What We Do When a Customer Doesn't Like the Coffee (No Voucher, No Upsell)

Most small roasters have a script for a bad email: a voucher, a discount code, an apology, and a gentle nudge to reorder. We don't have one — not because we're withholding it, but because a voucher closes a conversation, and most of those conversations aren't finished yet.

Here's what actually happens when a customer writes to tell me the coffee wasn't for them.

We do not run a love-it-or-free-it scheme

A love-it-or-free-it scheme promises free coffee for telling us it was bad. It reads generous from the outside. From the roastery side, it prices every mismatch as a loss: you didn't like it, so here's a bag, and if you didn't like that one, here's another.

We roast Wednesday, ship Thursday, and answer every email ourselves — that's how the work gets done here. When I read "this coffee isn't for me," I'm not trying to buy off a complaint. I'm trying to work out which of five things happened, because the response is different for each one and only one is actually our fault.

The first thing I do is nothing

Before I decide anything I read the email twice, and then I sit on it. Not to be polite. To be accurate.

1. What exactly they said. "Bitter," "sour," "too light," "smelled like cardboard," "my partner won't touch it." Each of those is a different conversation with a different answer.

2. How many cups they brewed, and whether they brewed it twice. Someone who brewed one cup on a Tuesday morning is describing Tuesday morning. The second cup tells you almost as much as the first.

3. What they brewed it with. A blade grinder and a French press is a completely different cup from an espresso machine and a hand grinder.

4. Whether they used the word "defect." A bean with a black spot in it is a defect. "Too acidic" is not.

That pause is the whole method. Most bad-taste emails are one bad brew, and if I answer inside the first thirty seconds I will confidently answer the wrong question.

Bucket 1: it's a brewing problem

This is the biggest bucket, and the one I feel least guilty about, because in most of these cases the coffee was fine and I handed it over badly.

The order I suggest matters. I send it the same way every time.

1. Grind it finer. Most "bad" coffee is under-extracted, which reads as sour, thin, harsh, and weak all at once. Fine, medium, finer. Stop when the bed holds together. Coffee that arrives pre-ground was ground for a drip machine, coarser than most espresso and much coarser than what makes a French press taste clean.

2. Fix the ratio. For a single cup, 18 g of coffee to 300 g of water is about 1:17, and that's where we start on a V60. Move from there in 2 g steps, not from the beginning.

3. Fix the water. Boiled water straight onto fresh coffee tastes flat and scorched at the same time. Let the kettle sit about a minute. Around 200°F, which is 93°C.

4. Fix the time. Three to four minutes for a pour-over. If it finished in ninety seconds, the grind was too coarse.

5. Fix the freshness. We roast Wednesday and ship the same day, and coffee wants about three days of rest before you brew it. If they brewed it the day it arrived, that alone explains a strange cup.

Grind is the variable people miss most, by a lot. We wrote a whole piece on choosing grind size for your brew method because the distance between a grind that extracts and one that doesn't is the distance between tasting the coffee and tasting the water it sat in.

Bucket 2: it's a roast-level mismatch

Some people want dark. I send light. Not because I'm ignoring them — because I don't advertise everyone's cup.

I roast in a range that runs mostly light to medium, and it's on the site. If you want a roast so dark the origin has been cooked out of it, I'm not the shop for you, and I'm not writing a product page that pretends otherwise.

The reason matters: the darker a coffee is roasted, the more of what you taste comes from the roast and the less from the place it grew. Dark and light roasts of the same coffee are closer to two different beverages than to two shades of one thing.

What I do with a roast-level email is simple. I tell them plainly which way we roast, I name the coffees in the lineup closest to what they're after, and I stop. I don't apologize for the range I've chosen.

Bucket 3: it's a taste we don't sell

This one gets mislabeled as a defect constantly, and it isn't one.

We don't roast flavored coffee. No oils, no syrups, no hazelnut. If someone expected a flavored bag and got an unflavored one, they bought something different from what they ordered, and I'm not going to be coy about it. Coffee with flavoring on it isn't coffee anymore.

We don't sell decaf. Decaffeination is a process somebody else does, on coffee somebody else selected, and we don't have a relationship with one we trust enough to put our name on. Same answer, same way: here's what we do, here's what we don't, no apology.

They're after a heavy body. We sell light-bodied coffee. If someone wants a syrupy, low-acid, thick shot that reads as almost nothing but chocolate, they're going to be disappointed, and I'd rather they learn that here than after two bags.

Bucket 4: it's a genuine defect

This is the short list, and it's the only bucket where we replace without being asked.

What actually counts:

• A bean with a black spot in it, or a hull fragment that didn't come out in roasting.
• A piece of equipment. A metal fragment, a bit of a belt.
• A whole bag that smells like paint thinner or wet cardboard when you open it.
• A bag with a broken valve, going stale in front of you.
• A bag that arrived and isn't the coffee you ordered — that's a packing error, not a taste complaint, and it's the same fix with less waiting.

What we do: I write back, I say it's ours, and I put a fresh 12 oz / 340 g bag in the mail, whole bean, no charge. I never ask for the bag back. No return label, no photo, no waiting on a form.

Asking someone to mail a 340 g bag back to prove they were right is a punishment for the person who was already wronged. I don't need the evidence either — we taste every coffee more than once before it leaves, so I know exactly what our own roast is supposed to sound like. If what's in the bag doesn't match that, it's ours.

Bucket 5: they just didn't like it

This is allowed. It is not a problem to be solved.

A person can buy coffee from a careful roaster, brew it properly, and simply not want it. Citrus is a real note. Floral is a real note. Some people drink coffee for the caffeine, and a roasted malty cup is the entire point of their morning, and an Ethiopia is going to register as fruit juice to them.

What I say in that case is short:

> Thanks for telling me. That's a fair read, and it's a coffee that doesn't land for everyone. If you'd like something closer to heavy, low acid, and dark, tell me and I'll point you at it. If not, that's fine too.

Then I stop. No voucher, no percentage code, no link to reorder. We don't have a subscription to nudge toward, no loyalty program to credit, no club to upsell you into, and we don't run sales, so there's no discount sitting there waiting to be deployed as a peace offering.

I've watched small shops turn "thanks, I didn't like it" into a retention maneuver. It teaches the customer that complaining is a transaction. I'd rather have someone leave because the coffee wasn't for them than stay because they banked a code.

Why I fix it in that order

The order matters as much as the buckets do.

Grind first, always. It explains more bad cups than any other single variable, and it costs the customer nothing but about thirty seconds.

Ratio second. 1:17 by weight, then small steps from there.

Temperature third. 200°F, not a rolling boil.

Time fourth. Three to four minutes, actually timed.

Freshness fifth. Three days off the roast date. The date is printed on the bag, which is what it's for.

Only after all five is it fair to call it a roast problem. Only after that is it fair to call it a defect.

There's a practical reason to run it in sequence. If I email a replacement for a grind problem, I've sent them a bag that will taste exactly like the first one, and now they're let down twice. If I email them a five-step brewing fix, they might brew it again tomorrow, like it, and write back to say so.

The one thing I do with every one of these

I log what they said, roughly, next to the roast. Not the name, not the address. Just the complaint: "too acidic, French press," or "wanted dark," or "bitter, espresso machine." Then I look at the pile every few weeks.

If the same phrase shows up three or four times, that stops being a service problem and becomes a description problem — the roast, or how it's written on the page, or the fact that we're selling a light-bodied coffee to somebody who wants a heavy one. Those I can fix.

Individual taste I can't fix and I've stopped trying. But a lot of what looks like a service problem is a product page that failed to say what it should have. Fix the page and you fix it for everyone who comes after.

The takeaway

There's no scheme. There's a list of five things that could have happened, and one of them — an actual defect — is ours to pay for. Everything else is either a fix I can send you in an email or an answer I'm allowed to give: this isn't the coffee for you.

I read the email twice before I reply. I ask what you were brewing with. I replace a bag without asking for the old one back. And when somebody simply doesn't like the coffee, I don't try to buy the feeling.

Sip slow.

— E