Hypothesis 1
There was no single winning recipe.
The top 3 were not small edits of one method. They split across low-water concentrate, high-bypass dilution, and fuller brew-water strategies.
Interactive analysis
WAC 2025 AeroPress recipes compared
The 2025 World AeroPress Championship recipes did not point to one winning formula. Dose clustered near 18 g, but bypass, brew water per dose, contact time, press behavior, water, and workflow spread widely. The top 3 looked successful in different ways, so the useful lesson is not a recipe clone. It is how competitors shaped extraction, dilution, and workflow around a target cup.
Top 3
Recipes
22
Unique included brews in the current dataset.
Comparisons
231
All brew-to-brew pairwise recipe comparisons.
Median distance
0.433
A middle-field recipe pair was meaningfully different, not nearly identical.
Bypass range
0%-59%
Dose stayed tighter: 18 g median.
Working hypotheses
Hypothesis 2
Dose was the stable part.
Most recipes stayed close to 18 g. The bigger moves came from brew water, bypass, contact time, press behavior, water, and workflow.
Hypothesis 3
Bypass was a major strategic lever.
Some brewers extracted close to their served volume. Others brewed a strong concentrate and rebuilt the cup with bypass water.
Hypothesis 4
Top recipes made the final cup separately from the brew step.
The podium recipes suggest that extraction setup and served-cup dilution should be judged as related but separate decisions.
Hypothesis 5
The weakest public recipes are under-specified.
Several recipes omit final beverage mass, press duration, water TDS, or bypass basis. That makes them harder to compare honestly.
Recipe universe
How far apart were the recipes?
This map projects the recomputed recipe-distance model into two dimensions. Nearby dots share more recipe structure; distant dots solved the cup in different ways.
Workflow divergence
More divergent
Core-pattern similarity
Less divergent
Less shared
More shared
This is still a projection of the full distance model, not a direct brewing-variable chart. Read the x-axis as how much recipes shared a common structural pattern, and the y-axis as how differently they reached that pattern in workflow terms.
Extraction vs dilution
The field split along bypass and extraction opportunity.
The x-axis is a relative extraction-opportunity proxy, not measured extraction yield. The y-axis is bypass share. Bypass is kept separate because it dilutes or shapes the served cup without extracting more from the bed.
Bypass share (%)
58.7%
0%
Extraction-opportunity proxy
0
100
High bypass
Higher proxy
Low proxy / low bypass
Recipes with missing bypass are excluded from this map only. They remain in the full distance model, table, methods, and pair comparison where the missing value is visible.
Variable explorer
Most recipe differences happened after dose.
Select one variable to see the full field. Each dot is one recipe, with podium recipes highlighted. The band shows the middle half of the available data.
Bypass
Coverage: 19/22 recipes · median 23.33 %
min
0.00
p25
13.63
median
23.33
p75
42.46
max
58.67
Available rangeMiddle 50%
Bypass (%)
Min 0.00 %Max 58.67 %
This is a distribution of available values, not a fitted bell curve. With 22 recipes and uneven missing fields, the honest view is the actual dot spread plus quartiles.
Selected recipe
Némo Pop
1st · Australia
41.2 %
Podium vs field
The top 3 did not agree on one template.
Australia, Switzerland, and India were all successful, but they solved extraction setup and served-cup dilution differently.
1st · Australia@nemo.pop
Dose18 gfield avg 17.78 g
Brew water/dose5.56 g/gfield avg 9.58 g/g
Bypass41.2%field avg 24.2%
Contact60 sfield avg 114.18 s
Extraction proxy25.5field avg 57.65
2nd · Switzerland@jan.aeropresser
Dose18 gfield avg 17.78 g
Brew water/dose5.56 g/gfield avg 9.58 g/g
Bypass56.6%field avg 24.2%
Contact112 sfield avg 114.18 s
Extraction proxy56.3field avg 57.65
I used AeroBox in my Switzerland workflow to watch the pressed output and make the bypass step easier to reason about. This page does not argue that AeroBox explains the result. It is simply part of how I kept my own press and final cup readable.
3rd · India@dharunvyas
Dose16 gfield avg 17.78 g
Brew water/dose13.00 g/gfield avg 9.58 g/g
Bypass5.5%field avg 24.2%
Contact80 sfield avg 114.18 s
Extraction proxy56.3field avg 57.65
Podium pair distances
Australia ↔ Switzerland
Némo Pop and Jan Ahrend
0.573
highly different
Australia ↔ India
Némo Pop and Dharun Vyas
0.686
highly different
Switzerland ↔ India
Jan Ahrend and Dharun Vyas
0.441
meaningfully different
Closest non-podium recipes
By average distance to all three podium recipes, Ammar Ali (Oman) was the closest non-podium profile.
Recipe distance
The recipes were spread out, not copies of one method.
The median pairwise distance landed near the middle of the scale. That means the field shared competition habits, but the recipes were not tightly clustered.
Pairs
231
Median
0.433
Closest
0.166
Farthest
0.712
very close
3
close
58
meaningful
109
highly different
61
Closest pairs
Most different pairs
Pair compare
Choose two recipes and see what actually changed.
The highlighted cells mark differences between the selected recipes. This is useful because two recipes can share a dose while diverging hard on bypass, press, water, or workflow.
| Variable | Némo Pop | Jan Ahrend |
|---|---|---|
| Dose | 18 g | 18 g |
| Position | upright | inverted |
| Grinder / clicks | Comandante Trailmaster x25 Tigershark · 31 | Comandante C40 MK4 · 24 |
| Brew temperature | 84 C | 88 C |
| Water TDS | 125 ppm | 60 ppm |
| Brew water | 100 g | 100 g |
| Brew water/dose | 5.56 g/g | 5.56 g/g |
| Bypass | 41.2% | 56.6% |
| Bypass timing | before brew | after brew |
| Final beverage | missing g | 152 g |
| Effective contact | 60 s | 112 s |
| Press duration | 20 s | 44 s |
| Filter setup | flow control plus double standard paper | double standard paper |
| Flow-control cap | yes | no |
| Sifting / fines | yes / yes | no / no |
| Extraction proxy | 25.5 | 56.3 |
| Confidence | high | high |
Recipe table
The full field in one view.
Use the table for scanning. The maps show shape; the table shows the recipe variables that created that shape.
| Competitor | Style | Dose | Brew water/dose | Bypass | Contact | Extraction proxy | Distance from center | Data notes |
|---|---|---|---|---|---|---|---|---|
1st · Australia | bypass-first concentrate upright | 18 g | 5.56 | 41.2% input water proxy | 60 s | 25.5 high | 0.534 outlier #1 | final beverage weight not stated; Trailmaster/Tigershark clicks not directly equivalent to C40 clicks |
2nd · Switzerland | ultra-concentrate + high bypass inverted | 18 g | 5.56 | 56.6% final beverage | 112 s | 56.3 high | 0.332 outlier #11 | press_duration_s; press duration estimated from 66g output at 1-2g/s |
3rd · India | low bypass inverted | 16 g | 13.00 | 5.5% input water proxy | 80 s | 56.3 high | 0.400 outlier #4 | press_start_s; exact press start time not stated; final beverage weight not stated |
Brazil | moderate bypass inverted | 18 g | 11.11 | 23.3% final beverage | 95 s | 62.9 medium | 0.000 outlier #22 | brew_water_g; concentrate_yield_g; whether second pour was +140ml or up to 140ml |
Costa Rica | fuller-volume extraction upright | 18 g | 12.89 | 0% input water proxy | 135 s | 93.9 high | 0.437 outlier #2 | position; press_start_s; press_duration_s |
Czech Republic | high bypass inverted | 18 g | 9.33 | 58.7% final beverage | 90 s | 56.4 high | 0.243 outlier #17 | No major flags |
England | unknown bypass inverted | 17 g | 10.59 | missing missing | 130 s | 80.9 high | 0.455 outlier #3 | bypass depending on taste not specified |
France | unknown bypass inverted | 18 g | 8.33 | missing missing | 105 s | 36.3 high | 0.322 outlier #14 | bypass amount not stated |
Ireland | low bypass inverted | 18 g | 8.89 | 13.3% final beverage | 105 s | 67.6 medium | 0.166 outlier #20 | press duration not stated |
Mexico | unknown bypass inverted | 18 g | 11.39 | missing missing | 175 s | 78.4 medium | 0.218 outlier #19 | bypass amount not stated because recipe says restore to 205g |
Netherlands | moderate bypass inverted | 18 g | 6.67 | 36.8% final beverage | 107.5 s | 50.3 high | 0.297 outlier #15 | concentrate yield not stated |
Northern Ireland | low bypass inverted | 17.5 g | 9.71 | 13.9% final beverage | 118.8 s | 75.8 medium | 0.287 outlier #16 | brew temperature not stated |
Norway | moderate bypass inverted | 18 g | 8.89 | 20% input water proxy | 127.5 s | 38.5 high | 0.290 outlier #13 | brew_water_g; whether second pour was to 160g total or +160g |
Oman | moderate bypass inverted | missing g | missing | 27.5% final beverage | 85 s | 25.9 medium | 0.296 outlier #8 | dose not stated |
Philippines | low bypass traditional | 18.2 g | 9.89 | 15.6% final beverage | 105.3 s | 31.1 high | 0.348 outlier #9 | concentrate_yield_g; header says 25g bypass but method says 15g |
Portugal | ultra-concentrate + high bypass inverted | 17.5 g | 5.71 | 50.8% final beverage | 121.3 s | 69.5 high | 0.279 outlier #12 | bypass_g; final beverage range 150-155g; midpoint used |
Qatar | moderate bypass inverted | 18 g | 8.33 | 23.3% final beverage | 115 s | 77.7 high | 0.229 outlier #18 | clicks_norm; swirl_count; method says upright but also says flip at 1:40; normalized as inverted |
Romania | high bypass inverted | 17 g | 7.06 | 43.8% final beverage | 105 s | 41.7 medium | 0.446 outlier #6 | dose_g; bypass_g; input dose 18g, used dose about 17g |
Singapore | fuller-volume extraction inverted | 18 g | 11.11 | 0% input water proxy | missing s | 49 medium | 0.379 outlier #5 | press timing not stated |
Thailand | high bypass missing | missing g | missing | 45.7% final beverage | 105 s | 25.1 medium | 0.373 outlier #10 | press_start_s; bypass_g; dose not stated |
UAE | low bypass inverted | 18 g | 10.00 | 14.3% input water proxy | 120 s | 75.5 medium | 0.204 outlier #21 | press duration and final beverage not stated |
Uruguay | fuller-volume extraction inverted | 17 g | 13.00 | 0% input water proxy | 110 s | 58.9 high | 0.340 outlier #7 | final beverage weight not stated |
Methods
What this analysis can and cannot claim.
This model is useful for comparing written recipes. It is not a lab measurement of extraction yield and it does not prove why a recipe placed where it did.
Extraction proxy
True extraction yield needs beverage TDS and beverage mass. Those were not available for most recipes. The page therefore uses a relative extraction-opportunity proxy from normalized grind, contact time, temperature, brew water per dose, and agitation. Bypass is not part of the extraction score.
Recipe distance
Pairwise distance uses a mixed-variable model over numeric recipe fields and categorical workflow fields. Missing values are skipped and the remaining weights are rebalanced. The result is a relative distance from 0 to 1 inside this dataset.
Data quality
The current dataset has 22 recipes. Water TDS, final beverage mass, and press duration are not equally complete across every recipe. Estimated or ambiguous fields stay visible in the table so the analysis does not hide weak spots.
Source basis: competitor-shared recipe text, public recipe pages where available, Instagram attribution, and manual identity checks from the WAC 2025 recipe dataset. The official podium recipe links are included in the table when available.
Practical guide
A short guide from the 2025 recipes.
These are not rules. They are the practical lessons I would carry into AeroPress competition prep after looking across the 2025 recipe field.
Do
- Start near a normal competition dose before chasing tiny dose changes.
- Track brew water and bypass water separately.
- Write down press start, press duration, and whether you stop before the hiss.
- Treat final beverage mass as its own target when using bypass.
- Test one major strategy at a time: extraction setup, dilution, water, or workflow.
Don't
- Do not copy the podium as one recipe family; the top 3 were meaningfully different.
- Do not treat bypass as extraction. Bypass changes the served cup, not what the grounds gave up.
- Do not compare recipes by dose alone. Dose barely moved compared with the rest of the field.
- Do not ignore missing source details. Unknown press or final beverage data changes the confidence of the analysis.
- Do not assume a higher extraction proxy is automatically better. The winner sat low on this proxy.
Read next
AeroPress championship prep
Use this to turn the WAC patterns into a practical prep framework before a competition.
AeroPress Wheel
Use the glossary and variable map when a recipe term needs a clear brewing definition.
AeroDial
Use the dial-in app to log your own test brews with the same separation between inputs and cup results.
Future updates
This page is built to accept more recipes later.
When more competitors share their 2025 recipes, the source rows can be updated and the analysis will recompute ranges, distances, rankings, and chart positions from the same model.
Caution on interpretation
A recipe can be close in the model and still taste different. A recipe can be far in the model and still score well. The model compares written recipe structure; judges score cups.