最有看点的互联网金融门户

最有看点的互联网金融门户
国际资讯

金融服务新时代——掘金“大数据”

大数据不仅是技术问题,也不仅是数据问题。大数据可以帮助商业用户做出有价值的决定。重要的是怎么才能让金融机构建立组织信任和大数据创新。开发大数据的企业也不应仅把关注点放在技术层面。他们的目标应该是帮助企业领导(CMOs,CFOs和其他)更好的理解他们的数据,并迅速地做出有价值的决定。

那么,金融机构怎么才能从大数据中抓住商业机会呢?这篇报告主要就是为了帮助金融机构从商业角度最大化其大数据的价值。要是想更好的理解大数据价值产生的技术部分,可以参考:大数据:金融领域变局者。

匹配大数据和商业目标

用数据来了解商业是不是个简单的过程,主要包括了以下6个步骤:设立商业目标、获取数据、清洁数据、分析数据、获得可操作性的见解、取得成果。每个步骤,公司都有可能会遇到陷阱。不正确存储、不正确清洁和非索引的数据都会妨碍数据获取。不正确的数据分类、没能正确的搜集数据、不能把好数据从坏数据中分离出来都会破坏数据的编组。另外,如果见解和商业目的不一致的话,也会导致其不可操作。

在很多情况下,大数据处理过程的陷阱是因为数据分析方法和早期设立的商业目标不一致。所以,金融机构首先一定要秉承目标导向的运营方式。也就是说,银行和保险公司需要依据其期望的商业成果来解决操作过程中遇到的问题。另外,金融机构在考虑可能的商业成果时,不能仅仅局限于其现有的技术支撑。比如大数据技术几乎可以实时处理大量数据,很多以前不可能的机会现在都变得可能了。

印度的ICICI银行正通过大数据技术提高其贷款回收率。印度的ICICI银行想要在不疏远客户的情况下提高其贷款回收率。它也想优化这个传统的人工实地操作过程。银行想利用电子邮件、电话和信件等非侵入的方式来搜集客户的早期数据,这样就可以维护客户良好关系。但比较难的是银行要依据违约的具体情况,为每种案例都找个合适的方式。

2.1 第1张

银行应用了一个分析系统,这个系统可以获得每个违约案件的详情并把它分配给合适的部门或沟通渠道。为了寻找最佳的搜集方式,这个模型涵盖了很多参数,包括风险承担、风险行为、客户档案甚至采集器的效率。这个系统已经从根本上减少了信用损失。例如,在汽车贷款方面,银行提高了50%的贷款回收。在一些方面,这个系统减少了80%人力需求,也把信息搜集周期从5至6天压缩到了几个小时。

一个美国银行用大数据技术改善客户体验。这个银行从不同数据库搜集了大量的客户行为和互动渠道的数据。银行想要把这些数据整合起来分析,以期对客户有个全面的认识。随着对客户数据的深入分析,银行可以开发出更多吸引客户的线上、线下产品和服务,最终改善客户体验。

银行安装了一个从线上、线下搜集数据的分析系统,以期更全面了解客户及其他们通过银行节点的互动方式。这些整合的数据被反馈到银行的客户管理平台,为其客服中心提供相关的指导。它也会指导银行设计更贴近客户的网站。

通过对消费者行为和偏好的深入分析,银行将建立起更好的客户关系,同时创造更个性化的服务体验。银行增加了100%呼入呼出的转换,也在18个月完成了3个主要网站的重新设计,利用导向性的数据优化了网站内容,改善了客户体验。

沙海淘金

充分考虑商业目标后,金融机构可以获得几个大数据价值。大多数银行和保险公司很可能推动一些特别积极的商业行为,而这些现在看来都是真实而可行的。

情感分析

情感分析用来从大量和复杂的非结构数据中提取股东(客户、雇员、监管者等)情感。从电邮到文件,所有的电子交流数据可以用来分析某个期间的情感趋势,情感时间序列和其他时间序列,并为用户(产品、图表)提供可操作的信息。

商业应用包括建立在电话记录、电邮、博客和社交媒体数据可用性上的“客户之声”平台,和结合情感分析提高预测模型对风险、欺诈、交叉销售的预测。客户情感也可以加速产品开发上市,雇员情感可以用来分析雇员满意度及其他HR目标。

在客户流失分析中,例如交易、客户档案、社交媒体等混合数据被储存在大数据平台。这些数据用来分析情感分数并确定客户流失的可能性。这个平台基于实时模型的客户行为重新计算可能性,帮助金融机构在正确时机为正确的人提供正确的产品。很多客户流失分析的商业应用(包括交叉销售和摩擦减缓)帮助银行和保险公司提高了其产品定价和客户接受度的可能性。

2.2 第2张

保险代位求偿

代位分析可以分析代位成功的可能性。在代位努力上很小的一个提高就会产生客观的投资回报。拥有有效代位部门的保险公司能为投保人提供更低的保险金。从代位过程获得的钱可以直接影响保险公司的盈利。

例如汽车碰撞索赔案件,凯捷已经可以通过分析警方报告、邮件、调查反映等非结构数据计算故障测试分。这个方法旨在帮助求偿部门专注于非投保人负有更大责任的更可能成功的案件上。事故测量可被用来计算代位成功可能性和成功代位获取的金额。

欺诈检测

大数据可以用来分辨特定期间的循环交易和交易操纵,用以增加对内部交易的识别。它也可以基于其他因素,比如账单、应税证明、杂志订阅等来实时识别潜在的欺诈交易,破译行为模式,分辨虚假身份。

所有这些好处都和金融机构的商业目标紧紧相连,都是在金融机构收集的数据基础上就可以获得。

价值创造:影响商业结果的目标行动区

技术和商业都会影响大数据的价值。从商业角度来看,准确及时的决定可以减少公司的风险和损失,降低管理成本,在目标市场更有竞争力,并利用商业机会提高投资者对其信任度。为了成功的利用他们支配的信息,那些挣扎在可信信息里的银行和保险公司高管需要在心里信任信息。他们需要创建可以有效利用数据的管理模式,采用商业目标引导的技术策略。最后,他们应该积极寻找影响商业结果的特别可行区域。

 

Big data is not just a technology issue nor is it only about data. Big data is about enabling business users to make decisions that  create value. It is important to define what the technology toolset must address to enable FS businesses to establish organizational trust and harness the innovation possibilities of big data. But companies tackling big data should not limit their focus to the  technology solution. Their goal should be to help business leaders (CMOs, CFOs and others) make better sense of the real data they have, get to it quickly and make valuable decisions.

What do financial services (FS) companies need to know to derive the right business opportunities from big data? This paper  focuses on helping FS companies maximize the value they generate from big data from the business perspective. To better understand the technology component of big data value generation, see Grabbing Value from Big Data: The New Game Changer for Financial Services.

Matching Business Objectives and Data Analytics

 The journey toward being able to use data to gain insights about your business is a challenging one, involving six distinct steps:  Setting business objectives, acquiring the data, cleansing it, analyzing it, gaining actionable insights, and achieving successful  outcomes. At each stage, firms are faced with pitfalls that arise. Improperly stored, improperly cleansed, or non-indexed data may obstruct the task of data acquisition. Similarly, improper categorization of data, not gathering data at the desired frequency and not being able to separate the good data from the bad data can make the task of marshaling data effectively non-achievable.  Furthermore, the insights generated might be rendered non actionable if they are not in sync with business objectives.

In most cases the pitfalls that occur in the big data journey are because of a lack of congruence between the data analytics  approach and the business objectives set early in the journey. First and foremost, it is critical that financial services firms follow a  business objective-led approach. This means that the types of problems the bank or insurer tackles should be based on the  business outcomes it most wants to achieve. Furthermore, when considering possible business outcomes, financial services firms  should be unconstrained in their thinking about what technology can support. As big data technology can process massive  volumes of data in near real-time, opportunities previously considered unattainable may now be possible.

ICICI Bank in India  focused a big data effort on operational efficiency by working to derive actionable insights to improve debt collection. India’s  ICICI Bank wanted to improve its rates of debt collection without alienating customers. It also wanted to bring efficiencies to a process that traditionally had been carried out manually by agents in the field. By taking advantage of non-intrusive channels, such as e-mail, phone calls and letters, the bank hoped to improve collections from early delinquents while still maintaining strong relationships with them. The challenge was to match each case to the most appropriate collection channel, based on the level of delinquency.

2.1 第1张

The bank adopted an analytics system that captures the details of each delinquent case and assigns it to the appropriate channel or agent. The model factors in a wide range of parameters, including exposure, risk behavior, customer profile and even the efficiency of the collector, to identify the best method of collection. The new system has helped to substantially reduce credit losses and improve productivity. In the area of auto loans, for example, the bank increased debt collections by 50 percent. In some areas, it has reduced its manpower needs by 80 percent. And turnaround time on collections has been condensed from five-to-six days, to a matter of hours.

A leading U.S. retail bank focused its big data effort on improving the customer experience. This bank collects massive amounts of data on customer behavior and channel interactions, but kept it stored in different data warehouses. The bank wanted to bring all the data together to analyze it and create a more holistic picture of customers. With the in-depth customer data it hoped to  develop more targeted product offers as well as more appealing online content. The ultimate goal was to improve the overall  customer experience.

The bank installed an analytics system that integrates data from online and offline channels, resulting in a more global understanding of customers and how they interact through all the bank’s touch points. This integrated data feeds into the bank’s customer relationship management platform, supplying the call center with more relevant leads. It also informs the bank’s decisions on how to design its web site to optimize customer engagement.

Through the detailed insights it offers into customer behaviors and preferences, the analytics system is helping the bank deepen customer relationships and create more personalized experiences. Specifically, the bank has increased conversions from inbound and outbound calls by 100 percent. It also has executed three major website redesigns in 18 months, using data-driven insights to optimize the content and increase customer engagement.

Diamonds in the Rough

With business objectives in mind, there are a few big data “diamond” initiatives financial services firms should consider—ones that, in most banks and insurance companies are likely to drive specific, positive business outcomes and are “real”and operational today.

Sentiment analysis

Sentiment analysis can be used to determine insights into stakeholder (customers, employees, regulators etc.) sentiment from massive, complex unstructured data. All types of electronic communication data from emails to documents can be analyzed to determine sentiment trends over a period of time, correlate sentiment time series with any other time series of interest and generate actionable messages for a given user or entity (product, geography).

Business applications include such things as a ‘Voice-of-Customer’ platform based on data available in call records,emails, blogs and social media data or to improve predictive models for risk, fraud, or cross-selling by combining Sentiment Analytics.  Customer sentiment can also be used in product development to accelerate speed to market and employee sentiment can be used to analyze data such as employee satisfaction surveys for staff retention insights and other HR objectives.

Customer churn and next best action In customer churn analysis multiple streams of data such as transactions, customer profiles, and social media data are sourced and stored on a big data platform. The data is then analyzed to create sentiment scores and determine the churn probability of customers. This platform also provides the insight for financial services firms to offer the right product to the right person at the right time (next best action) by determining and recalculating probabilities and offers based on customer actions in a near real-time mode. There are numerous business applications for customer churn analysis including cross-selling and attrition mitigation to increase profitability for retail banking or insurance, determining product pricing trends and determining how a particular offer will be received by a customer or set of customers.

2.2 第2张

Insurance subrogation

Subrogation analytics can be used to determine the probability of a successful subrogation. Even a small improvement in subrogation efforts can yield an attractive return on investment. An insurance company with an effective subrogation department can offer lower premiums to their policyholders. Any monies recovered through the subrogation process go directly to the insurance company’s bottom line.

For instance, for auto collision claims, Capgemini has built the capability to calculate a fault measure score using text analytics on unstructured data such as police reports, emails, survey responses etc. The idea behind this measure is to attribute the fault  probabilistically to either driver and then to inform the claims department to focus only on the likely to succeed claims where the fault of the non-policy-holder is deemed to be greater. Fault measure in turn is used to calculate two key outputs: Probability of the success of subrogation and the settlement amount from successful subrogation.

Fraud detection

Big data can be used to determine circular trading or trade manipulation patterns for a given instrument in a specific period of time and thus increase detection of internal trading. It can also be used to determine potentially fraudulent transactions in near real time based on multiple characteristics or factors which can be aggregated by combining other data such as utility bills, taxes paid, magazine subscriptions etc. to decipher footprint (in physical world) and behavior patterns and to differentiate false  identities from real ones.

Each of these diamonds is well aligned to financial services company business objectives, is readily operational, and can typically be implemented based on data the FS firm is already collecting.

Value Creation: Target Actionable Areas that Impact Business Outcomes

There are both technology and business dimensions to determining the value of big data initiatives. From a business perspective, accurate and timely decision-making allows firms to reduce risks and losses, lower regulatory capital requirements, compete more effectively in their target markets, capitalize on emerging business opportunities and enhance investor confidence. To successfully leverage the huge amount of data at their disposal, banks and insurers need to establish confidence in the minds of executives who struggle with trusting information used to make decisions. They need to create the governance required to leverage data effectively across the enterprise and adopt technology strategies that are led by business objectives. Then, they must pick specific actionable areas to address that will impact the desired business outcomes.


用微信扫描可以分享至好友和朋友圈

扫描二维码或搜索微信号“iweiyangx”
关注未央网官方微信公众号,获取互联网金融领域前沿资讯。

发表评论

发表评论

您的评论提交后会进行审核,审核通过的留言会展示在下方留言区域,请耐心等待。

评论

您的个人信息不会被公开,请放心填写! 标记为的是必填项

取消

王 少龙

2
总文章数

TA还没写个人介绍。。。

新金融时代区域性互金平台的资产优势

张彦军 | 网贷之家 21小时前

栈略数据:用大数据智能风控服务健康险

图图 | 未央团队 1天前

互联网金融狂潮简史

秋源俊二 11-18

[未央研究]本周互联网金融回顾 | 2017年第46周

未央研究 11-17

他的模式全球风靡,却在中国水土不服?

一本财经 11-17

版权所有 © 清华大学五道口金融学院互联网金融实验室 | 京ICP备17044750号-1