"I have a bean to cup machine, what should I buy?"
The questions are the same every week. Which coffee works in a bean to cup machine? What grind for a moka pot? Is this one too acidic for someone who takes milk? Does decaf work in an espresso machine? A knowledgeable member of the team answers each one carefully, which is a good use of their knowledge and a poor use of their time.
Many people do not email at all. They look at a page of Ethiopian, Colombian and Guatemalan coffees with notes of bergamot and stone fruit, cannot tell which suits their cafetiere, and leave.
The shop speaks roaster, the customer speaks machine
Specialty coffee shops are organised the way roasters think: by origin, process and tasting notes. New customers think in terms of equipment and habit: "I have a Sage machine, I drink flat whites, I do not like bitter coffee." Grind is the second gap, since choosing whole bean for a machine without a grinder, or espresso grind for a cafetiere, is an easy mistake.
- Product pages describe flavour, not suitability for a brew method.
- Grind choice assumes the customer knows what their method needs.
- Milk or black makes a big difference but is rarely asked.
- Bean to cup customers need whole bean, and many do not know it.
- The same answers are typed by hand every day.
The cost of a confused customer
Confused visitors leave without buying, and some who do buy get the wrong coffee or grind and never return. The emails take skilled staff time. And customers who would happily subscribe once they find a coffee they love never get that far because the first choice was a guess.
A chooser and an assistant built from your range
- A short guided chooser on your site asks about brew method, milk or black, and flavour preference, in plain language, with pictures of equipment so customers can recognise their own.
- It maps answers to your real products using rules you agree with us, and recommends a coffee and the correct grind, already selected when the customer adds it to the basket.
- An assistant on the site, built on a model such as Anthropic Claude or OpenAI, answers free text questions using only your product descriptions, brew guides and policies.
- The assistant recommends real products with links, and never invents coffees, prices or claims about health or caffeine that are not in your content.
- Questions it cannot answer confidently, and anything about orders or complaints, are handed to your inbox with the conversation attached.
- You see what people ask, which shows gaps in your product pages and brew guides.
| Customer says | Recommendation comes from | What goes in the basket |
|---|---|---|
| Bean to cup machine | Your rules for espresso style coffees | Whole bean |
| Cafetiere, black | Your filter and all rounder coffees | Coarse grind |
| Moka pot, with milk | Your darker, sweeter coffees | Fine grind for stovetop |
| Pour over, likes fruity | Your light single origins | Filter grind |
| Not sure | Your best seller for beginners | Choice explained |
After it goes live
New customers get a confident answer in seconds and the right grind without needing to understand grind. The team answers fewer of the same questions and more of the interesting ones. And the questions people type become a list of what your product pages should say, which improves the shop for everyone.
Is this your shop inbox?
- You answer "which coffee for my machine" emails every week.
- Customers buy the wrong grind for their equipment.
- Your shop is organised only by origin and tasting notes.
- New visitors browse but rarely buy on their first visit.
- Your brew guides exist but few customers find them.