決策型產品設計實踐

{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"決策型產品是通過理解業務需求，建立業務模型，提供數據展示與分析，進而實現自動決策的產品。"},{"type":"text","text":"本文將結合我們在嚴選供應鏈的實踐，分析在設計一個決策型產品時，最應該關注什麼，"},{"type":"text","text":"產品設計成功的關鍵點是什麼"},{"type":"text","text":"等問題。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"1. 什麼是決策型產品"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"大數據技術的成熟及其在企業決策流程中的廣泛應用，讓基於數據的各類"},{"type":"text","marks":[{"type":"strong"}],"text":"業務產品"},{"type":"text","text":"在企業產品矩陣中的地位越來越重要。此類業務產品按照對於業務決策支持能力的強弱，可以分爲三個發展階段："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"numberedlist","attrs":{"start":1,"normalizeStart":1},"content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":1,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"第一階段：報表型產品"},{"type":"text","text":"，爲業務人員提供各類數據報表，幫助他們瞭解業務的運行狀況，業務人員可以藉助報表展示的數據進行分析和決策；"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":2,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"第二階段，分析型產品"},{"type":"text","text":"，分析型產品更進一步，除了數據展示外，將成熟的業務分析思路嵌入產品，提供不同程度的數據分析能力，增強了輔助業務決策的能力；"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":3,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"第三階段，決策型產品"},{"type":"text","text":"，利用數據、規則、算法進行自動化決策，極大提高業務的自動化程度和決策水平；"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我們認爲，"},{"type":"text","marks":[{"type":"strong"}],"text":"決策型產品是通過理解業務需求，建立業務模型，提供數據展示與分析，進而實現自動決策的產品"},{"type":"text","text":"。本文將結合我們在嚴選供應鏈的實踐，分析在設計一個決策型產品時，最應該關注什麼？產品成功的關鍵點是什麼？等等問題。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/5c\/70\/5c0108b71bdb797c2539a76e9c919070.jpg","alt":null,"title":"","style":[{"key":"width","value":"100%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"2. 決策型產品的設計路線"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我們認爲，決策型產品的核心要素有三個：模型+數據+策略。與此相對應的，爲了能夠讓一個決策型產品成功落地，主要需要做好三件事："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"numberedlist","attrs":{"start":1,"normalizeStart":1},"content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":1,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"構建領域模型"},{"type":"text","text":"，即要充分抓住決策型產品所解決的真實的業務問題，構建精準且可擴展的領域模型；"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":2,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"挖掘數據價值"},{"type":"text","text":"，即要從業務、流程、系統等各個層面挖掘數據價值，以數據驅動業務決策；"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":3,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"打造決策閉環"},{"type":"text","text":"，即要讓決策系統是一個可以持續優化的閉環，以不斷滿足快速變化的業務需求；"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我們會在下面的章節展開對這三個核心要素的討論。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/8f\/5e\/8f56fb76ac9851fd87f648ba0e3ec95e.jpg","alt":null,"title":"","style":[{"key":"width","value":"100%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"3. 構建領域模型"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"決策型產品圍繞一個或一組業務場景進行設計，專注解決特定的業務問題，因此，"},{"type":"text","marks":[{"type":"strong"}],"text":"深挖業務規則、理解業務流程、明確業務指標"},{"type":"text","text":"是構建業務模型的突破口。可以說，一套精準又強大的業務模型，是一個決策型產品的靈魂。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我們所理解的構建業務模型通常是產品技術人員的工作。雖然這件事情很重要，但由於團隊責任劃分，通常負責需求的業務人員不會參與業務模型的構建，而是將需求通過層層傳遞的方式告知給產品技術人員，這可能造成關鍵信息丟失，進而導致構建的業務模型的精準性和擴展性不足。這種模型構建方式的直接後果是："},{"type":"text","marks":[{"type":"strong"}],"text":"業務人員腦子裏的模型、產品人員腦子裏的模型與開發真正實現在代碼裏的模型都不一樣"},{"type":"text","text":"，最終造成不同角色間溝通困難，代碼持續腐化，新需求開發週期變長等問題。那應該怎麼做呢？下面提出一些解決該問題的關鍵點。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/2d\/65\/2d163fa21508af2b02583713cb932a65.jpg","alt":null,"title":"","style":[{"key":"width","value":"100%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.1 達成廣泛共識"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"決策型產品設計之初，就應該要求參與產品的各方，都能夠對於業務模型、業務流程和業務指標達成共識。該實踐的重點是，構建業務模型的工作不能僅僅是由產品技術人員來完成，而是應該由整個團隊（包含業務人員、產品、技術、數據等角色）共同完成，讓每個角色都能夠對業務模型所包含的模塊，模塊之間的關係等有深刻的、一致的認知。在領域驅動設計（DDD）的實踐中提出了一種稱爲“事件風暴”的方法：業務專家向團隊所有成員描述詳細的業務流程，整個團隊積極參與討論，並在過程中找到核心域、非核心域，明確實體、聚合、事件等關鍵元素，最終完成業務模型構建。在具體的實踐中，能否高質量地進行事件風暴，除了很考驗每一位參與者的專注度與執行力之外，"},{"type":"text","marks":[{"type":"strong"}],"text":"還特別需要一名業務專家的積極投入"},{"type":"text","text":"。因此必須要讓業務專家意識到，一方面他的專業知識對於業務模型的構建至關重要，另一方面一套清晰、精準、強大的業務模型也能大大幫助他完成業務目標。即在這個過程中，"},{"type":"text","marks":[{"type":"strong"}],"text":"要讓業務專家由提需求的人，變成協助\/甚至主導構建業務模型的人"},{"type":"text","text":"。除了對業務模型的構建達成共識外，"},{"type":"text","marks":[{"type":"strong"}],"text":"還需要各方對於整個業務未來的發展取得基本共識"},{"type":"text","text":"，例如未來業務會重點朝哪個方向努力，哪個業務域的可挖掘價值最高等等。這麼做的目的是能夠讓產品技術團隊進行一定的提前佈局：例如是否需要拆分微服務、是否可以將非核心的模塊外包、是否應該在某個模塊開發時投入主要精力等。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.2 統一溝通語言"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"達成廣泛共識的最終狀態應該是：在未來需求迭代、系統演進的過程中，各個相關方能夠使用同一套話語體系（Ubiquitous Language）進行溝通。這不僅避免了各方在溝通過程中的信息失真，更重要的是可以讓大家都能夠在同一個思維框架內去認知和討論。當各方對於領域模型達成廣泛共識，並能夠用同一套語言進行溝通時，就可以大大提高業務需求從提出到落地的效率。除此之外，還有其他諸多好處："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"numberedlist","attrs":{"start":1,"normalizeStart":1},"content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":1,"align":null,"origin":null},"content":[{"type":"text","text":"當面臨一個業務需求時，有兩種思考方式，一種是根據系統現狀，直接實現業務需求；另一種更好的方式是，對照領域模型，對業務需求進行結構化分析："}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"numberedlist","attrs":{"start":1,"normalizeStart":1},"content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":1,"align":null,"origin":null},"content":[{"type":"text","text":"爲了支持業務需求，核心域是否需要進行能力擴充（即對核心業務模型進行擴展）"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":2,"align":null,"origin":null},"content":[{"type":"text","text":"爲了支持業務需求，支撐域是否需要新增處理邏輯（即對支撐業務邏輯進行調整）"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":3,"align":null,"origin":null},"content":[{"type":"text","text":"各個子域之間的交互方式是否需要升級"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"numberedlist","attrs":{"start":2,"normalizeStart":2},"content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":2,"align":null,"origin":null},"content":[{"type":"text","text":"當提出需求的人和實現需求的人對於領域模型有共同的認知時，那他們也能更好地就需求的投入產出比達成共識（而與此相對應的是，提需求的人對於需求實現的難度沒有概念，造成開發資源的浪費）；"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"3.3 挖掘核心業務指標"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我們首先需要明白一個道理：“"},{"type":"text","marks":[{"type":"strong"}],"text":"若無法衡量，則無法優化"},{"type":"text","text":"”。對於一個決策型產品而言，我們必須找到一系列指標，來衡量產品做出的各類決策的質量優劣，進而能夠圍繞這些指標，來持續進行決策功能的優化。同時，對於一個團隊來說，共同挖掘業務指標的過程，也是一次發現價值的過程。大家需要回答這些問題："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"numberedlist","attrs":{"start":1,"normalizeStart":1},"content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":1,"align":null,"origin":null},"content":[{"type":"text","text":"什麼是這個產品需要滿足的最核心價值？"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":2,"align":null,"origin":null},"content":[{"type":"text","text":"什麼是這個產品取得成功的最關鍵因素？"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":3,"align":null,"origin":null},"content":[{"type":"text","text":"什麼指標可以幫助我們衡量產品是否取得成功？"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":4,"align":null,"origin":null},"content":[{"type":"text","text":"什麼因素真正影響着這些業務指標的變化？"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"還有一些構建指標過程中的注意事項："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"numberedlist","attrs":{"start":1,"normalizeStart":1},"content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":1,"align":null,"origin":null},"content":[{"type":"text","text":"業務指標應該是分層的，哪些是核心指標，哪些是非核心指標，一個核心指標下面關聯哪些非核心指標。分清核心指標和非核心指標，對於產品的設計也有很大的影響，例如在我們的一些供應鏈產品中，主頁面只需要展示最核心的庫轉和缺貨等核心指標即可；"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":2,"align":null,"origin":null},"content":[{"type":"text","text":"業務指標彼此可能存在衝突，例如成本和用戶體驗，通常爲了優化一個，就必須在一定程度上犧牲另一個。我們應該結合業務發展現狀，來選擇當前業務更加關心的指標，找到影響這些指標的關鍵因素來進行優化；"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":3,"align":null,"origin":null},"content":[{"type":"text","text":"業務指標的評價口徑不是一成不變的，而是應該隨着業務的發展而發展，定義業務指標的目的是爲了更好地理解業務的運行狀態，挖掘系統下一步的優化目標，因此當指標的統計口徑不能很好地指導業務發展時，就應該改變指標的統計口徑；"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"4. 挖掘數據價值"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"對於互聯網這樣的數字化業務，業務流程中的方方面面都需要利用數據。一個經得起市場檢驗的決策，是基於儘可能廣泛的數據收集和儘可能嚴謹的邏輯分析基礎上得來的。因此，決策型產品的第二個關鍵要素是對業務數據的分層抽象與使用，深度挖掘數據價值。我們可以將數據的加工與使用分爲三層："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"第一層，"},{"type":"text","marks":[{"type":"strong"}],"text":"數據只是一串通過觀察而記錄下的符號"},{"type":"text","text":"，所含內容十分有限，無法回答特定的問題。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"第二層，通過加工解釋、分析數據間的關係獲得了信息，"},{"type":"text","marks":[{"type":"strong"}],"text":"信息是具有邏輯關係的數據"},{"type":"text","text":"，但此時信息中可能包含大量的冗餘內容，還不足以指導用戶決策。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"第三層，進一步對龐大無序的信息進行管理和分類，"},{"type":"text","marks":[{"type":"strong"}],"text":"知識是從相關信息中過濾、提煉及加工而得到的有用資料"},{"type":"text","text":"。具備知識的決策產品能夠直接推動用戶的決策和行爲，加速行爲過程。"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"舉個例子來說明。例如，某個商品在A倉庫有10件，在B倉庫有20件，C倉庫有15件，一個用戶訂單需要購買5件。這是數據，只是單純的記錄，無法幫助業務決策應該從哪個倉庫給用戶發貨是最優的。繼續分析發現，用戶地址離A倉庫最近，配送時效更快，但是A倉庫的庫內操作費和配送費都比B、C倉庫更高，同時，B倉庫存在產能不足的問題、C倉庫的商品即將過期……可以看到，信息開始變多變雜，我們開始面臨新的問題：如何確定這些信息的價值或者說優先級，從而制定最佳決策呢？此時，我們可以進一步分析信息之間的關係，並確定業務最關心的指標。例如，將影響調度結果的信息歸類爲成本、時效、產能、效期等因素，明確信息之間的協同或互斥關係。當業務目標是確定以用戶體驗最優爲導向時，我們便可以根據業務模型快速構建決策鏈路，最終得到決策建議，從最近最快的A倉庫爲這位用戶發貨。設計決策型產品的一大核心在於，"},{"type":"text","marks":[{"type":"strong"}],"text":"需要圍繞業務目標對數據進行“數據-信息-知識”的逐步提煉"},{"type":"text","text":"，最終系統可以在一個完善的知識體系下實現“因地制宜”的決策效果。在這個過程中，我們認爲，主要應該關注數據在三個方面的價值表現。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/6d\/7d\/6dfd68ae1204ba90c26c21f0ecf8c47d.jpg","alt":null,"title":"","style":[{"key":"width","value":"100%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"4.1 業務價值"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"作爲面向企業內部的產品，決策型產品的一大重點在於爲業務指標服務。"},{"type":"text","marks":[{"type":"strong"}],"text":"滿足業務需求、解決業務痛點是決策型產品需要具備的核心能力，而數據則是這項能力的關鍵載體"},{"type":"text","text":"。在規劃決策型產品時，不僅要清楚數據在整個業務流程中的運轉規律，更需要釐清原始數據在每一次使用和運算後產生的新信息的含義，再結合業務規則和邏輯對新信息進行加工。用一個例子來說明。在庫存平衡的業務場景中，調撥人員每天都要創建倉庫與倉庫之間的調撥單據，當系統只記錄和提供各種商品在每個倉庫的數量時，在調撥場景中是沒有業務價值的，我們不僅需要提供庫存的實時數據，更重要的是滿足調撥核心需求——該從哪個倉調撥哪些商品調撥多少數量到哪個倉，纔可以幫助庫存平衡，從而達成業務指標。我們同樣可以通過以下幾步來解決問題："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"第一步，挖掘倉庫、商品、庫存、時間、路線等多維度數據之間的關聯關係，獲得更多信息；"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"第二步，基於以上信息，構建合理、模塊化的業務模型，規範統一輸入參數，也就是搭建知識體系；"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"第三步，綜合以上所有條件並集合算法計算出符合核心目標的最佳組合；"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"最後，覆盤每次調撥決策中存在的異常和問題，作爲下一次對模型或參數優化的輸入。"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"決策型產品要讓數據爲目標所用，冗餘的數據會讓用戶困惑，每一個展示給用戶的數據在特定場景下都應該有它的業務價值所在。隨着系統的自動化程度變高，我們向業務人員提供的數據重心也會逐步從信息展示類轉向風險預警類和異常分析類，一方面提高系統輔助業務決策的前瞻性，另一方面也爲下一步的自動化決策做準備。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"4.2 流程價值"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在系統還未發展成熟的階段，業務人員往往會通過各種各樣的方式達成業務目的。即使在同一個業務場景下，不同的業務人員操作流程也很有可能是不一樣的。千人千面的用戶習慣對於整體業務流程來說其實是一種資源損耗，業務人員可能花費大量精力在數據的獲取和清洗上，甚至是僅根據少量數據拍腦袋做決策，可能導致一些意想不到的異常情況和後續很難解釋的補丁流程。企業內部產品實現提效的根本在於作業流程標準化，因此，"},{"type":"text","marks":[{"type":"strong"}],"text":"找出不同業務行爲背後的根本問題，通過構建合理的數據模型重組業務流程，促進流程標準化，是決策型產品應該從數據中挖掘到的第二重價值。"},{"type":"text","text":"我們還是以調撥爲例。在人工處理時期，業務需要根據採購計劃、銷售活動、庫存水位、倉庫路線、調撥成本等信息綜合判斷，人工篩選、計算每條路線需要調撥的商品數量，計算結果受個人影響大且難以監控。可以看到，在整個過程中，使用到了銷售、商品、採購、庫存、配送等多個維度的數據，而業務如何使用這些數據其實是不明朗的。決策型產品需要非常清晰地定義數據使用分幾個階段、每個階段使用哪些數據、使用的順序是什麼、不同數據的權重如何等問題，我們要明確將哪些流程封裝到系統進行自動化，哪些流程仍然需要業務人工處理，而這個設計也非常依賴於數據的可靠性、完整性和準確性。通過對數據的定義與抽象分析，業務流程中的共性與特點也將脫穎而出，成爲我們標準化流程的關鍵。此外，在我們設計產品的過程中還需要保持獨立的產品視角，明確產品的階段與發展形態，在承接業務需求時，儘量與產品的發展路線相匹配，而不是一味地進行功能點的堆砌。對業務需求和規則做必要的抽象，沉澱爲系統的通用能力，從而可以滿足未來更多的業務場景。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"4.3 系統價值"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"打造數據流轉閉環、反哺系統建設是決策型產品發展歷程中的高級形態"},{"type":"text","text":"。當前我們使用的多數內部決策型產品基本還處於輔助決策的狀態，這裏的“輔助”不僅僅指系統對於業務來說，是一種數據分析與計算的工具，不具有完全決策能力，也表示對於產品本身而言，還缺乏一種自我迭代機制，用戶的行爲數據還未實現反向輸入或分析，產品的更新與進步仍然需要依賴於用戶主動的自我意識，導致產品品質與使用者的專業素質關係綁定很深。反向分析用戶行爲數據的過程其實是一種對用戶已有經驗知識的逆向解析，目前大多數情況下，我們通過人爲儘可能地廣泛收集用戶工作方法與經驗，將用戶通過經驗形成的知識解構爲信息和數據，再嵌入到產品中供系統識別和使用。這樣做的缺點在於往往受個人偏好的影響較大。智能決策型產品則需要具備自我學習能力，可以基於更多角度的、更深度的、更實時的信息和知識進行解析和覆盤，獲得新的知識與技能，持續優化決策。此時，決策型產品便完成了正向“數據分揀-信息聚類-知識架構”和逆向“知識解構-信息轉化-數據捕捉”的完整數據閉環。這也是決策型產品設計的第三大核心“打造決策閉環”的前提。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"5. 打造決策閉環"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"決策型產品的最終目標，是能夠實現自動化決策，取代業務人員日常決策的工作。實際上，實現自動化決策本身其實並不難，"},{"type":"text","marks":[{"type":"strong"}],"text":"難的是如何實現一套可持續迭代優化的自動化決策系統"},{"type":"text","text":"。我們把這項工作稱爲打造決策閉環，即實現“決策→覆盤→優化→再決策”的閉環。不難理解，只有完成了決策閉環的構建，纔算真正意義上實現了自動化決策。進一步，通過一輪輪決策閉環的迭代優化，完善策略，強化算法，才能發揮決策型產品的價值，將業務人員從日常的決策工作中解放出來。下面詳細介紹，要實現決策閉環，有哪些關鍵步驟。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/ce\/6e\/ce7b83a75e212374598c817fb0bae26e.jpg","alt":null,"title":"","style":[{"key":"width","value":"100%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"5.1 構建評價體系"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"一套"},{"type":"text","marks":[{"type":"strong"}],"text":"完備的、標準化的"},{"type":"text","text":"評價體系至關重要。上文我們已經討論過需要挖掘關鍵的業務指標，而評價體系裏除了業務指標外，還包含兩類指標："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"面向業務的中間指標 —— 例如供應鏈的核心業務指標爲默認倉滿足率，而我們可以根據系統優化的需要，定義諸如最優倉滿足率等指標，幫助我們對系統進行更準確的評價；"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"面向系統的技術指標 —— 例如系統的訂單處理能力、算法執行速度等指標；"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在構建評價體系的過程中，有幾點需要注意："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"numberedlist","attrs":{"start":1,"normalizeStart":1},"content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":1,"align":null,"origin":null},"content":[{"type":"text","text":"構建的評價體系是需要多層次的、完備的，能夠對系統的方方面面進行評價。舉例嚴選的倉配決策系統，我們有針對訂單層級的指標，如拆單率等，也包含運單層級的指標，如次日達率等；"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":2,"align":null,"origin":null},"content":[{"type":"text","text":"在評價體系裏實現的業務指標，一定要和業務方認可的核心指標對齊（即需要被業務方認可），最好是能夠直接接入真實的業務核心指標計算邏輯裏。這麼做的好處是，評價體系容易被業務方認可，同時一旦業務指標的計算邏輯有更新，也能夠馬上反應到評價體系裏；"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"5.2 覆盤與模擬"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"爲了能夠讓決策型產品做出的決策質量更高，必須持續對每一次決策進行復盤，而覆盤的關鍵，就是要看每一次決策對於評價體系的各項指標產生什麼影響。更進一步，能夠分析出指標變化的原因，從而在下一次的決策過程中調整參數，以期望得到更好的決策結果。上面所說的這個“決策→覆盤→優化→再決策”的過程，在實際執行的過程中會比較複雜，主要原因是："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"numberedlist","attrs":{"start":1,"normalizeStart":1},"content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":1,"align":null,"origin":null},"content":[{"type":"text","text":"實際中的業務指標，經常受到很多因素的共同影響，所以很難簡單的歸因於某一個具體的決策導致了業務指標的變化。例如對於用戶訂單的拆單率，可能受到用戶訂單數量、庫存分佈情況、倉庫產能等數十個因素的影響，如果我們想要找到拆單率變化與某個決策之間的關係，其實是很困難的；"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":2,"align":null,"origin":null},"content":[{"type":"text","text":"實際中的業務指標，可能需要很長時間才能夠真正體現出影響來，這導致了從覆盤、優化到再決策的週期很長，系統優化的效率低下。例如商品的採購策略，可能會影響未來幾周甚至幾個月的銷售情況、庫存成本等，我們不可能真的等到這批庫存賣完，才能對採購的下一次決策進行優化；"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"爲了解決這個問題，系統在設計的時候，可以考慮引入仿真模擬系統。簡單來說，仿真模擬系統是基於業務模型構建的一套虛擬運行環境，通過控制整個系統變量的數量，來對系統決策如何影響指標變化進行量化。仿真模擬系統能夠大大加速覆盤週期，例如快速將用戶過去半年的訂單以事件流的方式進行重放，讓“決策→覆盤→優化→再決策”的閉環能夠真正落地。因此，我們應該考慮圍繞仿真模擬系統去打造決策反饋閉環。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"5.3 圍繞模擬的反饋閉環"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"需要說明，我們所說的“決策”，實際上是決策產品在面對某個業務場景時，所執行的一套規則或一組算法，而不論是規則或算法，都是由一系列的參數來控制的。因此本質上，我們說的覆盤，除了對於規則或算法本身的更新修正之外，更多的是來複盤面對某個業務場景是，對應的各項參數配置是否是“最優的”"},{"type":"text","marks":[{"type":"strong"}],"text":"（注意：我們討論規則是否“最優”，一定要和業務場景相結合，不難理解，庫存管理系統針對大促時期和平銷時期的最優參數配置一定是不同的）"},{"type":"text","text":"。爲什麼需要圍繞仿真模擬系統打造反饋閉環呢？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"numberedlist","attrs":{"start":1,"normalizeStart":1},"content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":1,"align":null,"origin":null},"content":[{"type":"text","text":"仿真模擬系統可以根據歷史數據和未來期望數據，在決策前即對可能的業務結果進行仿真模擬，明確決策對於業務各項指標的影響，達到決策有預期，避免錯誤決策；"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":2,"align":null,"origin":null},"content":[{"type":"text","text":"在決策後，仿真模擬系統可以持續對影響決策的參數進行模擬，嘗試尋找更優的參數配置。一旦找到了更優的參數，就可以對線上的決策參數進行替換了。當然，在產品設計時，當發現更好的參數配置，有兩種做法：一種是將更好的參數及其模擬結果告知業務方，由業務人員確認，另一種是直接由系統替換配置，再告知業務人員配置變更的原因。選擇哪種方式，取決於系統的成熟度，通常都是從第一種方式向第二種方式逐漸過渡；"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":3,"align":null,"origin":null},"content":[{"type":"text","text":"即使在多輪模擬後、甚至第一輪配置時就找到了決策對應的最優參數配置，但是由於決策面臨的外部場景會發生變化，決策依賴的參數配置也需要隨之改變。通過仿真模擬，我們可以以更短的週期（例如每天）結合最新的業務變化情況與爲未來的預測情況進行模擬，從而找到與未來情況更匹配的決策參數；"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"結尾"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"隨着企業各類業務線上化、自動化和智能化的推進，會出現越來越多的決策型產品，幫助業務人員做好各類業務決策。我們相信，在推進決策型產品落地的過程中，如果圍繞我們提出的核心要素進行設計和實踐，能夠幫助大家少走彎路，一次把事情做對。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"作者簡介"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"麥格，網易嚴選供應鏈策略系統產品經理，負責嚴選供應鏈策略系統建設，專注於供應鏈訂單調度、庫存平衡、仿真模擬等方向。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"斯內普，網易嚴選供應鏈技術架構師，負責嚴選供應鏈策略系統建設，目前專注於供應鏈系統建設、仿真模擬系統實踐。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"horizontalrule"},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"頭圖"},{"type":"text","text":"：Unsplash"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"原文"},{"type":"text","text":"："},{"type":"link","attrs":{"href":"https:\/\/mp.weixin.qq.com\/s\/08PU8nGhhalmzVIMH2T-4A","title":"","type":null},"content":[{"type":"text","text":"決策型產品設計實踐"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"來源"},{"type":"text","text":"：嚴選技術產品團隊 - 微信公衆號 [ID：YanxuanTechProd]"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"轉載"},{"type":"text","text":"：著作權歸作者所有。商業轉載請聯繫作者獲得授權，非商業轉載請註明出處。"}]}]}