【Cold Brew as A/B Testing】Optimizing Water Recipes for Better Coffee Outcomes

In digital marketing, A/B testing is second nature: change one variable, observe the result, and optimize. What if I told you that making cold brew coffee works the same way?

More than 98% of your coffee is water. Just as a marketer tests headlines, images, or CTAs, a coffee lover can test water mineral ratios to “optimize conversions” in flavor. By systematically adjusting water composition, you run controlled experiments — with every brew as your dataset.

Coffee as an Experiment

When brewing cold brew coffee, the main constants are beans, grind size, and steeping time. The key variable is water quality.

• Magnesium (Mg²⁺): enhances sweetness, like boosting engagement.
• Calcium (Ca²⁺): adds body, like improving retention.
• Bicarbonates (HCO₃⁻): buffer acidity, like balancing user experience.

👉 In short: water chemistry is your testable variable — change it, and your cup changes.

Experimental Setup: Smartwater × Kirkland

To simulate “variant testing,” we use two readily available waters:

• Smartwater — distilled, electrolytes added (Ca ~10 mg/L, Mg ~15 mg/L, HCO₃⁻ ~0). Too “thin” alone.
• Kirkland (Chase Spring) — mineral-rich (Ca ~38 mg/L, Mg ~12 mg/L, HCO₃⁻ ~135 mg/L). Too alkaline alone.

Neither works perfectly on its own. But when combined in different ratios, they form effective test groups.

A/B Test: Water Ratios and Flavor Outcomes

Variant	Ratio (Kirkland : Smartwater)	Best For	Flavor “Results”	Interpretation
A	1 : 1	Medium-dark roasts (Brazil, Sumatra)	Full body, chocolatey, smooth sweetness	Optimize for conversion depth — rich, less acidic.
B	1 : 2	Light-medium roasts (Ethiopia, Colombia)	Balanced, fruity, refreshing	Balanced funnel: acidity + sweetness coexist.

Keep It Scientific: One Variable at a Time

In both A/B testing and coffee experiments, the golden rule is simple: change only one variable, keep everything else constant.

• If you change beans, grind size, water quality, and steeping time all at once, you’ll never know which factor caused the difference.
• In this cold brew experiment, the only variable we tested was water composition (Smartwater × Kirkland ratios).
• Everything else — beans, grind size, ratio, steeping — remained constant, so the results could be directly attributed to water.

💡 Just like in digital marketing, if you test two ads with different headlines and images and budgets, the data is useless. One variable at a time = clean insights.

Methodology (Cold Brew Protocol)

1. Beans (Audience Selection) — Medium-dark roast = chocolate/nutty (broad appeal). Light-medium roast = fruity/floral (niche segmentation).
2. Grind (Variable Consistency) — Coarse, like French press. Too fine = over-extracted bitterness (poor UX).
3. Ratio (Coffee : Water = Offer : Audience) — Standard: 1:8 – 1:10 (approachable, everyday brew). Concentrate: 1:5 – 1:6 (strong base for dilution).
4. Steeping (Campaign Duration) — Room temp: 8–12h → brighter, lighter. Fridge: 12–24h → smoother, sweeter.
5. Filter (Post-Processing) — Paper/mesh; double filter to remove “noise”.
6. Storage (Retention Phase) — Refrigerate up to 3–4 days; serve plain, with milk, or tonic/cocktail.

Remarks on Variant C (1:3)

While Variant C (1:3) works well for hot pour-over (highlighting bright acidity), in cold brew it often tastes too thin. We include it here for completeness, but most brewers will find Variant A (1:1) or Variant B (1:2) more effective for cold brew optimization.

Practical Application: Hario Cold Brew Bottle Recipe

Want to put this experiment into practice? The Hario Filter-in Bottle (650 ml) is perfect:

• Standard Cold Brew (1:10) — 65 g coarse-ground coffee + 650 ml water; steep 12–18h in fridge → balanced, everyday drink.
• Concentrate Cold Brew (1:5) — 100 g coffee + 500 ml water; steep 12–18h in fridge → stronger body, ideal with milk/ice.

💡 Pair with water recipes: Variant A (1:1) → richer, chocolate-heavy; Variant B (1:2) → refreshing, fruit-forward.

Key Insights for Professionals

• Treat water ratios as experiments. Don’t trust a single water source — blend, measure, compare.
• Segment by roast profile. Light roasts → Variant B (1:2); Dark roasts → Variant A (1:1).
• Iterate based on taste feedback. Taste = your KPI. Adjust ratios as you would bid strategies.

Conclusion

Cold brew is not just coffee; it’s a live experiment. With Smartwater + Kirkland in controlled ratios, you can run your own A/B tests: Variant A (1:1) → bold, chocolate; Variant B (1:2) → balanced, refreshing.

💡 In digital marketing, we optimize campaigns. In coffee, we optimize water. Principle: Test → Measure → Iterate.

This cold brew experiment is just one side of the story. For a deeper dive into brewing methods and variables beyond water, check out my previous post: Extended Guide to Perfect Coffee Brewing.

---

【冷泡咖啡的 A/B 測試實驗】用水質比例優化你的咖啡成果

在數位行銷中，我們習慣透過 A/B 測試來優化績效：一次只改變一個變數，觀察結果，再做最佳化。你知道嗎？冷泡咖啡其實也是一場 A/B 測試實驗。

一杯咖啡裡有 98% 是水。就像行銷人會測試廣告標題、圖片或 CTA，咖啡愛好者則可以測試 水中的礦物質比例，來「優化轉換率」——也就是風味。透過有系統的調整水質，你就能像跑數據一樣，得到一杯又一杯不同的實驗結果。

咖啡 = 一場實驗

在冷泡咖啡的實驗中，豆子、研磨度、浸泡時間幾乎是固定變數，唯一的 可控變數 就是水的成分：

• 鎂 (Mg²⁺)：帶出甜感，就像提升互動率。
• 鈣 (Ca²⁺)：增加厚度，就像提升留存率。
• 碳酸氫鹽 (HCO₃⁻)：緩衝酸度，就像平衡使用者體驗。

👉 換句話說：水質就是你的實驗因子 ——只要改變它，結果就會不同。

實驗設計：Smartwater × Kirkland

• Smartwater：蒸餾後加電解質（Ca 約 10、Mg 約 15、HCO₃⁻ 幾乎 0），單用偏薄。
• Kirkland (Chase Spring)：礦物質較高（Ca 約 38、Mg 約 12、HCO₃⁻ 約 135），單用鹼度偏高。

兩者單用都不理想，但以不同比例混合，就能形成清晰的測試組。

A/B 測試：水質比例 × 風味結果

實驗組	比例 (Kirkland : Smartwater)	適合烘焙	風味結果	行銷語言解讀
A	1 : 1	中深焙（巴西、曼特寧）	厚實、巧克力感、甜感圓潤	專注「轉換深度」→ 飽滿、低酸。
B	1 : 2	淺中焙（衣索比亞、哥倫比亞）	酸甜平衡、清爽果香	「平衡型漏斗」→ 酸甜並存。

保持科學：一次只改變一個變數

無論是 A/B 測試還是咖啡實驗，都有一個黃金法則：一次只改變一個變數，其他條件保持一致。

• 同時改變豆子、研磨度、水質、浸泡時間，將無法判斷是何因素造成差異。
• 本實驗唯一測試的變數是 水質比例（Smartwater × Kirkland）。
• 其他條件（豆子、研磨度、粉水比、浸泡方式）皆固定，才能把差異明確歸因於「水質」。

💡 就像行銷裡，如果同時改動標題、圖片和預算，你的 A/B 測試數據就沒有參考價值。一次只改一個變數，才能得到乾淨的洞察。

實驗流程（Cold Brew Protocol）

1. 豆子（Audience Segmentation）：中深焙＝巧克力/堅果（大眾）；淺中焙＝果酸/花香（利基）。
2. 研磨度（變數一致性）：粗研磨近法壓壺；太細會過萃、苦澀。
3. 粉水比（Offer : Audience）：標準 1:8–1:10；濃縮 1:5–1:6（後續可稀釋）。
4. 浸泡（Campaign Duration）：室溫 8–12 小時較清爽；冷藏 12–24 小時更順滑甜感佳。
5. 過濾（後期清理）：濾紙/濾袋；建議二次過濾以提升澄清度。
6. 保存（Retention Phase）：冷藏 3–4 天內飲用最佳；可加冰/牛奶/氣泡水或調酒。

備註：C 組 (1:3)

C 組 (1:3) 在熱手沖時能突顯酸質與乾淨甜感，但在 冷泡咖啡中常顯得過於單薄。因此僅作補充參考；多數情況下，A 組 (1:1) 或 B 組 (1:2) 更適合用於冷泡最佳化。

實際應用：Hario 冷泡瓶配方

想把實驗落地？Hario 冷泡瓶（650 ml） 很適合：

• 標準冷泡（1:10）：咖啡粉 65 g + 水 650 ml；冷藏 12–18 小時 → 酸甜平衡、日常直飲。
• 濃縮冷泡（1:5）：咖啡粉 100 g + 水 500 ml；冷藏 12–18 小時 → 濃郁厚實，適合加牛奶/冰塊。

💡 水質搭配：A 組 (1:1) → 厚實巧克力風；B 組 (1:2) → 清爽果香型。

專業人士的啟發

• 水質比例 = 實驗變數：別只用一種水源；混合、量測、比較。
• 依烘焙分眾：淺中焙 → B 組 (1:2)；中深焙 → A 組 (1:1)。
• 迭代優化：味覺就是 KPI；像調整出價策略那樣調整比例。

結論

冷泡咖啡，不只是飲品，而是一個活生生的實驗。透過 Smartwater × Kirkland 的 A/B 測試，你能快速找到自己的最佳解：A 組 (1:1) 厚實甜感、巧克力感強；B 組 (1:2) 酸甜平衡、清爽果香。

💡 就像數位行銷一樣：Test → Measure → Iterate。

這次的冷泡實驗只是故事的一部分。如果你想更全面了解沖煮方法與其他變數，建議閱讀我之前的文章：完美咖啡沖煮延伸指南。

---

Author / 作者：— Dr. Ken FONG

Keywords： Cold brew A/B testing, coffee experiment, Smartwater coffee, Kirkland cold brew, digital marketing coffee analogy, 冷泡咖啡水質, A/B 測試咖啡, Smartwater 冷泡, Kirkland 冷泡