Green coffee, fade, and cannibals at dinner

12 minute read

Better pass boldly into that other world, in the full glory of some passion, than fade and wither dismally with age.

James Joyce

At her class in Seattle during SCA expo in 2017, I helped Lucia Solis present a cupping of two coffees she processed 14 months earlier at an estate in Hawaii. The difference between the two was staggering: one tasted like wet cardboard with a dry finish and little acidity or structure (hallmarks of past crop coffee), and the other tasted intense, round, bright and sweet (they scored 80.5 and 84 respectively on the Q scale, for those of you keeping score at home). Both coffees had been processed, dried, milled, and shipped at the same time, and for their entire life in the U.S. sat in a plastic bag at room temperature (full disclosure: they sat forgotten on my desk until we needed to put together a cupping for her class: a happy accident, like bubblegum or microwave ovens)—not vacuum sealed, not really protected in any way.

So what was the difference between the two?

One of the the coffees—the 84—was inoculated with a selected strain of yeast during fermentation, while the other coffee underwent a spontaneous “wild” fermentation, like most coffee fermentations. This is a curious effect of yeast inoculation that Lucia and I have talked about for years: it seems to cause coffee to fade more slowly.

Importers, particularly specialty importers dealing with ever-changing consumer demand, often face the issue of their positions getting long: coffee sitting in storage, unsold, growing older and more faded. Eventually, they’ll need to discount the coffee to sell it, eating margin just to clear the position rather than continuing to pay financing on dying inventory.

But what if you could slow fade?

The potential upside for buyers like me would be enormous: I could purchase larger positions of coffees from a producer without worrying about the quality of the coffee dropping off over the life of that position. This means I could spend more time and energy on fewer coffees. I could travel less. I could spend less money and time consolidating and shipping fewer containers. I wouldn’t have to worry about timing the changing of my blend components nearly as often.

Coffee as an industry doesn’t exactly have a robust training system—disciplinary, interdisciplinary or otherwise. Knowledge is handed down between generations of baristas or roasters like totems or folklore. We’re not the craftspeople or blacksmiths we romanticize ourselves to be; instead of spending a literal lifetime learning to properly make sushi rice we accept half-baked pop science and fully baked roasts as the signatures of our trade.

I came of age in coffee in that same system. Most of what I learned during those early days of the “third wave” I taught myself from posts on or learned from temperature surfers shredding the gnar on a heat-exchanger machine during morning rush. And when I became a buyer, no one ever taught me why coffee faded. Hell, it was barely acknowledged that it did (do you ever wonder why the coffee that an importer is trying to sell you spot and says is an 86.5 doesn’t taste like an 86.5? Here’s a hint: They probably haven’t re-cupped it recently.) 

Not in the CQI’s Q course, nor on any of my buying trips, nor in talking to any of my importers, nor in reading even pretty good industry material such as Cafe Imports’ water activity study was there really a sufficient explanation for the obvious attrition of cup quality over time exhibited by green coffee. People talked about how coffee needed to “rest” in parchment for some arbitrary period of time, or how coffees tasted “fresh” and implied some sort of connection to shelf life, but it was all anecdotal. I knew that coffees from Ethiopia seemed to last longer than coffees from Peru, and coffee from certain regions of Colombia seemed to lose quality slower than others. But I never really knew why.

Fade was always sort of presented in the following way: “some coffees fade faster than others and it seems to have to do with origin or density, and definitely with moisture.” Maybe we’d throw in some hand-waves, or woo-woo mysticism, or some romantic, neo-colonialist bullshit for good measure or to make it sound like we really gave a shit about “quality” however we defined it.

Studies conducted by manufacturers of barrier bags such as GrainPro (and validated by researchers like Flavio Borem) demonstrated the ability of GrainPro and Ecotact to retain coffee quality over time versus control (jute only) samples, supporting the relationship between moisture stability and shelf life.

Prior to shipping coffee, coffee is dried ideally to stability, whether processed using wet or dry methods. It’s quite simple: not only is it more economical to ship a product without moisture (water is heavy—coffee begins as around 60% moisture if cherry or 45% moisture if parchment), but drying protects the coffee from molds or toxins (such as Ochratoxin A) that require a high moisture and high water activity environment to grow. Additionally—and directly relevant to the point of this post—dehydrating coffee prior to storage or shipping, just like with other seeds or grains, renders it relatively inert and prolongs its shelf life.

According to ICO (International Coffee Organization) minimum export standards, coffee must be between 8% and 12.5% moisture content at time of export, “with the proviso that this should not affect established, good and accepted commercial practice. Thus, where moisture percentages below 12.5% are currently being achieved exporters should endeavour to maintain or decrease these.”

Broadly, though, ICO recommends 11% moisture for coffee at time of export.

Over time, as I developed my own protocol for green-coffee analysis (a topic for a later post), I began to notice patterns in the several thousand points of data I’d accumulated from hundreds of coffees. Indeed, a few trends that were suggested by Holy Coffee Doctrine seemed to emerge:

(Also, coffees with less UV fluorescence exhibited less fade—but that, too, is a post for another time) [EDIT: Here’s that post, as promised.]

Density as a prophylactic against fade makes sense, particularly if you view fade as quite simply a loss of organic material. Extraction of coffee solids during brewing requires the presence of those solids in the first place—the result of transformations during the roasting process predicated upon the initial presence of those chemical precursors in green. Density is related to nutritional development of the seed, and the more starting material you have the more tolerant it is of somewhat predictable rates of decay (since theoretically coffee that has a bulk fill density of 0.68 g/mL won’t decay more slowly than one with bulk fill density of 0.73 g/mL if stored under the same conditions, though the organoleptic difference between the two is quite divergent). 

Then I had a thought—one that you can read between the lines of the FAO’s report about mycotoxins in coffee (tl;dr: it’s not really a concern in specialty coffees, and inoculation makes it even less so), and one that occurred to me after I watched Aimee Dudley’s (fantastic) talk at Re:Co in Seattle in 2017 (seriously, take the time to watch it). What if moisture isn’t the primary cause of fade—it’s just the environment for the things that make it possible?

What if it’s the microbes?

Is it just coincidence that drying past 11% moisture seems to prolong the shelf life of coffee and that many yeast and bacteria remain active only until ~11% moisture?

And why do coffees that are “too wet” but that have been inoculated with selected yeasts during fermentation hold up better over time than their wild-fermentation counterparts?

I must reference my earlier comments about coffee’s reign of amateurism and caution that all of the following should be approached with some appropriate level of caution: I am not a microbiologist. Fuck, I didn’t even take organic chemistry—I’m just a dude with a microscope and broadband connection (I see your woo-woo and raise you a bit of Google-fu).

Let’s take a step back.

If you were to take a swab of a coffee cherry during a dry process, or swab a fermentation during wet process, for example—like how Silva et al. did (2000, 2008a), you’d find somewhere in the realm of: 2.5 x 105 colony forming units per gram of bacteria, 200 types and 15 species of yeast, and 100 colony forming units per gram of other fungi.

There are literally millions of different microorganisms present on pulped coffee.

And, if you were to continue to swab the coffee as it dried to monitor microbial succession, you’d find that as coffee gets drier, it moves from being primarily dominated by bacteria to being dominated by yeast and other fungi—and that yeast and fungi including fermenters like Saccharomyces Cerevisiae continue to be active all the way to 11% moisture. Meaning, their metabolism continues. In the case of S. Cerevisiae, for example, this means consuming glucose and converting it into alcohol, carbon dioxide and secondary metabolites (some of which feed other microorganisms).

But where does that food for the microbes come from? That’s right: the coffee itself.

If coffee is not dried to a stable moisture content of below 11%, microbes will continue to consume available nutrients from the seed and slowly, over time, produce what we recognize as “fade” and “past crop character.” But if we dry it to 10 or 10.5%—the microbes go into a state of suspended animation. It’s like giving everyone you’ve invited over for dinner an aperitif laced with melatonin and Ambien right before you pull the meal out of the oven (can you imagine the Yelp review?). They might eat a few bites but won’t get very far.

And of course, different microbes feed on different molecules—so if you have a diversity of them living on your coffee, it’s like you’ve invited carnivores, omnivores, vegans and cannibals to your dinner (someone please turn this into a short film). Everything on the table is sure to get wrecked. This means, no leftovers: nothing for us to roast and extract later on—other than the cellulose the meal was served on (paper plates! Delicious. And classy).

But what if instead of inviting everyone, you had a very exclusive guest list—you only invited one organism with a very particular diet—and didn’t leave space for others?

(You could give them the sedative cocktail and dry to 10.5%, too, for good measure. No brussels left, but mashed potatoes for weeks!!!).

In order to test this theory of mine, I plated two samples using the method described by Dr. Dudley and her collaborators and cultured coffee on CHROMagar Candida media for three days at 30°C and then examined them to determine, at least visually, what sort of activity was present.

The idea was to show contrast—the easiest, most direct way to do this is to select a sample that challenged conventional wisdom. The coffees I chose landed in the U.S. over 5 months late, and with a moisture content over 12% upon arrival. These are red flags to any coffee buyer as they would tend to be associated with faded coffees. But, they were from Ethiopia, a country renowned for producing high-quality, high-grown, dense coffee with a long shelf life.

And sure enough—one of the coffees did taste faded, with its hallmark papery taste and lack of sweetness. The other, though, did not: it had been inoculated during its fermentation with a selected strain of yeast.

The samples selected were:

WILD sample under blacklight


Bulk fill density: 0.75 g/mL
SCA Defects (primary/secondary): 0 / 0
UV fully glowing (# / %): 0 / 0%
UV speckled (# / %): 24 / 3.6%
Moisture content: 12.5%
Water activity: 0.56
SCA Cupping score: 85.5

CIMA sample under blacklight


Bulk fill density: 0.76 g/mL
SCA Defects (primary/secondary): 0 / 0
UV fully glowing (# / %): 0 / 0%
UV speckled (# / %): 25 / 3.9%
Moisture content: 12.4%
Water activity: 0.56
SCA Cupping score: 86.75

After 3 days in the incubator at 30°, the plates did, indeed, show contrast.

WILD and CIMA plates side-by-side

Holy shit, right? The difference between the two is stunning.

The CIMA inoculated plate only shows one color of colony and relatively little activity. The color is lavender (which according to Dr. Dudley, is how Saccharomyces Cerevisiae, the species we inoculated with, will present). This suggests that the only micro-organism present on the coffee (that could grow on this plate preparation, anyway) is whatever appears lavender on the plate. We only invited the vegans so we could greedily save everything with butter for later.

The wild fermentation however, showed a frenzy of activity—including similar activity to the singular activity on Sample B (but more so), and also fuzzy brown and mauve structures and whitish fuzzy growths. We invited the cannibals to dine with the vegetarians

There’s no food left. Just the paper plates on which it was served.

Let’s play with microscopes

Under a microscope, I saw the same phenomenon— just one type of structure in the CIMA plate and multiple (including a diversity of structures that appeared to be fungus, yeast and bacteria) in the WILD plate (note that I didn’t use any antibiotics in this treatment so it’s possible I contaminated the sample from the outset, though the results proved consistent across replicates):


Saccharomyces Cerevisiae under microscope on the CIMA plate
Saccharomyces Cerevisiae under microscope on the CIMA plate


Several species of yeast, bacteria and fungus present on the WILD plate
Several species of yeast, bacteria and fungus present on a smear from the plate, plus some Agar. Remember how I said I didn’t take organic chemistry? Someday I’ll be skilled with an inoculation loop.
Another shot – several species of yeast, bacteria and fungus present on the WILD plate

This supports the notion that microbes may be implicated in fade to a greater degree than previously acknowledged and that inoculation may serve as an effective means to reduce the speed or overall effects of age on green coffee quality. 

NB: I’m not discounting the role of moisture—remember, giving your dinner guests Ambien will save the meal for later, too—but I’m hoping to merely offer other tactics for producers, exporters and importers to protect the quality of their coffee over time and retain value in a volatile marketplace. 

Other thoughts, limitations and conclusions

There are limitations and challenges to this approach. Coffee production is an unpredictable (and often unprofitable) enterprise. Sometimes, trucks break down while coffee is still wet. Sometimes it rains when there’s no one around to cover it. And not everyone can afford a moisture meter or a platform scale.

And sometimes roasters use the C-market as a price index to negotiate the price they’re willing to pay for coffee, so how is a producer supposed to afford the extra labor or infrastructure?

Unless you, as a buyer, are willing to pay (significantly) more and guarantee the purchase of your requested lot or process (yes, I’m asking that you share in the risk and forfeit your rejection rights), it’s unlikely that a producer is going to somehow magically be able to afford the raised beds or solar house or steel tanks or microbiology lab or whatever-the-fuck-it-is you’re requesting—up to and including drying to 10%.

The price that everyone around the world is used to paying for coffee was established when coffee was produced using slave labor. We need to think about that—we need to engage with how we buy coffee differently (more on this another time).

And yeast is expensive. The yeast we used for this experiment was subsidized at significant expense by the importer who guaranteed purchase of the lot as well as roasters who pre-contracted the coffee. CIMA is almost $0.12 per pound of green coffee, not including shipping or customs fees, when applied at the standard 1g yeast to 1kg of pulped coffee ratio. While that may not seem like much to us in consuming countries (what’s another GrainPro charge?), it might increase a producer’s cost of production by 10% in and of itself—and it’s not even necessary for producing high-quality coffee (and might not even change or improve the quality, at least in the short-term).

There are other strains of yeast that are much more affordable (Red Star Premier Cuvée, for example, or Laffort Rosé or Spark, to name a few) and I happily use those at their lower cost of $0.03 per pound to extend the shelf life of the coffees I buy (caution: not all yeasts work well with coffee, and not all coffee yeasts are right for you)—but it’s really just insurance for producers I have been working with for years. And of course, I pay for the yeast myself, bring it in my luggage, and do what I can to share in that risk (including giving up that lovely SAS-NANS language).

And this inoculation strategy doesn’t work with every coffee. Coffee that has been dried too hot and thus no longer has a viable germ, or coffee that isn’t super dense—let’s call it 0.72 g/mL or below (based on my observations)—is going to fade no matter what you throw at it (hell, I’ve got one of those on my sheet right now). But it won’t fade as quickly, and at the end of the day, the question is: what will be left? If you’ve still got acid, and you’ve still got structure—that might be a coffee that still has value in the marketplace.


That's just, like, your opinion, man

  1. This is super interesting! This post works with many of my suspicions as a buyer of many years. Also, I am happy to ready your humor and humility 🙂 Humor and humility may be as important as the questions you raise. Great thinking and great questioning!

  2. Fantastic insight here. It was nice to see Premier Cuvée name dropped since that’s been our yeast of choice and I hadn’t heard of others using it (Lucia hadn’t either in a consultation I had about two years ago).

    We dry our coffee slower and drier than others in the region (11% in Sulawesi/Indonesia) and use GrainPro for everything including domestic sales, which isn’t common. We have been very pleased with the shelf life, as have roasters, and I thought it was all the MC, aW, and GrainPro providing that benefit, but this is a super logical explanation for why the yeast may have been a major contributor as well.

    Stay curious 😃

    1. Right on!! Happy to hear you’ve found the same thing. Premier Cuvée is inexpensive, readily available and neutral—a great choice. My general attitude toward stability is that we should take a “yes and” approach, and it sounds like you have. Thanks for sharing!

    2. Greetings from Burundi! We are planning to use Premier Cuvée this year on some experimental batches, and would you be willing to share details about your dosing and procedures?

      Thanks and regards,

      1. Hey Dan! There are a couple ways you can go about it — either way, I’d use a dosage of 1g per kg of coffee. If you’re doing a dry process, rehydrate your yeast in clean, potable water (10L for 500g of yeast) at 20-25ºC for 20 minutes, then spray the coffee with the yeast water mixture.

        If you’re doing a wet process, you can do the standard Burundi method (12-16h dry ferment followed by a soak) and just add the rehydrated yeast directly to your dry fermentation — but what I think is substantially more interesting is to do a sequential fermentation: start with a 12 hour dry fermentation, then add water and inoculate the tank with your rehydrated yeast. This will allow you to steer and control the fermentation and prolong it if you choose.

        Finally, you can do a cherry fermentation ahead of your wet process—in which case you might wish to float the cherry, drain the water, add rehydrated yeast to the cherry and allow it to sit in shade below 30º for 24h prior to pulping.

        Please note that if you desire, you can reuse the liquid from submerged fermentation to inoculate a second batch (ideally of lower quality coffee) but it’s risky to reuse beyond 48h of total fermentation time.

        Hope that gives you some ideas!

  3. Hello! I’m greatly sorry, I didn’t get the comment in the ending about visble germ. How could it refer to the process of fade? Would you be so kind to make it clear for me?

    1. Viability of the seed — this is one metric people look at to talk about quality potential and viability has connection to shelf life. However, typically the industry is looking at the embryo when we should be talking about the endosperm

  4. I think I’m falling in love with you. I got here because of Scott Rao and I’ll be coming back until I’m up to date. It’s hard to believe you don’t have any kind of biology or chemistry background… I’m a Chemical Engineer doing PhD in materials science and your approach is more professional and scientific than that of many of my colleagues. I’m also a roaster, trying to run a small coffee roastery. If you have any project/idea in mind and need an experience hand to do the work, I’m up for it. Thank you for writing these super interesting and pleasant texts

  5. Hey Christopher, thanks for the read.. It makes me think of the really great miso/shoyu makers in Japan, as well as some wild ferment wineries here in Canada I’ve been to.. They ferment in the same rooms year after year, so that the rooms/building itself will cultivate their own little microbiome that will naturally inoculate the moromi or grapes with beneficials. There’s also often usually a lot of wood present for the microbes to live in between batches..

    My question for you.. Where does coffee fermentation physically take place (usually)? Having never seen it sadly, most of the photos I see seem to take place in uncovered, stone/tile/concrete tanks? Not really an ideal place for beneficial microbes to live between batches.

    I wonder if you can think of any producers that always ferment in wood? Then could be interesting to investigate their yeast levels, and fading trends..

    Obviously fermenting coffee in wood may not be possible everywhere in the world, but while inoculating is expensive, if you can somehow get your environment to naturally inoculate for you, you can get these benefits without the added cost.

    Also, I’m about halfway through Aimée’s talk so this may change my question/answer it for me..


    1. Hey Dylan,

      I do know of a few coffee producers who ferment in wood, and in general, the results aren’t very good.

      It’s true that it’s a potential source of inoculation for subsequent batches of coffee—however, the sheer numbers and diversity of microorganisms present in coffee fermentations create a bit of a hazard, since you don’t necessarily know *which* microorganisms are adapted to that particular environment, or which you’ve selected for. It also will accelerate microbial succession which gets you closer and closer to the microorganisms that tend to produce defects in a tank.

      I’m reminded of some mills that use recycled water (as a way to reduce water usage). At the beginning of the harvest, the fermentations will run quite long, anywhere from 24-30 hours—but by the end of the season, fermentations will complete in as few as 4 hours and the water will be dark, murky and smell strongly of sewage. The high microbial charge there means that fermentations occur both (a) faster (not good for quality, in general), and (b) with less control.

      Faster generally does not mean better when it comes to coffee. If you view the removal of mucilage as a mechanical process alone, it has an advantage—but if you view it as one capable of changing and either improving or degrading flavor, time becomes an important factor worth understanding and controlling.

      Regardless: At the scale we’re talking, inoculation is significantly cheaper than a defective lot (or one receiving a lower quality premium).

      In other mills, I’ve heard of producers have worse results after they’ve switched from bare concrete or porous tiled tanks to nonporous tiles, steel, or plastic — and they attributed it to some phenomenon like the one you’ve described. But it turns out that the kinetics of the fermentation are quite different in each of these tank materials (and the change to the tank shape as well) and as a result, the protocol of the fermentation would need to be adapted to achieve the same sort of resulting flavor profile.

      In general the best guidance is for tanks to be highly cleanable and cleaned prior to usage, whether or not inoculation is used. There is enough native microflora for the fermentation to spontaneously occur on the tanks, in the air, on the cherry.

  6. Staling through oxidation is a major concern in beer brewing. Wet cardboard aroma is probably the most commonly cited sign of a heavily oxidized beer. Other signs include muted, dull flavors and loss of hop aromatics.

    Brewers have learned that certain metals catalyze the oxygenation of beer. Of particular concern are iron, copper, and manganese, all of which are present in malt and brewing water. However, it has been learned that yeast are effective at eliminating iron and to a lesser extent copper from the beer. (Manganese, unfortunately, tends to ride through the entire fermentation and can contribute to staling the beer.)

    It occurs to me that yeast could be playing a similar role in coffee production, absorbing iron and (to a lesser extent) copper and thus eliminating catalysts that stale the coffee. So in other words it might not be the presence of other microbes, but rather the absence of yeast, that is staling one of the coffees.

    1. Interesting — Hadn’t thought of this. I remember Lucia Solis telling me that in wine fermentation, when a yeast get stressed and start producing bad-smelling aromas they ADD copper to the fermentation to get rid of the smell, so it’s interesting that in beer copper is assimilated by the yeast. I don’t think in a dry solid media like coffee oxygenation would happen quickly — but perhaps in the interstitial liquid and effluent?

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Back to top