing the wound given by a potion might give it an illusion of being useful, but even then I'd probably not bother personally.
Moderating the usefulness of high power wound heals by making them heal wounds slowly or split them into smaller wounds is a good way to keep healing alchemy in check if it gets silly.
Personal takeaway to improve alchemy from the alchemist's point of view:
There's been some UI issues floating around, those are covered in other threads.
Herbal grinds/calcifications should have their own recipes as primaries, to give the potential for more or less any perfect potion to exist if somehow someone can find that magic combination. This has the interesting impact of making some potions almost unique due to the ridiculous number of possible combinations. It could give a notable, and almost un-replicatible, advantage to one side that finds it. Alternatively these more processed ingredients could come with a potency boost, that'd make them interesting as well. It doesn't get around the frustration of "well if I can't find something that gives me exactly what I want I can't ever make it" though.
After lye buffs, the production of lye for alchemy feels tolerable, no longer an issue imo.
Unless it gains a new trait (like buffing potency) brimstone combustion is out of line cost wise for the alchemy system.
Mercurial potion needs a notably reduced cost to see frequent use. 1/5th would probably be tolerable, but 1/10th would make it affordable.