Person

Nineteen bars I have fixed, and what was wrong with them

TM Tyrell Mason

Since summer 2025 I have done nineteen on-site espresso jobs around Portland: new cafes, restaurants, a brewery taproom, two offices and a bakery. I keep notes on every one, partly because I am trying to work out whether this can be a real business and partly because the same three problems keep showing up. This is the honest version. What was actually wrong, what it cost to fix, and the two jobs where I could not help. There is also a section on the roasting, which is the smallest and least profitable part of what I do and the part I would give up last.

01 The three problems, in order of how often I find them

Nineteen jobs, and seventeen of them came down to one of these three. None of them are exotic. All of them are cheaper to fix than the owner expects.

One: nobody has touched the grinder

Worn burrs, eleven of nineteen. It raises your cost per drink and flattens the taste at the same time, so you pay for it twice and cannot taste why.

Burrs wear. As they wear they cut less cleanly, so you grind finer to hit the same yield, so you use more coffee, so your cost per drink creeps up. Meanwhile the fines make the shot taste dull and slightly bitter no matter what you do with the recipe. Eleven of nineteen shops I visited had burrs past their useful life. One taproom had burrs from a 2018 install that had ground an estimated 1,400 kilos of coffee. A set of 64mm burrs is about 180 dollars and forty minutes. The brewery taproom's espresso went from something the bartenders apologised for to something they started offering. Same machine, same beans, same people.

Two: there is no recipe, just a habit

Four of nineteen had no written dose and yield anywhere. Every shift made a different drink and nobody could say which one was correct.

Ask a restaurant what their espresso recipe is and you often get a shrug and the words we just fill the basket. A recipe is three numbers and a card: dose in, weight out, time. Eighteen grams in, thirty-eight out, twenty-nine seconds. Written down, taped to the machine, checked twice a day with a scale that costs thirty dollars. Without it, the morning person and the evening person make two different drinks and neither knows they are drifting. The fix takes an hour and costs nothing but the scale. It is the single highest-return thing I do and the one owners most often think is unnecessary.

Three: water

Two of nineteen. Portland water is soft and low in alkalinity, which is not neutral for espresso, and it will make good coffee taste thin and sharp.

Portland tap is soft. That is great for scale and hard on flavour. Low alkalinity means less buffering, which pushes acidity forward and makes a good coffee taste sour and thin, and people then blame the roast. Two shops had this and both had been chasing it through their beans for a year, changing roasters twice. A proper filtration cartridge with some mineral adjustment, around 400 dollars installed, fixed both. I am careful here because water is where consultants love to sell expensive things nobody needs. I only raise it after the burrs and the recipe are sorted, because those two explain almost everything else.
02 The two I could not fix

Four of the nineteen were equipment problems I could not solve in an afternoon. Two of them I want to describe, because I think turning down work is part of the job description.

The office with the eight thousand dollar machine

Beautiful two-group machine in an office lobby, plumbed wrong, no drainage, and nobody whose job it was to clean it. No recipe would have survived.

A tech office bought a serious commercial machine as a perk. It was plumbed to a line with no proper drain, so the drip tray was emptied by hand whenever someone noticed, which was rarely. Nobody owned it. The group head had not been backflushed since install. I cleaned it, dialled it and trained four people. I also told the facilities manager plainly that in three months it would be back where it started unless one named person owned it and had twenty minutes a week scheduled for it. They did not do that. I drove past a year later and it was gone, replaced with pods. I charged for the visit and I do not think I earned it.

The new cafe that bought the wrong grinder

A great machine paired with a domestic grinder that could not keep up past twenty drinks an hour. That is not a dial-in problem, it is a purchase problem.

A new shop in Montavilla spent well on the espresso machine and then saved money on the grinder, which is backwards. The grinder they bought was fine for a home kitchen and overheated inside half an hour of a real rush, at which point the grind wandered and every shot after that was a guess. There was nothing to fix. I told them so in the first hour, gave them two grinder options at 1,600 and 2,400 dollars, and charged them for one hour instead of three. They bought the cheaper one. I went back two months later and dialled the whole bar in properly, and that job worked.
03 The roasting, which barely pays

Thirty pounds a week on a shared one-kilo drum. It is the least profitable thing I do and the part of the week I look forward to most.

Why one kilo at a time

Because it is what I can afford and what fits the bay. Small batches mean I can change a profile in a week instead of committing to four hundred pounds of green.

A one-kilo drum is a slow way to make coffee and a fast way to learn. I roast Monday and Tuesday mornings, about fourteen batches across the two days, which is thirty pounds finished. Green comes through a small importer in Seattle who publishes farm-gate prices. I buy two coffees at a time, a Colombia and an Ethiopia, in 30 kilo lots, because that is what I can pay for at once and what I can move before it goes flat. The economics are thin. After green, gas, bags and rent on the bay I clear maybe four dollars a pound, and that is before I count my time, which I do not, because if I counted it I would stop.

Public spreadsheet of every batch I have roasted: charge temp, first crack time, development, drop, and my tasting note three days later. Including the failures.

🔗Roast log, every batch since Marchmasoncoffeeworks.com↗