由個人化走向預測化：如何駕馭下一代的「大體驗潮」

	全球型的創新公司創辦人說：為什麼在整個顧客體驗的週期中，由機器來代替行銷人員是比較聰明的做法？
	Peter Sena II, Founder of Global Innovation & Design Firm Digital Surgeons, shares why leveraging machines for customer experience is smarter than we think.
	彼得席納，全球型的創新及設計公司 Digital Surgeon 的創辦人分享給我們，說明為何由機器來追蹤顧客經驗，會比我們想像的更加聰明。
	Marketers have been trying to master personalization since the moment the gesture driven interfaces and small screens of mobile devices rendered desktop customizations useless.
	自從取代桌上型裝置的移動裝置介面被手勢動作和小螢幕所主導以來，行銷人員一直試圖著要了解使用者的性格
	Who’s my audience? How do I personalize the creative? How do I personalize the offer? How do I figure out how to engage consumers in a more meaningful way? What can I show that’s going to drive purchase?
	誰是我的受眾？我要如何讓創新更加個人化？讓提供的內容更加個人化？如何讓人們更有意義地參與進來？要怎麼表現才更能驅動人們的購買行為？
	Amazon wrote the book on personalization, using the comprehensive shopping data they have about each consumer to personalize everything from their homepage to the offers they receive—all with the goal to drive purchase behavior and increase loyalty.
	亞馬遜所寫的書中提到「個人化」，它利用了他們所擁有的、每一個顧客廣泛的購物數據進行歸納，將每一個人「個人化」的內容，從首頁到每個人收到的購物條件，連結到「達成購買行為」和「增加忠誠度」這兩個目標上。
	But just as personalization made customization look trite, smart(er) AI powered by machine learning and natural language processing (NLP) engines is poised to replace the personalized with the predicted in our user experience.
	但光是個人化，顯得顧客化的做法有些老掉牙了，比較聰明的人工智慧，藉由機器的自我學習和語言自然化程序，可以平衡掉人類在個人化經驗方面先入為主的預測。
	Empowered by machine learning and natural language processing we can distill the unprecedented amount of consumer behavioral data now available to form predictive analytics that can begin to make sense of today’s fragmented consumer journey. When paired with advances in automation technology, these predictive analytics allow us to deliver tailored consumer experiences in real-time that put the personalized to shame.
	由機器自我學習和語言自然化程序賦予的功能，可以使我們提煉出前所未有顧客行為的數據量，並可用於目前的預測性分析，合理解釋現今消費者各種片段化的消費軌跡。藉著搭配自動化科技，這些預測性的分析可以讓我們及時地為消費者訂製出足以讓個人化做法相形見拙的消費經驗去。
	The Devil In In The Defaults
	預設值中的惡魔
	People overwhelming use default settings. Sure, there’s always a small percentage of power users, but most people take what is given to them. With that in mind, it’s not about customization or even merely personalization, it’s about the ability to have a predictive layer in your application that leads to perfect personalization.
	人們普遍地使用預設值的設定，當然，也有一小部分的自主性強的使用者例外，但大多數都沿用既定的設定值。照這樣來看，它的關鍵既不是客製化更不是個人化，它的關鍵在於如何能預先在應用中知道使用想用的可能選項，然後再引導到完全的個人化設定。
	If we understand where and how our users are interacting with our experiences, we can make our messages, and how they are delivered, that much more relevant to their environment— after all the medium is the message.
	如果我們能了解使用者如何跟在什麼地方與我們的經驗進行交流，那麼我們就能將要傳達的訊息，依照我們要的方式傳達出去，而且是和當時所處的環境更加有關的訊息。
	Understanding a user’s context also has interesting implications for supply and demand, and improved pricing. As Economics 101 tells us, the fairest price occurs at the point where consumer demand matches product supply — supply increases, prices go down; buyers demand more, the price goes up.
	了解使用者的溝通上下文，同樣會產生對供需來講有趣的含意，並有助於訂價。正如同一百零一條經濟學原理告訴我們的，最公平的價格發生在消費者的需求與產品供應相交的那個點— 供給增加，價格就會下降，購買者越多，價格就會上升。
	Demand pricing informed by digital inputs has been around forever, just think of the countless hotel and airline booking sites that price offers in real time based on market supply and demand.
	需求者的價格可以隨時在數位資料輸入後產生出來，就像無數的線上預訂飯店和機票那樣，價格都是基於市場的需求和供給而產生的。
	Or Uber’s surge pricing that raises the price of rides to match driver supply to rider demand. When there are more riders than drivers, increasing the price puts more drivers on the road to reduce customer wait times and improve its customer service.
	或者如優步的「動態訂價」那樣，以提高搭乘方的需求價格來符合司機方的供應價。如果搭乘方需求數多於司機方需求數，價格便會提高，然後吸引更多的司機過來載客以減少搭乘者的等待時間，提高服務水準。
	Uber’s head of economic research recently disclosed that riders are more likely to pay for surge pricing if their battery is low. Uber claims that they don’t use this information to gauge riders with low battery life, but it is an interesting glimpse into the ability of our digital products, informed by digital inputs, to understand the context from which consumers are making each purchase and how we can price offers based on predicted demand.
	優步的經濟研究主管最近發表一項結論表示，搭乘者當他們手機電力不足的時候，多半喜歡採用「動態訂價」。優步宣稱他們不會利用這些資料調查那些手機快沒電的搭乘者，但這透露了一項對數位化商品（動態訂價）有趣的面向，藉由收到數位的輸入資料，可以了解到消費者在做每一項購買決定時，相關的對話以及是如何透過預估需求給出訂價的。
	While context is important, a message is truly predictive when it’s personalized based on intent — this allows you to not only understand what, when, where, and how, but why? To determine intent, strengthen your inputs and data warehouse by supplementing click-stream digital analytics with machine learning and natural language processing.
	當然這些對話都頗具預測性，因為它的個人化選擇都是基於「意願」—這不但讓你了解誰、何時、什麼、如何、何地購買的，為了確認「意願」，利用學習機器和語言自然化程序來強化輸入數據的資料庫中，對點擊紀錄的統計分析。
	The step from the personalized to predictive occurs when your user data is processed through a machine-learning library like Google’s TensorFlow to determine trends, or a natural language processing platform like IBM’s Watson that can actually generate consumer profiles.
	當使用者的資料經過類似Google的TrensorFlow這種自我學習機器的圖書庫處理之後，數位行銷便從個人化進步到預測性的境界，可用以確認趨勢；或由類似IBM的Watson語言自然化處理平台，能實際產生客戶檔案。
	If I’m a clothing store, I may warn a customer it’s going to rain, or simply tailor my sales messaging to emphasize they are missing their last chance to buy a pair of pants on sale that they keep browsing past.
	It is important to note that predictive tech can help eliminate the paralyzing paradox of choice that can occur now that the world of decision sits at our fingertips.
	如果我是一家服飾店，我可以提醒某個客戶快下雨了，或製作一個提醒的訊息，通知客戶某件促銷商品快要售罄。要了解在目前這個彈指之間即可做出購買決定的時代，利用預測性科技可以減少因為「抉擇的矛盾」所造成購買時舉棋不定的情況。
	Predictive technology is even making a big splash in the B2B software industry, Salesforce’s CEO Marc Benioff spoke at the Forbes CIO summit in March and shared his belief that software that can analyze data and recommend the best course of action is the next wave of opportunity in his industry.
	預測性科技在B2B的商業軟體產業中也有很大的著墨，Salesforce公司的執行長Marc Benioff在三月份富比世的投資長高峰會中表示。同時他還說他相信軟體可以進行分析並認為最好的行動契機將會在下一波的產業機會中。
	This isn’t a hollow proclamation, Salesforce has walked the walk and acquired a number of machine learning and data startups in the last several years, including PredictionIO, MinHash, and Tempo AI.  Just two year ago, it spent nearly $400 millions to buy RelateIQ, an intelligent email client, calendar, and work dashboard that can automatically determine which salesperson has the best relationship with a client or predict when you need to start filing an upcoming project.
	這可不是一個空洞的宣言而已，Salesforce公司跨出步伐並且在過去幾年中，收購了一些機器學習及資料處理的新創公司，包括：PredictionIO, MinHash和Tempo AI等，就在兩年前花了將近400萬美金收購RelateIQ，一家整合智慧處理電子郵件、行事曆、和工作備忘面板的公司，它能自動地辨認出哪個業務人員與客戶維持著最良好的關係，或預測出你何時該推出新的專案計畫。
	Less Marketer, More Cyborg
	少一點行銷人員，多一點電子機器人
	Automated predictive technology can and will enable us to provide the ideal consumer experience at each touchpoint.
	A well-oiled bot built with confidence matching, neural networks, and deep learning will soon be able to predict and deliver better consumer experiences. And when the bot isn’t as well-oiled as you hoped, a human customer service rep that takes over will be starting informed by deep insights about the consumer’s pain points and how they have made it through the journey thus far.
	The purpose of data isn’t to collect it and assign it buzzwords, it should be used to solve real consumer problems, sometimes before they even happen.
	自動預測科技能夠在每一個接觸點上，幫助我們推測理想的消費者體驗。運作良好的BOT程式，結合了信心匹配運算、神經網絡及深度學習，便很快地預測和提供較好的消費者體驗。當這個BOT方案運作的情況不如你的預期，真人的客服系統便會進來銜接處理，同時深度的學習機制同時啟動，以便進一步洞察客戶不能滿足客戶的痛點在哪、如何形成的。蒐集到的數據，其目的不在於蒐集後把它變成標語，而應該能確實解決消費者的問題，有時甚至是問題發生之前！
	This week I was at Starbucks trying to get a coffee and I couldn’t log into the app. How personalized the in-app experience was no longer mattered, if anything it just made me that much more upset that my rewards were out of reach.
	這禮拜我去星巴克想點一杯咖啡，結果APP登錄不進去。這個APP設計得有多個人化其實已經不重要，如果用這個東西只會把我搞得更火大，那根本得不到我想要的期望。
	I let Starbucks know with 140 characters of fury.
	我那次讓星巴克了解了140種的憤怒。
	Okay admittedly I was polite, but that didn’t help matters because I received no response. I’m annoyed now and this interaction is going to leave a bad taste in my mouth (despite the delicious iced coffee).
	好吧我承認我一開始很禮貌，但是這樣沒用因為我得不到回應，等到我有些惱火的時候，講出來的話就不會太好聽（雖然冰咖啡的味道很棒。）
	In the grand scheme of things, this is isn’t a huge deal, and it hasn’t kept me from drinking Starbucks everyday. But I’ve got a decent sized Twitter following, and if a certain percentage of them see the tweet, and that Starbucks didn’t respond, it is not hubris on my part to think I could affect purchase behavior.
	這個事件的主軸，本來不是什麼大不了的事，我也不會因此而不再天天喝星巴克，不過因為我推特的追蹤者還蠻多的，而且如果有一部分的人看到我的貼文，而星巴克又一直不回應，那麼我們這邊的人會開始改變購買行為是一點都不誇張的。
	The experience economy has arrived and the hundreds, thousands, or in Starbucks case, millions of these touchpoints that regularly occur don’t shape your brand, they are your brand. Brands now stand in the center of the colosseum, subject to the will and favor of the angry connected mob. It’s impossible for any human marketer armed with social listening to stand their ground alone, but with the help of machines, they’ve got a chance.
	體驗經濟的時代已經來臨，數以百計、千計或類似像星巴克的案例，數以百萬計的體驗接觸點已正在常態化地發生進行中，他們所代表的不是塑造影響你的品牌，他們就是你的品牌！品牌現在正站在劇場的中央，它們的表現全憑廣大飢渴的群眾在接觸點所呈現的意願和喜好。這種場景已經不可能由任何一個真人的行銷人員可以藉由「社群聆聽」而一意孤行的方式可以做到，但如果有「機器」的協助，或許還有可能。
	Starbucks could have improved that touchpoint with smart error reporting that notified their technology team that my username had received an error message. Then, it’s just a matter of automating an apology.
	星巴克或許已經藉著可以使用聰明的錯誤報告系統，因為在系統中我的用戶名收到了一個錯誤訊息，而讓技術團隊的人察覺，但那最多只會多出現一個「自動化」的道歉信息而已。
	At the end of the day, it’s all about customer lifetime value. It’s about improving the relationships and messages that we create, about giving people what they want. And when consumers have complaints, concerns or issues, it’s about addressing and appeasing them with a gift.
	當一天結束，所有的發生的事終將歸結於顧客日常體驗的價值罷了，關係的改善和我們傳達出去的信息才是關鍵，提供顧客想要的才是關鍵。而當顧客產生抱怨、疑慮或是出現問題，修正它們和提供補償來安撫顧客才是關鍵。
	The idea of marketing customer service “cyborgs” fascinates me because since it’s half-human/half-machine. Theoretically, its got all the efficiencies and effectiveness of a machine, but also all the empathy of the human condition.
	客服行銷的「機器人」構想很吸引我，因為「他」是半人半機器，理論上是一部集效能與效率於一身的機器，但同時又具備人類的同理心。
	What if there was a distinct hand-off pre-baked into the process? A consumer could begin chatting to a brand bot through Facebook messenger but once the bot—using natural language processing—determines that the consumer has a high lifetime value, a human actually picks up the dialogue.
	假使我們設計出一套全自動化的程序並且任憑它自己運作，那結果會怎樣呢？ 某個消費者可能開始透過臉書messnger，對著一個品牌bot程式講一些話，一旦這個bot程式運用語言自然化程序，辨識出消費者有很高的體驗評價時，真人即可獲取這類的對話了。
	This would allow us to scale the one-to-one customer service conversation that the consumer craves. A human experience, but all the grunt work is provided by a smart AI that understands how to analyze inputs to determine the outputs that will drive positive outcomes.
	And of course, the brand with the most positive consumer experience wins. Amazon was first to market, but I’m willing to bet that Google Home will ultimately overtake Alexa for market share because we all use Google Search, GMail, Calendar, Apps, and maps. Home will have a better understanding of the end-user and the context to provide predictive—not personalized—experiences.
	這讓我們得以實現客人所企盼的「一對一行銷」顧客服務。由真人體驗到全由聰明的AI（人工智慧）來處理日常普通的工作，並且知道分析怎麼樣的投入與產出，可以得到好的結果。
	當然最後是能取得顧客最正面體驗評價的品牌勝出。Amazon目前居於領先地位，但我相信Google Home最終將取代Alexa獲取市場佔有率，畢竟我們大家都使用谷歌搜尋、Gmail、行事曆、應用程式及谷歌地圖。Google能更瞭解終端使用者並依照來龍去脈作出預測，而非「個人化」的體驗。
	As for executing personalized messaging, virtual assistant Amy, from Dennis Mortensen andX.ai, has proven that the technology is there to create intelligent agents, not bots, that can complete tasks from end-to-end. Providing seamless service, Amy coordinates and schedules meetings without any oversight. Once the agent is CC’d on a relevant email, it takes care of the rest.
	至於處理個人化的訊息，來自Dennis Mortensen和X.ai人工智慧系統，虛擬助理Amy已經證明科技創造的目的是智慧代理，而不是完成終端對終端的bot程式。提供無縫接軌的服務，Amy無需任何監督的情形下，便可協調排定會議日程，只要將郵件CC給她，其他都可以交給她搞定。
	Operator is an iPhone app that gives users recommendations based on their personal taste. Users text Operator a request and the app connects them with a (actually human) expert that asks follow up questions. Not surprisingly, the app is headed by Uber Co-Founder Robin Chan who understands the “uberification” of today’s consumers who crave “one tap” solutions to the problems that occur in micro-moments.
	Operator是一項iPhone的APP，可以根據個人喜好給消費者提建議。使用者寫下一項要求，APP便會連線到一個（真人的）專家來跟進這個提問。這項APP毫無意外地是有Uber優步的共同創辦人Robin Chan所領軍的，特別是他瞭解到現今的「優步化」的消費者，如何希望在片刻之間便能快速地得到問題的解決答案。
	A machine could start, or finish the customer journey and reduce the number of steps required from both the consumer and their support representative. The more we can use data to inform and educate, the more we can create valuable customer service relationships.
	機器它能啟動並完成顧客的消費過程，減少消費者和服務回應過程中的步驟。我們運用數據在通知和教育的情況越多，顧客關係的服務價值就越大。
	Humans aren’t going to be replaced any time soon, but maybe it’s time we let machines do even more of the work.
	真人在短期內並不會被取代，但或許是時候讓機器多做點事了。