This is a prototype in development. Everything here — the questions, the wording, the coffee picks — is a draft put up so you can react to it. Nothing is decided, nothing is live on cafecampesino.com, and no coffee data has been changed. Expect rough edges; that's the point of showing it this early.
A tool to help customers find the coffee they'll actually like. Built by Lowthian Design · notes last updated 30 July 2026
Most coffee quizzes ask you to describe what you like in coffee language — "do you prefer bright, balanced, or bold?" The trouble is that most people genuinely don't know, and the words don't mean much until you've done some tasting. So the quiz gets a more or less random answer and returns a more or less random coffee, which is why these things often feel like a novelty.
We're trying to build something that earns its place instead. Three ideas are doing the work:
People can't reliably say "I prefer medium acidity." They can absolutely say what's in the cupboard right now, how they brew it, and whether they put milk in it. Those answers are reliable because they're facts about a person's morning, not opinions about flavour theory.
This turned out to be the strongest question in the set. "Too bitter," "too watery," "too sour" are things people say with real confidence. So the tool takes their current coffee as a starting point and treats the complaint as a direction to move — rather than trying to pin down their ideal cup from scratch.
Every recommendation is calculated from the Roast Level, Body, Acidity, Flavor and Aroma fields already on your products in BigCommerce. Nobody wrote "if they answer X, show them the Guatemala." That means when you change a coffee's details, the tool's recommendations change with it — no developer needed.
| # | Question | Why it's there |
|---|---|---|
| 1 | How do you brew at home? | Tells us which grind to recommend — a genuinely useful output, since you sell four. |
| 2 | Caffeine? | Asked early on purpose. See below. |
| 3 | How do you take it? | The most informative taste question by a distance. Milk changes which coffees work. |
| 4 | What are you drinking now? | Sets the starting point. |
| 5 | How do you feel about it? | Sets the direction to move. |
| 6 | Which of these sounds best? | Flavour direction, asked with food words rather than coffee jargon. |
An early draft asked it last. Geoffrey pointed out the problem: the quiz spends five questions implying it's searching carefully on your behalf, then asks about decaf — where there's exactly one coffee to give. The customer's effort bought them nothing.
It turned out to be broader than decaf. Counting every combination of brew method and caffeine, 17 of 32 leave a single coffee or none — decaf, half-caff and pods are each one product, and there's no such thing as a decaf pod. So the tool now asks the two narrowing questions first and stops as soon as the answer is settled:
"No quiz needed. We roast exactly one decaf. So rather than ask you five more questions that can't change the answer — this is it."
Roughly one customer in four now answers two questions instead of six. It still gives them the grind recommendation, because that part is still useful.
Café Campesino roasts Medium, Full City, Viennese and French. In the wider trade those run from medium through to properly dark — there is no light roast in the range. That's a real characteristic of the roast style, not a gap.
But it means a customer who says "I want something bright and fruity" cannot be given what they're picturing. A quiz that hands them a Full City roast with a cheerful description is, plainly, misleading them. So the prototype says this instead:
"Your answers point lighter than anything we roast. Café Campesino's lightest is a Medium roast — closer to a traditional City roast than to a modern light roast. This is the lightest, brightest coffee we make, but if you love pale Scandinavian-style roasts, we're honestly not that roaster."
This is the biggest open question on the project. It will occasionally talk somebody out of a purchase. The argument for keeping it is that the customer who's told the truth and buys anyway doesn't send the bag back, and the one who's told "we're not your roaster" remembers a company that didn't waste their money. But that's a judgement about your brand, so it's yours and Matt's to make, not ours.
Working through the catalogue turned up some things worth a look. None of this is criticism — it's the normal state of product data that's grown over years, and most of it is quick to put right. We haven't changed anything; these are for you to decide on.
They have no Body, Acidity or Flavor filled in, so there's nothing to match on and we've deliberately left them out rather than guess. The one that stands out is the Honduras Bird Friendly Honey Processed — arguably the most distinctive coffee on the list, and its own description already contains the answer ("pronounced sweetness and less acidity"). Five fields and it becomes findable. The other two are Historic Grounds and the Single Origin Sampler.
For example, one blend's description says "medium body, low acidity" while its Body field says Light and its Acidity field says Medium. Both can't be right. The tool reads the field; the customer reads the description — so they'd see the two contradict each other. This one needs somebody who actually tastes the coffee, not a developer.
Georgia Organics, Good Morning Americus and Carter Butterfly Trail currently carry exactly the same Roast, Body, Acidity, Flavor and Aroma values. If they really are the same coffee under different labels, that's fine and worth knowing. If not, they need their own notes — at the moment the tool can't tell them apart.
| Field | Why |
|---|---|
| Process (washed / natural / honey) | One of the strongest predictors of how a coffee tastes, and currently recorded nowhere. |
| Caffeine | Not set on any product. We're currently reading it out of the product name, which works but is fragile. |
| Milk-friendly (yes/no) | One word from whoever cups these is worth more than anything we can infer. About a third of customers take milk. |
| Impact tags | Women-led, Bird Friendly, indigenous-led, supports-a-nonprofit. It's all in the descriptions, but only on some products. |
You already sell a Single Origin Sampler — four bags spanning Honduras, Guatemala, Colombia and Peru, which happens to cover your whole roast range. For the customer who honestly doesn't know what they like, that's a far better answer than a guess dressed up as a recommendation.
Longer term it could be better still: let them come back and say which of the four they liked, and that turns a guess into something we actually know about them. That's the strongest single improvement available, and it uses a product you already stock.