2024 retrospective: 1631 batches, 184kg roasted

In 2024, I ran 184.5kg (406lb) of green coffee through a tiny little Roest sample roaster across 1,631 roast sessions, totaling around 230hrs.

2023 was the year I tried to log as many coffees as possible. To maximize my spread, I picked up home roasting and became the sample roaster for Hydrangea, through which Bill would funnel me green samples to up my count. A year and some major purchases later, and I’d also become the sample roaster for Moonwake, hopefully learning a thing or two about sample vs. profile roasting along the way. This won’t be a post about roast technique or methodology, but a retrospective analysis on this year’s coffee roasts.

First up - for those unfamiliar, the Roest is a sample roaster rated for up to 200g (7oz), realistically topping out at around 180g. I received the unit at the end of 2023, fully ramping up by the start of 2024. With most of my roast profiles running about 6min and several minutes of between-batch protocol, I could pump out 6-7 roasts per hour, regardless of batch size. Different batch sizes benefit different approaches due to the total thermal mass of the bean pile and their relative convective/conductive heat transfer. The Roest community has been a strong sounding board and place for new ideas, concepts, and methodologies; thanks Chris, thanks Denis.

As a sample roaster, the reality is that most green samples from producers come in 100g per offering, but the bean temperature probe on the Roest is not optimally positioned to be fully reliable until around 120g. You could learn to disregard the data and trust your roast approach. However, because purchasing decisions are made off of sample roasts, there is merit to optimizing an even smaller 50g profile with a backup, course-correcting re-do if the first attempt at a blind roast fails.

On the other hand, as a home roaster firing up beans for fun, 150-180g is the sweet spot for measly “production”. Using bean temp vs. inlet temp (BT/IT) profiles, a predictable bean can be run back to back on near auto-pilot while producing beautifully congruent roast graphs. Typically purchasing greens in 1b increments, the ~453g splits up beautifully into three 150g runs. Divvying up 1kg bags of green were a bit trickier, requiring more thoughtful summation to use up neatly.

Here’s an alternative breakdown for how much of the overall 184kg was ran across different batch sizes. Unsurprising to me, around half was produced through my confident 150g profiles. But wow, I did not expect to roast even 100kg this year on this tiny little Roest. While I find 180g batches to feel like I’m pumping more material out, it’s not that much more and I already struggle to split up and give away 150g roasts, so I’m likely to continue mostly on 150g next year.

It’s no surprise to me that Colombia took top spot among roast session count. Both in personal 150g batches and 50g samples roasts for Hydrangea and Moonwake, we go through a lot of Colombian offerings. Ethiopia isn’t too far behind, but there’s a lot less direct trade and wide swaths of sample offerings received at a time; the numbers mostly come from my own desires to regularly have a bag of something Ethiopian on hand, 150g at a time.

It dips and levels out a bit from there, with Panama samples from celebrity farms coming in 5-10+ at a time, and new partnerships with several Ecuadorian producers. What surprised me is that I thought I did a lot more Kenyan roasts, it certainly feels like I handed out a bunch to friends. I was on a chase to try every Kenyan offering to find something special, but I ended up roasting Honduran offerings more times, which I don’t have too many memories of except a couple 5lb bags I purchased to practice roasting medium-dark.

Towards the bottom, I notice that I hardly touched any Brazilian offerings, I neither bought greens nor did we work with Brazilian producers this year. Compared to my explorations last year with some yeast-fermented offerings, Burundi was also nearly missing this year.

I’ll say it - I am a funk enjoyer. I came into specialty coffee through a transition from funky IPA beers, and it was an exciting time when a new process seemed to be making its way onto a bag label almost every week; give me that triple nitrogen flushed anaerobic white honey yellow gesha. Since last year, I’ve cooled off on chasing maximum funk, and it really shows here - just over half of my roasts were washed.

At Hydrangea, we started out mostly offering funky stuff not easily obtainable in North America, but we joke that one day Bill realized he might be better at roasting clean, washed offerings. I might be in the same camp here, I struggled for much of the year to roast naturals the way I wanted while the washed offerings came out pleasant with minimal effort; perhaps also a testament to the stability of a good washed offering.

In this segment, the bottom row marked ‘others’ are either blends or bags only shipped with mystery product codes.

Wow. As luck would have it, the count of geshas is exactly the same as the various types of Ethiopian beans roasted. As these counts were made on 12/31, I could tip one to be higher than the other, but I’ll revel in the coincidence. The categories here are a bit vague - Kenyan SL28, SL34, Ruiru 11, Batian combos fall under the same SL bucket as a Colombian SL28. Pink bourbon I deemed unique and high enough in count to be separate from the 5 bourbon ajis, 6 red bourbons, and 6 orange bourbons that got counted towards the broader bourbon row. I tend to enjoy wush wush and sudan rumes, but they remain fairly uncommon; sidras are in a similar bucket for me but were more plentiful.

‘Others’ were offerings that weren’t marked, are parts of blends, or are too low to represent here. For example, the Coffea Diversa group sent samples where a many were one-off varietals I’d never heard of, such as dilla alghe, cioccie, ghimbi, and dalle amarillo; we ended up going with the wush wush and sudan rume.

This effort would’ve been a lot easier if I’d known I’d be writing up this year-end digest. It took a silly amount of manual tallying. When you’re in the thick of back to back roasts and sampling them off roast for initial evaluation, logging every detail becomes a lower priority. While I find the Roest software workflow feature-rich and outstanding overall, a couple more little things like logging varietal and process on-the-fly without creating dedicated inventory items could help track logs and enhance data analysis. Though perhaps not too many roasters are this sentimental.

Receiving samples isn’t about feeling like you’re getting free green beans - there is the expectation to apply your expertise to determine in one (maybe two) blind attempts to see if the producer’s effort and quality matches your vision of a business offering. There’s a fair bit at stake - poor sample roasting means a roaster business is missing opportunities for good finds. Sample roasting that strongly deviates from the production roast style means there’s more assumptions and mental translations that can result in disappointing output. Thousands of dollars in transactions are being committed to from a tiny reference amount, and fingers crossed that the full bag matches the 100g sampled. Samples are not cheap for producers and importer/exporters to coordinate either; it’d be disrespectful to think of it like a free spree of early access beans. Thanks to the producers who trust Hydrangea and Moonwake to represent your work.

Samples constitute potential relationships but are also an opportunity to provide feedback to producers. There are emails exchanged, spreadsheets and PDF inventories to comb through, and that’s before the logistics of purchase orders and shipment. But those don’t give indication of why you rejected or passed on some samples - information an importer or producer may be hungry for. A relationship founded on complacency or fear of giving constructive feedback doesn’t bode well for either side in the long term. Keep notes, and if solicited, that info can be a data point to help shape their future work. As someone who does not work in coffee, I have the luxury of focusing just on trying to make the beans taste good. Thanks kimoi for being a liaison and front-man to many of our partners.

As a home roaster, optimizing a roast profile has been a trial of learning through repetition. I’m happy to report that I’ve built the confidence to be happy with my own roasts, but it’ll take some time for me to formalize those thoughts to accelerate an aspiring home roaster. For better or worse, I’m on track to being the hermit-type that only drinks their own roasts; I’ve purchased a total of 21 bags roasted by someone else this year.

A lingering question - am I drinking all the stuff I roast? Of course not. Locally, our community has expanded quite this year to become a network of sharing beans, and a majority of beans I end up passing off to friends and community members (please do not reach out asking). I see no signs of our local community slowing down in sharing beans and knowledge, but in the meantime, I’ll have to find something else to start logging for an end of year post next year.

2024 retrospective: 1631 batches, 184kg roasted

The inaugural BAC brewing competition

P200: 2 years of 2 Option-O P100’s