Quid Spro Quo
Train the best Baristas.

Quid Spro Quotables

Espresso Flavor Models

IMG_0099.PNG

When teaching espresso extraction, much of the conversation is troubleshooting. “My espresso tastes too much like [blank]” and the like. QSQ is designed to stabilize a shop style by creating parameter sets within which a Barista can troubleshoot flavor. And the models that we use to teach them come with road maps for dialing in so the Barista doesn’t spend their entire shift spinning their wheels trying to make it taste less meh. As an introduction to that conversation, here’s a quick rundown of the basic principles we use to teach espresso extraction.

Coffee Strength answers:

  • How much of this cup of coffee isn’t water?

  • How does the non-water make the mouthfeel feel?

Coffee Extraction answers:

  • How effective and/or efficient was the brewing?

  • Did the brewing create a beverage that is pleasing to drink?

Extraction Flavor Modelling is:

  • A rationale for the flavor results of Extraction and Strength.

  • A working model (like model trains or blue prints) that makes a few assumptions about the brewing methodology in order to offer direction when the Extraction or Strength are unpleasant.

Espresso Extraction models are most helpful when they work. Many don’t work. Here’s why.

Key Assumptions:

  1. No two cafes are making coffee the exact same way.

  2. Nobody two cafes are using the exact same water - even in the same company. There is similar, but there is no same… except for one really expensive example that doesn’t apply here.

  3. Coffee is dynamic. The coffee I’m dialing in today is in a different state of degradation than that I’ll use next week. The roast will be slightly different even if the roast profile isn’t different.. The density will be slightly different. The exact makeup of the blend (within variety or origin) will be different. If I try to approach a dynamic product with a static strategy, I’m rarely going to hit gold twice in a row.

  4. People aren’t machines. Despite my best efforts, I am not a machine. But when I approach dialing in with an articulated strategy and rationale, I consistently get pleasant results in short periods of time.

Okay.

You may be familiar with this time based model of understanding extraction. In this model, we assume that dose and yield are constant in order to use grind size to change the elapsed time. As time progresses within the extraction, more flavor is developed.

Flavor Development over Time

Or you may be familiar with this ratio based model. In this model, we assume dose and time are constants so that grind size can be changed to affect yield outcomes, which change the Yield:Dose ratio. As yield increases, the ratio increases, and flavor develops along the following spectrum.

Flavor Dev over Ratio

Or you may be familiar with an even more elaborate model like these. In the basic version, we understand that strength is an extension of the Dose and Yield relationship, while we understand that grind size affects both surface area and flow rate. We can micro adjust extraction flavor along a timeline by using grind.

Manual Extraction Model

Or the same version but within a Volumetric or Gravimetric model. In this model, we still use grind size surface area as a lever for increasing or decreasing extraction, but we see a shift from straight up and down dosing to understanding the Ratio impact of dose on yield. Here, we attempt to stabilize extraction percentage with ratio then use micro adjustments of grind size to fix flavor problems.

VOL Grav Extraction Model
Chronometric Model

Or an even more abstract version within a Chronometric model. In this model, you may notice a big shift away from using grind size surface area as primary lever for increasing or decreasing extraction. Here, we assume stable extraction ballpark percentage because time is constant and use ratio to micro adjust for flavor and strength.

This post is getting a bit long so I’ll sum it up with the following:

When training staff to understand these models, we make some critical assumptions about what information the Barista is retaining. One of those assumptions is that once you’ve learned how to ride the bike (make the coffee), you can jump on just about any bike (espresso machine) and you’re g2g. QSQ runs contrary to that assumption. We have training models for each of these Extraction Models so if your machine breaks down or you open a new cafe or you start a catering program, your staff can work on any machine you give them. To finish up this bike analogy, we teach Baristas to ride three different bikes.