I’d like to share a brief overview of the quality-control protocols we use at Prodigal. I think you’ll find them interesting, and probably a little extreme. We take great pains to buy and roast the most delicious coffee we can, and admittedly, some of these efforts make us less, not more, profitable. Hopefully that will change with time if the market comes to value things like quaker sorting and buying from a roaster able to replicate flavor batch after batch. 

Note: I have not been paid, and will not be paid, by manufactureres of any of these equipment I mention in this post.

Some of our most costly QC protocols include: 

We go to great pains to purchase almost exclusively coffees processed within the past month. The limitations and costs associated with choosing only recently processed coffees, and getting them to Prodigal in a matter of days or weeks, and not months, are extreme. 

We remove quakers to improve cup quality, but is painfully expensive. So far, the market hasn’t rewarded that effort, but we sleep better at night knowing that we have a lower percentage of quakers in our bags than any other roaster. Some of the tedious “knee-jerk moral outrage” crowd online is “offended” by our intense quaker sorting. Any consumer reading that nonsense will think “I guess those roasters sell coffee with a lot of quakers,” so I welcome their complaining. We wouldn’t be a target if we weren’t doing something special. We want to raise the bar in every way we can.

For a given coffee we have a ground-coffee color-range tolerance of four points, an accepted weight-loss range of +/-0.2%, and a cup-score floor of 87.5 points. Any batch that doesn’t meet those standards goes in our heavily discounted “first batch” offering. Selling First Batch kills our margins, but it allows us to sleep well at night, and it absolutely thrills customers who can’t quite afford to pay full retail for lovely coffee. There are times when we spend a small fortune on green coffee and the arrival coffee is below 87.5 points. That can be very costly, as we have to find a way to sell that green rather than roast it and put it in a Prodigal bag.

Green Buying

Green buying is the most challenging and time-consuming part of running our roastery. We roast dozens of samples each week, often repeating a sample roast if the first attempt did not land in our desired ranges for ground color and weight loss. 

We measure the moisture content of each green sample before roasting, using the Agratronix Coffee Tester. It’s probably not the most accurate moisture meter on the market (it seems to read a bit high), but it’s consistent enough to be reliable and useful. We roast samples in the Roest L100 Plus. (Disclaimer: I have a small financial interest in Roest.) Immediately after roasting, we calculate weight loss, and measure color using the DiFluid Omni. The Omni is consistent and wastes only 3—5 grams of coffee per ground-color reading. We are currently testing a beta version of the DiFluid Omix, which has exciting potential as an all-in-one color/moisture/density/screen size/water activity meter. We have yet to test its accuracy against other devices. If the weight loss or color of a sample is not within our target range, we will immediately re-roast that coffee. We save weight loss and color data in the Roest software.  

The next morning we cup our sample roasts blindly, often with production roasts and samples from other roasters on the table. We remove quakers from each cupping bowl and for any coffee we record the percentage of quakers by weight of any coffee we are considering for purchase.

We carefully manage our water chemistry and weigh both the water and beans in the cupping bowls. We break the crust at 4:00 and begin slurping at around 13:00. We take copious notes on each coffee and score them before discussing our findings. We debate the merits of various cups. After this we reveal the coffees, take notes, and discuss further. We cup an average of 125 bowls every week.

We will often re-cup a coffee if we missed a substantial quaker and tasted its influence in a cup. We re-cup any coffees we consider potential buys, if we scored the coffee 87.5 or higher, or we think the coffee may have 87.5+ potential if we change something in our processes. We will often brew potential purchases as Filter3 to get a different impression of its flavor. 

We pass on some coffees that are 87.5. Reasons for passing can include price, funky flavors, various types of risks, and marketability. We’re not price sensitive, but we don’t accept when a supplier asks for double the price of better alternatives. While better coffees tend to cost more overall, some fantastic green is a relative bargain, and some mediocre green costs double or triple what better alternatives do. A few middlemen buy green, mark up the price fabulously, and wait to see who bites. We’ve seen a few lots offered by various middlemen at wildly different prices. Cost of production, politics, logistics hurdles, and government policies all contribute to what sometimes seems like irrational pricing. The quaint notion many people have of a farmer asking X price, and then shipping coffee directly to a roaster exists; we have and love those relationships, but they are not the norm, especially not with certain origins or larger-production lots. We try to have as many such direct relationships as we can, but the laws and logistics in some countries make such simple, direct, transparent transactions challenging or impossible. Thankfully, technology is bringing farmers and roasters closer together in many ways, and we love being able to wire money directly to farmers when possible.

Risk Management

Not a day goes by that we don’t discuss risk at the roastery. Risk can take many forms, and without good risk management, I don’t think any producer, middleman, or roastery can survive. One risk we learned about the hard way last year was the risk that sample material had been prepared to a much higher standard than the arrival coffee. This is a common practice that somehow never favors the roaster, and undermines trust, which ultimately holds down prices and inhibits productive relationships. Anything that promotes transparency, security, and trust paves the way for potentially higher prices, in any industry. Think about this: every rich country in the world is what economists call “high trust” and every poor country is “low trust.” There may be some chicken and egg in the dynamic, but no one can argue against honest, open practices that promote trust.

There are many other sources of risk to a roastery: will a coffee degrade during transit? Will coffee take far longer than expected to arrive? Will a roaster purchase too much green and watch it age, not knowing how to sell it? Will a key account go to a competitor, causing a roaster to get stuck with extra, expensive, unsellable, aging inventory and negative cash flow? Will a coffee fade prematurely or taste baggy soon? Will a coffee simply not sound appealing to customers? Will a roastery run out of cash because it has too much money tied up in green inventory and contracts? I am not in any way downplaying the risks farmer face; those are massive and almost unthinkable. But that doesn’t change the fact that many roasters and green importers have gone out of business due to being long on expensive green or receiving low-quality, unsellable green. 

We have had coffees arrive with so many quakers that we had to sort 40% of the material in order for the coffee to meet our quality standards. (The samples were somehow almost completely free of quakers.) Obviously, that was not a profitable purchase. We have had coffees we could not sell because they arrived with fade. One coffee was beloved among customers, but we began noticing fade in about 1 of 5 cups, so we reluctantly yanked it from the menu. 

In 2023, we purchased five dud coffees. None of those coffees reached our menu. Since our business model calls for an extremely high standard for green coffee, when green coffee arrives below our standard, we need to sell it to someone else. Thankfully, we’ve always found a new home for green coffee we didn’t want to keep. 

We store our green in airtight bags in a refrigerated and humidified room. We monitor green for changes in moisture content over the weeks we store the coffee. We store a very small amount of exceptional green in vacuum-sealed bags in a freezer. Cold storage temperatures retard aging and decrease a major source of risk at a roastery. 

We buy relatively modest amounts of each coffee we purchase. Not a day goes by that I don’t have Ryan Brown in my head, telling me it is better to buy too little green than too much. 

As Ryan wrote in Dear Coffee Buyer: “It takes mere days to buy a quality coffee when necessary (call your importer, ask for samples, cup them, approve the best—or the least-awful—and ship it to your roastery), but it can take up to a dozen months to roast it all. Because of this, it is much easier to correct an underbought situation than to correct an overbought situation.”

I may have taken Ryan’s admonition to an extreme, but I’d rather run out of coffee at Prodigal than ever sell something that tastes a little aged. 

So far, we have purchased very few coffees based on preship samples (PSS) or contracts for future harvests. At some point as we grow, we will have to increase our time horizon and risk tolerance. Our tiny size has allowed us to avoid that until now. New businesses are inherently risky; we were bound to make many errors in our first year, and we did. We bought bags that needed replacing. We bought an unusable optical sorter. We ran out of money a couple of times. We bought five dud coffees. We got screwed for $25,000 by a supplier who pulled a bait-and-switch with some green, and then spent five weeks avoiding our messages. Live and learn. Once we grow and feel more secure in our purchasing relationships, we’ll be able to shoulder more risk. Had we taken the amount of risk in year one that most roasters take in an average year, we would have had to change our business model or go out of business. Being a niche roaster without a retail cafe business or a large, not-so-picky wholesale account, we don’t have the usual outlets roasters use for their mistakes (cheap blends, iced coffee, etc). 

No one says this out loud, but every roastery has a fair number of subpar green purchases each year, and they either sell the coffee as-is and hope customers don’t notice the low quality, or they hide it in a blend.

We are irrelevant in the market 

For those who think it’s “wrong” to have high standards for green, remember: every quality roaster rejects more than 90% of the samples they cup. Unlike most companies, we do not buy green based on price. We prefer to pay high prices for extraordinary coffees. And there will always be a home for those 87-point coffees we reject; it’s not like we are the only possible buyer, or even an important buyer. Any supplier holding an 87-point coffee possesses something relatively easy to sell. 

We take our responsibilities to our suppliers and our customers seriously. We pay perhaps the highest average green prices of any roaster in the world (>$10/lb FOB in 2023), we deliver it to customers who expect consistently flawless results, and we always offer a money-back guarantee. 

We are small, and do not have a material impact on the market. Our best chance of having an impact is to spend lavishly on green, to put pressure on competitors to follow suit. A few times over the years (before Prodigal) I was accused on social media of influencing green prices. Such absurd accusations indicate many people don’t realize the size of the green market. Saying I can move green prices is like saying a single EV driver can influence petrol prices. If you were to make a list of the 1,000 people in the world who most influence green prices, I would not be on that list. If you think otherwise, please get out of your bubble and read some statistics about the $40,000,000,000 per year green-coffee market. (As an aside, about 10 years ago I visited an Italian roastery that almost no one outside of Italy has heard of. At that time, the roastery produced as much coffee per week as *all* US-based third-wave roasters combined. That roastery could have swallowed the combined production of Intelligentsia, Stumptown, Counter Culture, Blue Bottle, and Onyx without getting indigestion.)

From sample roasting to production roasting

While one cannot transfer settings or temperatures from a sample roaster to a production roaster, we use sample roasting to inform how to approach a new coffee in the IMF. We roast each arrival coffee 5—8 times in the Roest and cup those samples blindly to decide our production-roast targets for color and weight loss, and we note any unusual behavior a coffee displays in the sample roaster. We could not use this prediction system with most sample roasters, but the excellent control and data collection provided by the Roest make it possible. 

Once we have chosen a profile and targets for our first production roast, we roast it in the IMF, with predictions for development time, roast duration, weight loss, and color. We rarely miss our targets by more than a trivial amount in our first production roast. After measuring color and weight loss of a roast, we may adjust our time targets and/or recipe in the IMF for subsequent batches. After our first batch of a coffee, we expect to be within +-0.1% of our weight loss target and +/-2 points of our color target on every subsequent batch. If we have any concerns, we brew a Filter3 or pour a cupping bowl of a batch right out of the roaster. Such rapid sampling isn’t ideal, but it is usually enough inform us of how to improve a coffee. 

All batches that do not quite meet our standards get blended and sold as “First Batch.” If something were ever uncomfortably off-spec, we would sell it as grinder-seasoning beans. First Batch is frankly pretty good, and an incredible bargain. 

Sorting

We run all roasted coffee through the MINI-125 optical sorter from Coffee Machines Sale.

We choose sorting weight loss by cupping several different sort levels. This process is time consuming and expensive, but our goal is to maximize cup score relative to cost by removing quakers. Quakers contribute astringency, vegetal, peanutty, and often stale flavors to coffee. They are difficult to avoid, and we do not want them in the coffee we sell. 

Very few specialty roasters sort quakers, and none to my knowledge sort the percentage of quakers we do. We average 20% sort loss on most coffees, but have gone as high as 40% weight loss to ensure the result cups at 87.5 or higher in blind cuppings. 

Quakers and odds and ends get bagged and sold as grinder-seasoning beans. We waste nothing. 

Bagging

We bag coffee using a Dupre Simplex 1H3L weigh and fill. It’s fast and accurate to within one gram. We use a continuous band sealer, DS770, also from Dupre. 

Production Cupping

The morning after production roasting, we cup every batch from the previous roast day. We skip cupping perfectly replicated batches. In other words, if three batches of a coffee all roasted in 8:31 with a weight loss of 11.1%, identical ground colors, and identical curves, we only cup one of the three batches. Our tolerances are such that almost all batches of a coffee are indistinguishable.

We save a sample of every production roast and attempt to re-cup the coffee after one, two, three, and four weeks, in order to taste the coffee as it rests and matures. We find our coffees taste best between 3–5 weeks off roast, which is why we sell several “rested’ offerings on our site. The rested coffees are popular, but really should be far more popular. We often get messages from customers who say “I finally waited four weeks to open a bag, and I wish I had been doing that all along.” 

Thanks for reading.