独家|机器学习帮助摩根士丹利理解客户需求(Exclusive machine learning helps Morgan Stanley understand customer needs)

  原文标题:

  How Machine Learning Is Helping Morgan Stanley Better Understand Client Needs

  作者:homas H. Davenport; Randy Bean

  翻译:吴昊

  术语校对:王伟玲

  全文校对:王红玉

  本文长度为1800字,建议阅读4分钟

  巴布森学院管理和信息技术专业的杰出教授Thomas H. Davenport和NewVantage Partners公司的首席执行官Randy Bean为你解读摩根士丹利最新的智能投顾“next best action”是如何在该公司的业务中发挥作用的。

  智能投顾,即为金融公司提供投资建议的自动化系统。虽然行业中大多数人对这个名词并不感冒,但是它但是其流行的势头仍然难以阻挡。不过,摩根士丹利最近宣布了一款经过机器学习增强的人性化咨询系统,它的智能程度远远超出了“机器”这一标签,可能在未来将这个词扫入历史的废纸筐中。

  位于纽约的摩根士丹利自1935年开始运营以来一直被誉为零售投资行业中最具人本精神的企业之一。摩根士丹利拥有16,000名财务顾问,他们通过诸如面对面会议和电话等传统渠道与客户保持着稳固的工作关系。然而,该公司深知传统渠道劳动密集型的性质限制了潜在客户数量的增长,而且主要吸引的都是年长的客户。根据德勤研究所得报告(

  因此,摩根士丹利的财富经营管理部门近年来持续致力于研发“next best action”系统,这个系统可用来帮助财务顾问更高效地向客户提供更有效的服务。该系统的初始版本功能仅为使用基于现有规则的方法来提出投资选择的建议。升级换代之后采用了基于机器学习的系统,这样便可以更好地将适配投资潜在选择与客户的偏好。对于财务顾问来说,当前有太多的投资选择要追踪并呈递给客户。一旦在市场上发生重大事件,例如,英国脱欧投票及由此造成的英国股票下跌,此类重要信息很难在短时间内由财务顾问亲自传达给所有客户。

  摩根士丹利的下一代“next best action”系统则着眼于三个独立的目标。其中只有一个在智能投顾市场上常见,即为对客户投资的提供投资洞察和选择。在大多数现有的智能投顾所提供的服务中,推荐的投资建议都是完全被动的,即针对共同基金或交易所交易基金。摩根士丹利的系统可以根据客户的意愿提供这些信息,也可以根据公司的研究成果提供个人股票或债券(的投资建议)。财务顾问借此可以获得数个想法以供客户选择,并且通过自己判断决定是否传递其中任何一个或全部。

  系统的第二个方面是提供操作警报。这些可能包括保证金通知,低现金余额警报或客户投资组合大幅增加或减少的通知。它也可以留意到金融市场上值得注意的事件,例如上述的脱欧投票。财务顾问可以将个性化的文本与警报信息附在一起并通过多种通信渠道发布。

  最后,摩根士丹利系统还包括了针对用户个人生活信息的服务。例如,如果一个客户有个孩子得了某种疾病,该系统可以推荐最好的当地医院和学校,以及用于治疗疾病的财务策略。在其他智能投顾系统还没有该种生活服务的时候,这一独有的特性将有助于客户和财务顾问之间建立一个累积信任和价值的关系。

  这个系统的特点和功能都令人印象深刻,但一旦涉及到推向市场,它的初次体验往往决定了它未来是否成功。因此摩根士丹利在推广过程中也一直保持着谨慎,机敏和开放的态度,所以系统在设计过程中引入了数个财务顾问。目前该系统的开发已经完成,正处于测试状态之中,并将在9月份推荐给500个财务顾问使用。在首席数据和分析主管Jeff McMillan的领导下,财富管理分析和数据组织(the Analytics and Data Organization within Wealth Management)完成了对此系统的研发。Jeff McMillan一开始就知道让财务顾问使用该系统是一个巨大的变革管理项目。一般来说,财务顾问的工作依赖于他们的过往经验,并且最初他们也不了解机器系统的运作方式。

  “next best action”系统将在一开始先通过财务顾问进行仲裁,但客户也可以访问最新的线上信息。摩根士丹利计划最终发行具有投资组合管理功能的纯数字版本,这样就可以用更低地成本向更偏爱纯数字版的顾客提供服务(客户中很多人为千禧一代)。为了协助顾客,以及接受了该系统的投资顾问,一位供职于呼叫中心外的数字顾问还将就系统的使用提供专家级别的建议。

  McMillan强调人类将保持在财富管理方面的作用。他认为“智能投顾”这个术语特别令人讨厌。通过电话他告诉我们:

  在可预见的未来,像这样的系统将会作为顾问和客户之间人际关系的一个补充。在整个行业中,“人机混合”产品已经越来越成功。但是人可以理解上下文,处理客户情绪,处理不同的数据集,所以人在财务咨询方面仍然扮演着非常重要的角色。

  McMillan和他的同事们做了大量工作,以使公司所有的投资知识都可以通过该系统灵活运用。例如,他们发现当前没有一种人工智能系统提取运用投资分析报告中蕴含的知识,以此来支持他们向客户所提供的投资策略。所以McMillan正在与公司的研究部门合作,努力使报告中的知识更加结构化以便于机器来学习运用。这一挑战一旦达成,它至少将有助于财务顾问更加有效地使用 “next best action”系统。

  当然,这个新的系统和流程之中,机器层面的占比只有很小的部分。无论是摩根士丹利的商业模式,还是其文化使然,都不会采用完全基于机器且不提供人类支持的财富管理解决方案。我们也同样认为行业中的大多数公司也都认同这一理念。

————————

Original title:

  How Machine Learning Is Helping Morgan Stanley Better Understand Client Needs

  作者:homas H. Davenport; Randy Bean

Wu Hao

Term proofreader: Wang Weiling

Proofread by: Wang Hongyu

The length of this article is 1800 words. It is recommended to read it for 4 minutes

Thomas H. Davenport, an outstanding professor of management and information technology at Babson College, and Randy bean, CEO of newvantage partners, explain how Morgan Stanley’s latest intelligent investment adviser “next best action” plays a role in the company’s business.

Intelligent investment adviser is an automatic system that provides investment advice to financial companies. Although most people in the industry are not interested in this term, its popularity is still unstoppable. However, Morgan Stanley recently announced a humanized consulting system enhanced by machine learning, which is far more intelligent than the label of “machine”, and may sweep the word into the waste paper basket of history in the future.

Morgan Stanley in New York has been known as one of the most humanistic enterprises in the retail investment industry since it began operation in 1935. Morgan Stanley has 16000 financial advisers who maintain a strong working relationship with customers through traditional channels such as face-to-face meetings and telephone calls. However, the company is well aware that the labor-intensive nature of traditional channels limits the growth of potential customers, and mainly attracts older customers. According to Deloitte Research Report(

Therefore, the wealth management department of Morgan Stanley has been continuously committed to developing the “next best action” system in recent years, which can be used to help financial advisers provide more efficient and effective services to customers. The function of the initial version of the system is only to use the method based on existing rules to make investment selection suggestions. After upgrading, the system based on machine learning is adopted, which can better adapt the potential investment choices to customers’ preferences. For financial advisers, there are too many investment options to track and present to customers. Once major events occur in the market, such as the brexit vote and the resulting decline in UK stocks, it is difficult for the financial adviser to personally convey such important information to all customers in a short time.

Morgan Stanley’s next-generation “next best action” system focuses on three independent goals. Only one of them is common in the intelligent investment consulting market, that is, providing investment insight and choice for customer investment. Among the services provided by most existing intelligent investment advisers, the recommended investment suggestions are completely passive, that is, for mutual funds or exchange traded funds. Morgan Stanley’s system can provide this information according to the wishes of customers, and can also provide individual stocks or bonds (investment suggestions) according to the company’s research results. In this way, the financial advisor can obtain several ideas for the customer to choose, and decide whether to transfer any or all of them through his own judgment.

The second aspect of the system is to provide operational alerts. These may include margin notices, low cash balance alerts or notices of significant increases or decreases in the client’s portfolio. It can also note noteworthy events in financial markets, such as the brexit vote mentioned above. Financial advisers can attach personalized text and alert information and publish it through a variety of communication channels.

Finally, Morgan Stanley system also includes services for users’ personal life information. For example, if a customer has a child with a disease, the system can recommend the best local hospitals and schools, as well as financial strategies for treating the disease. When other intelligent investment advisory systems do not have this kind of life service, this unique feature will help customers and financial advisers establish a relationship of accumulated trust and value.

The features and functions of this system are impressive, but once it comes to marketing, its first experience often determines its future success. Therefore, Morgan Stanley has always maintained a cautious, smart and open attitude in the promotion process, so several financial consultants have been introduced into the design process of the system. At present, the development of the system has been completed, is in the test state, and will be recommended to 500 financial advisers in September. Under the leadership of Jeff McMillan, chief data and Analysis Director, the analytics and data organization within wealth management has completed the research and development of this system. Jeff McMillan knew from the beginning that letting financial advisers use the system was a huge change management project. Generally speaking, the work of financial advisers depends on their past experience, and initially they do not understand the operation of machine systems.

The “next best action” system will arbitrate through the financial advisor at the beginning, but customers can also access the latest online information. Morgan Stanley plans to eventually release a pure digital version with portfolio management function, so that it can provide services to customers who prefer the pure digital version at a lower cost (many of them are millennials). In order to assist customers and investment consultants who have accepted the system, a digital consultant working outside the call center will also provide expert level advice on the use of the system.

McMillan stressed that mankind will maintain its role in wealth management. He found the term “intelligent investment adviser” particularly annoying. He told us by telephone:

In the foreseeable future, systems like this will complement the interpersonal relationship between consultants and customers. In the whole industry, “man-machine hybrid” products have become more and more successful. However, people can understand the context, deal with customer emotions and deal with different data sets, so people still play a very important role in financial consulting.

McMillan and his colleagues have done a lot of work so that all the investment knowledge of the company can be flexibly applied through the system. For example, they found that there is no artificial intelligence system to extract and apply the knowledge contained in the investment analysis report to support the investment strategy they provide to their customers. So McMillan is working with the company’s research department to make the knowledge in the report more structured so that machines can learn and use it. Once this challenge is met, it will at least help financial advisers use the “next best action” system more effectively.

Of course, the machine level accounts for only a small part of this new system and process. Neither Morgan Stanley’s business model nor its culture will adopt wealth management solutions that are completely machine-based and do not provide human support. We also believe that most companies in the industry agree with this concept.