Start With the Difference, Not the Label
For one dose, input mass minus collected output mass equals the net difference for that observation. Weigh 18.0 g of beans in, collect 17.8 g of grounds, and the deficit is 0.2 g. That figure describes what you could not account for at the cup. It does not describe how much coffee the machine holds, because the burr chamber, chute, and other cavities can keep a store of older grounds that one dose never fully empties.
Three ideas are easy to mix up:
Concept | Plain meaning | What a scale shows |
|---|---|---|
Total resident grounds | Everything inside the grind path at a given moment | Nothing directly; a dose balance cannot count that store |
Exchanged grounds | Older grounds leaving with the new dose while some new grounds stay behind | Mass alone cannot identify old/new composition |
Net mass balance | Input minus collected output for one observed dose | One number for one observation, with spills, tare, and handling limits folded in |
Barista Magazine asked grinder makers to define retention. DF64's Anna described it as the grounds that remain after use, and Niche Coffee said retained grind goes stale "a lot like bread." These are manufacturer statements, not independent tests.
Three Hypothetical Mass Balances
The examples below are arithmetic only, not measurements taken by us, and not a description of any particular grinder.
18.0 g in, 17.8 g out. The net deficit is 0.2 g. That is the mass unaccounted for in this observation, not the mass inside the machine. It could sit in the chute, it could be an unseen spill, or it could come from how the scale rounds and how the cup was handled.
18.0 g in, 18.0 g out. Zero net difference. This reading is often mistaken for proof of zero retention, but it only shows that the mass leaving matched the mass entering. Equal input and output does not demonstrate zero retention or no exchange; old grounds can leave as new ones stay behind, so the output can weigh the same while still containing coffee from an earlier dose. Niche Coffee calls a completely zero-retention grinder impossible, because oils and very fine particles cling even in good machinery.
18.0 g in, 18.2 g out. Here 0.2 g more came out than went in. That is not negative physical retention. It could reflect release of grounds from an earlier dose, inconsistent collection or taring, or measurement and handling differences. This is arithmetic, not an observation, so it cannot say which happened.
What Limits the Inference
Five things routinely qualify a reading:
- Display resolution: display resolution is not accuracy. Rounding, tare, repeatability, and handling all affect the uncertainty of a difference. A displayed 0.2 g is an observation under those limits, not a direct count of all grounds inside.
- Spilled grounds: coffee that lands on the counter is still part of the output, and easy to miss.
- Beans left unground: beans sitting in the hopper throat or anywhere outside the burr path are neither ground nor collected.
- Receiver handling: grounds that cling to the cup by static may be weighed twice or not at all, and an inconsistent tare shifts the baseline. These are possible measurement artifacts, not an observed explanation. Static spray is a separate symptom from retention, and our guide to coffee grinder static explains how to separate the two.
- Release of previously retained grounds: a dose can flush out coffee that was already there, inflating the output without adding new retention.
Comparable observations mean the same bean, the same setting, the same dose, and a recorded idle interval. Record the idle interval because it changes the age and context of whatever coffee is sitting inside, not because a longer wait proves more retained mass or greater exchange. One interval comparison cannot establish a taste or staling effect. This mirrors the repeat-the-recipe logic in our espresso channeling guide: one odd result is a hint, repeated results are evidence.
A Blank Repeated-Dose Ledger
Copy this ledger and fill one row per dose. The first row is illustrative arithmetic, not a measurement. The blank rows are the worksheet.
Trial | Bean / setting | Idle interval | Input (g) | Output (g) | Visible spill | Input minus output | Notes |
|---|---|---|---|---|---|---|---|
Example only | Example bean, setting 12 | 10 min | 18.0 | 17.8 | None | 0.2 | Illustrative |
Blank worksheet:
Trial | Bean / setting | Idle interval | Input (g) | Output (g) | Visible spill | Input minus output | Notes |
|---|---|---|---|---|---|---|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Reading the Ledger and Following Up Safely
Read the spread before the average. If your deficits cluster once spills are controlled and idle intervals recorded, you have evidence about how much mass this setup fails to deliver per dose. You still do not have a count of everything inside, and you cannot infer freshness from the totals. Repeated comparable observations can show a pattern in delivered mass, but they do not establish every causal mechanism or prove that the coffee is fresh. One cleaning session, or one cup that tastes different, cannot establish repeatability, prove staling, or show that cleaning caused a change.
For follow-up, use the exact manual for your model. Unplug the grinder before any cleaning the manual permits, and clean only what it says is user-accessible. The maker representatives quoted in Barista Magazine's retention article recommend routine cleaning and scheduled maintenance. Skip universal purge-gram targets, water-on-beans routines, electrical work, dismantling, burr adjustment, and bellows routines: a mass ledger does not justify them, and machine-specific steps belong in your own manual. Niche Coffee's grinder frustration guide describes stale exchange as a mixing problem: a useful warning, not something a single dose balance can measure.