成果範例
Data analysis: 9 months of bookings at a family-run guesthouse
Key numbers
| Metric | Value |
|---|---|
| Bookings (excl. cancelled) | 251 |
| Room nights sold | 689 |
| Revenue | $78,540 |
| Average rate per night | $114 |
| Cancellation rate | 11.6% (33 of 284) |
| Average stay | 2.7 nights |
Summary by channel
| Channel | Bookings | Share of revenue | Avg. rate | Commission | Net per night |
|---|---|---|---|---|---|
| Booking platforms | 142 | 54% | $109 | 15% | $93 |
| Direct (phone, email, website) | 87 | 38% | $126 | 0% | $126 |
| Local travel agent | 22 | 8% | $102 | 10% | $92 |
Direct guests pay more, stay longer (3.4 nights vs 2.3) and bring $33 more per night after commission than platform guests.
Trends and outliers
- July and August made 41% of revenue; occupancy reached 93%.
- March and April were weakest (occupancy 38%), but weekend nights still sold at 79%: the gap is midweek.
- Cancellations were 19% for platform bookings with free cancellation and 4% for direct bookings with a deposit.
- Outlier: 6 bookings in May at $45 per night. These are all one corporate group with a manual discount; they were kept in totals but flagged.
Findings
- Your most profitable guests are direct, returning guests, yet only 14% of platform guests were invited to book direct next time.
- Midweek in spring is the only real gap; weekends sell themselves.
- Free cancellation on platforms costs about 27 lost nights over 9 months.
Recommendations
- Add a "book direct" card at check-out with a 10% returning-guest rate: still $20 above platform net.
- Create a midweek spring offer (3 nights for the price of 2, Sun–Thu) for March–April.
- Switch platform listings to a non-refundable rate at 8% off alongside the flexible one.
- Track the booking channel for every guest from now on: 17 rows had no channel and were grouped as "direct".
All totals were cross-checked against the monthly sums in your sheet; the difference is $0.
Client: a 12-room guesthouse by a lake, run by two owners Data: 284 booking rows, January to September (date, room type, nights, channel, price paid, cancellation) Question asked: "Which channel and which months actually make us money, and where are we losing it?"
客戶與所有細節皆為虛構。
關於此服務
一週後看起來好 12% 的版本，可能純屬偶然；在第一個好日子就停止測試，是上線一個毫無作用的改動的常見原因。我們根據各版本的訪客數與轉換數，計算轉換率、提升幅度、統計顯著性與信賴區間，並檢查樣本是否足夠。您會得到明確的結論：採用新版本、保留原版，或視為尚無定論而延長測試。
包含內容
- 您會收到的內容測試報告，含轉換率、提升幅度、顯著性、信賴區間、樣本數檢查、進階方案另附族群細分結果，以及明確結論，以格式化文字搭配表格呈現在訂單頁面。另提供 Word (.docx) 報告，表格以 Excel (.xlsx) 提供。
- 轉換與提升每個版本都有
- 顯著性以及信賴區間
- 樣本數檢查測試時間是否足夠
- 族群細分裝置、來源或新使用者
- 明確結論採用、保留或延長測試
服務流程
- 1需求說明由您貼上或附上各版本的數字
- 2分析我們檢驗顯著性與樣本
- 3結論您會收到決定與理由
需要您提供的資料
各版本的訪客數與轉換數（最好依日期）、您測試的指標、測試進行了多久，以及各版本之間改動了什麼。
我們會詢問您：
- 測試了什麼？
- 哪個指標決定勝負？
- 測試進行了多久？
- 是用哪個工具進行測試？
比較方案
付款保障
下單時付款，但只有在驗收成果後，款項才會撥付。
- 固定價格付款後，方案價格不會變動。
- 款項先行保留付款時，金額只會先保留，在您驗收後才會撥付給服務。
- 交付前檢查每份成果在交到您手上之前都會經過品質檢查，您也可以在驗收前於訂單頁面查看成果。
- 免費修改在需求說明範圍內，可依方案允許的次數提出修改。
- 3 天後自動驗收若您在交付後 3 天內未驗收也未提出修改，成果將自動驗收。若您在交付後於訂單聊天室留言，訂單會暫停自動驗收；我們回覆後，可能會重新起算 3 天，並會在聊天室中告知。
- 重新製作或全額退款若成果不符合您的需求說明且修改後仍未改善，或製作時間超過交付期限的兩倍，請在訂單聊天室留言：我們將重新製作，或全額退還款項。
常見問題
我的工具已經說這個版本贏了，為什麼還要檢驗？
各工具的方法不同，有些會太早顯示「贏家」。對原始數字做獨立檢驗，能讓您知道有多大把握。
可以告訴我下次測試要跑多久嗎？
可以，我們會根據您的流量與目前的轉換率，估算偵測到您在意的提升幅度所需的樣本數。
結果如果尚無定論怎麼辦？
這也是有效的答案。我們會說明還需要多少流量，或差異是否小到不足以在意。
沒有找到您要的答案？