Last updated May 9th, 2026.
Tune in soon for updates on drop 4! If you purchased Drop 4: A/B Experiment, leave your profiles reviews here.
Instructions for participants
- Wait until the coffee is at least 10 days rested before brewing. Ideally, the coffee is 18-20 days rested.
- Brew each coffee (your way) and take notes on flavors, roast character, and structure.
- No you do not have to brew each coffee on the same day. If you do want to do this, I propose you try cupping them (see Base Coat IG for a how to).
- When tasting, think about structure of flavors, acidity, sweetness, and overall cup complexity. The details here should be pretty nuanced between each up.
- Share with us any feedback on taste you have in the Profile Review Form! We plan to update this page with your feedback (anonymously of course).
Profiles that were available in this experiment
| Profile Name | Color | Method |
| 1: Full Soak | Blue | This is the most standard approach where we cut gas and wait for the barrel and bean mass to find an equilibrium temperature before they rise together. For us, this usually takes around 1 minute. It also slows the roast down a bit to outside the window (+15-30s specifically) of roast times we typically like for our extra light profiles. |
| 2: No Soak | Red | In this case, we're leaving our gas application at full gas when we drop. This can work well when we charge lower and are roasting darker. I think we could still get interesting results since this will surely be the fastest roast time. I expect that we lose some complexity in this coffee though. |
| 3: Artificial Soak | Yellow | This is our standard method. It's similar to how we sample roast. Essentially, we pick a bottom out temp where we start applying gas to catch the roast. At that temp, you apply half your full gas pressure. Just before turnaround, you apply full gas. You can also fix this to time though. I could do 30s in half gas, then at 1 minute apply full gas. It leads to a quicker drying phase than a full soak and a higher Peak RoR. It's been great for wetter Colombias this year. |
| 4: TW (Tim Wendelboe) Method | Green | TW is on a loring so in all honesty, this is more a product of "common Loring approaches". The approach is to keep gas (or heat application) just barely on at drop and then at turnaround, full gas. It means you can catch the bottom out of your drop in temp earlier, which is exciting in the winter when bean mass is much lower in temp to start. I was interested in this method because it actually keeps the gas (just barely) on. On the Proaster, the igniter can be really twitchy and can cause you to lag on your soak. If I like this method, it may become the new default on this roaster. |
| 5: Step Up Method | Purple | A more gradual approach to artificial soaking that actually has full gas application come in after turnaround. Not a method I've used before but one that a friend in the co-roasting uses pretty often. The result is usually a longer maillard but still good Peak RoR. Her coffee is awesome too so I wanted to give this a try with our curves. |
Experiment Design
Each purchaser received 3 different profiles of the same coffee in 60g quantities. Each profile has the same charge temperature, drop temperature, gas adjustments, and airflow settings. Rate of rise will vary since our peak heat velocity will differ between profiles. Purchasers were not blind to profiles this time around. Instead, the packaging let them know what was different between each roast. Total roast times, and development time differences (if any) were shared on each bag.
Profiles
Below are the 5 general profiles we ran for the varying soak methods. As you may notice, some side effects of the methods included differing environmental temps (top lines), as well as differing rates of when "40 point" burner adjustments were incremently made downwards later in the roast.
The most dramatic differences being Artificial Soak (yellow) and Step-up (purple) methods. The most obvious reason being that they had the biggest differences in time it took for full gas to be applied (aside from No Soak). One thing we notice about Full Soak (Blue) as well as Artificial Soak (yellow) and Step-up soak (purple) is that our inlet, and subsequent turnaround on environment temps appear significantly lower than the other coming into the roast.
Highest peak RoR was No Soak at about 50 RoR @ 1:45. Lowest peak RoR was 47.5 @ 1:43. These differences were pretty insignificant. However, roast length varied from 6:47s with Full Soak (blue) and 7:26s with Step-up soak (purple).
RoR was kept at 24 to 21 degrees per minute going into FC, which is very fast! One thing that we did not consider in this study was that first crack would happen at very different times given the difference in stored and accumulated energy during the roast, which we largely attributing to soak. The earliest bean temperature to crack was the No Soak (red) at 383°F. The two latest to crack on bean temp were Artificial Soak (yellow) and Full Soak (blue) at 387°F. It is surprising artificial soak cracked later.
One potential limitation of the sensory results was the rate at which we enter first crack is pretty dramatically high compared to our normal, and especially other normal approaches. In retrospect, we should have probably slowed down RoR to a healthier range (16-18 RoR) going into first crack. When we roasted Full Soak method, this was the speed at which we left maillard after keeping gas as consistently high as possible. We could have done a better job stepping down the gas adjustments earlier. You can even see a spike in RoR later on that specific curve. This could explain some perceived smokiness in some roasts.
Sensory Feedback
There was only an 11% response rate for the customers that bought drop 4. The favorite profile was surprisingly the "No Soak" method. The least popular was the "Artificial Soak". In user feedback, the perceived most acidic and complex was the "No Soak" method. The perceived sweetest profile was the "Step Up" profile. We did not actually get any responses for "Full Soak" or "TW/Loring" method.
The individual feedback section shared that the method that ran the fastest (No Soak), may have leaned slightly under-developed. There was also an acrid/smokey taste to the artificial soak method, which may not have been an attribute resultant of the soak but rather the pace at which heat momentum occurred later in the roast.
We agree too that what this study really lacked was agtron. Normalizing this comparison by agtron rather than by drop temp would have probably led to better comparison. At the time, were limited by not having access to a color reader.
Limitations
Many! We mentioned before that not having agtron really limited consistency in taste. Also, our first profile we spiked RoR. We have low sensory feedback response rate. We also did not get much info into how people were brewing (cupping, espresso, drip, pourover) and if they were tasting blind or not.
What I really wanted to do with this study was measure how built momentum initially naturally decays in the same time window in these 5 roasts. Time got the better of me. Maybe one day I'll get to repeat this experiment. Or, I'll go back to my cropster export and do some data analysis on rate of decays.
Happy to share that data if anyone is interested and has the time!