Eight months chasing a decent espresso at home
I bought a used espresso machine in March because I was annoyed at spending four euros a day on coffee I didn't even love. The math was supposed to be simple: machine pays for itself, I drink better coffee, done. Eight months later I have spent considerably more than the cafe would have charged me, I own a refractometer (more on that humiliation later), and I have finally — finally — pulled a shot I'd happily serve to someone whose opinion I respect.
This is the post I wish I'd read in March. Not a buying guide. More of a "here's the order in which things actually mattered, and where I wasted my money" sort of thing.
The machine was never the problem
I started with a secondhand single-boiler machine — a Gaggia Classic someone in the next neighborhood over was selling because they "didn't have time." It came with a pressurized basket and a plastic tamper that felt like a toy. For the first month I blamed the machine for everything. Shots gushed, shots choked, crema looked great and tasted like nothing.
Here's the thing nobody tells you loudly enough: the machine is the least interesting variable. The Classic is genuinely fine. What I actually needed was the ability to grind consistently, and I didn't have it. I was using pre-ground supermarket espresso for the first two weeks, which in hindsight is like trying to debug a race condition by staring at a screenshot.
Where the money actually went
Rough order of impact, best to worst, after eight months:
- A real grinder. This was 80% of the improvement. I held off for two months because good grinders cost more than my used machine did, which felt absurd. It is not absurd. It is the whole game.
- A bottomless portafilter. Cheap, and it turned my extraction into something I could actually see. Spritzing, donuting, channeling — all of it suddenly visible.
- A WDT tool. Basically a handle with acupuncture needles glued in. Stirring the grounds to break up clumps. Looks like a joke, works like magic.
- A decent tamper that fits the basket. The toy one was leaving the puck uneven and I never knew.
- The refractometer. Last and least. We'll get there.
The grinder is the espresso machine
I keep coming back to this because it took me embarrassingly long to accept it. With the pressurized basket the machine fakes the pressure for you, so any grind sort of works and nothing tastes good. The moment I switched to the unpressurized basket, grind size became the difference between water and tar.
I went with a hand grinder with a flat-ish burr set. People online will fight you about flat versus conical burrs like it's a text-editor war, and like the editor war, most of the people fighting have never measured the difference. For me, by hand, in my kitchen, on the beans I can actually buy here — the upgrade from "any grinder" to "a grinder with adjustable, repeatable settings" mattered enormously. The upgrade from "good grinder" to "expensive grinder" I genuinely cannot taste, and I've tried.
If you're going to overspend on exactly one thing, overspend on the grinder and buy the cheapest machine that can hold temperature. I did it backwards and paid the tuition.
Dialing in, or: I am a backend dev and this is just tuning a system
What finally made it click was realizing this is the same loop I do at work. There's a small number of inputs, a measurable output, and you change one thing at a time or you learn nothing. My recipe converged on something boring and reliable:
dose: 18 g in
yield: 36 g out (1:2)
time: 28-32 s
temp: ~93 C (off the boil, no PID, I just time the warmup)
The dose I keep fixed. The grind is my one real knob. Too fast and sour? Grind finer. Too slow and bitter? Coarser. That's it. That's the whole feedback loop, and I spent two months overcomplicating it by changing dose, grind, and tamp pressure simultaneously and then being confused about why nothing was reproducible. Classic mistake. I'd review that PR harshly.
The refractometer detour
So. In month six I bought a refractometer — the little device that measures total dissolved solids so you can calculate your "extraction yield" as a percent. I told myself this was rigor. It was actually procrastination dressed up as science.
Here is what the refractometer taught me: my good shots measured around 20% extraction, and my bad shots measured around 20% extraction. The number did not correlate with whether the cup was delicious. My tongue already knew which shots were good. I'd built a dashboard for a metric I didn't need while ignoring the alert that was already firing. It now lives in a drawer. I'm not sorry I bought it, exactly. I'm just clear-eyed about what it was: a way to feel productive without drinking more coffee.
The beans, quietly, were doing half the work
The last shift was beans. I'd been buying whatever dark roast was on the shelf because it was cheap and forgiving. Once I had the grinder dialed, I started buying smaller bags from a local roaster — fresher, lighter, with a roast date actually printed on the bag. The difference was bigger than any piece of gear after the grinder.
Lighter roasts are less forgiving, which annoyed me at first. They want a finer grind and they punish a sloppy puck. But when they land, they taste like something specific — a particular fruit, a particular place — instead of just "coffee, but strong." I let the bag rest about a week after roast date now. Day-one beans gas off and the shot gushes. I learned that by ruining several mornings.
Was it worth it?
Financially, no, obviously not, don't let anyone tell you home espresso saves money. I am thousands of cups away from breaking even and I'll have replaced something by then.
But that was never really the point, was it. The honest version is that I wanted a small, bounded system I could fully understand, where the feedback loop is fast and the reward is immediate and you can hold the result in your hands at 7am. I get that at work too, occasionally, but at work the puck takes three sprints to pull and someone always changes the dose on you.
This morning's shot was good. Even pour, no spritzing, tasted like blackcurrant for some reason I can't fully explain. I'll probably never get the four euros back. I think I'm okay with that.